News - Niouzes

BEL 20

CAC 40

DAX

S&P 500

FTSE 100

NASDAQ

Nikkei 225

Dow Jones Industrial

Crude Oil

Gold

EUR/USD

EUR/GBP

Monthly Food Inflation

27/07/2024 15:12 – The Art of Saying No: Live a More Focused, Higher Quality Life
#Goal #SocialPsychology #Psychology #Gratitude #BehaviouralSciences #Triage #SelfCare #PsychologicalConcepts #Cognition #PrimateBehavior #Reason #Productivity #BranchesOfScience #MScottPeck #BehaviorModification #SelfEsteem #Empathy #Physician

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
BlogAll ArticlesThe Art of Saying No: Live a More Focused, Higher Quality Life

Happiness & Fulfillment

***THE ART OF SAYING NO: LIVE A MORE FOCUSED, HIGHER QUALITY LIFE***

July 24, 2024 • 7 Min Read

This post may contain links from our sponsors. We provide you with accurate, reliable information. Learn more about how we make money and select our advertising partners.

Have you felt overwhelmed from being so busy, both in your professional and personal lives? There are only so many hours in a day, and weâ€™re often trying to fit too much in. For a long time, I was in this cycle.â

At a certain point, while running multiple businesses and keeping crazy physician hours, it became too much. I just couldnâ€™t do it all. Instead of doing a few things well, it was like I was falling short on everything. I took a step back and realized I was saying â€˜yesâ€™ to almost everything. In the end, that diluted my ability to balance work with what truly matters most for me: friends and family. It was time to figure out what was important, what was not, what to delegate, and how to prioritize my time.

I realized I needed to say â€˜yesâ€™ to less and â€˜noâ€™ to more.

Today, weâ€™re learning how to reclaim our time and taking a closer look at how to set boundaries, rejuvenate your energy, prioritize whatâ€™s most important to you, and to do it all without feeling guilty. That will include helpful tenets and actional steps to empower yourself and prioritize your life goals.

**Understanding The Value Of Your Time**

Author of The Road Less Traveledâ€”a wonderful book on achieving fulfillmentâ€”M. Scott Peck once said, â€œUntil you value yourself, you wonâ€™t value your time. And until you value your time, you will not do anything with it.â€ Thatâ€™s an important lesson: value your time. Time is the one resource you canâ€™t get back once you spend it.â

Artfully saying â€˜noâ€™ to reclaim your time is part of this process. But I get it. I love saying â€˜yesâ€™ to as many things as possible. It got to the point where I felt obligated to always say â€˜yes.â€™ And when I couldnâ€™t say â€˜yes,â€™ I felt guilty.

But when you value your time, saying â€˜noâ€™ can be a strategic decision that prioritizes your high-value tasks and responsibilities, especially those essential for busy doctors.

**Freedom To Choose**

One way to regain your freedom is to choose your commitments. This allows better focus and higher productivity which, in turn, helps you accomplish what you have in front of you. The results will be immediate, and youâ€™ll work better.â

Hereâ€™s another way to think about it. When you say yes, youâ€™ve locked yourself in, losing your freedom to choose. James Clear, author of Atomic Habits: Tiny Changes, Remarkable Results, said it well: â€œWhen you say no, you are only saying no to one option. When you say yes, you are saying no to every other option.â€

Saying â€˜yesâ€™ all the time doesnâ€™t give you more freedom; itâ€™s the opposite. By being selective, you ensure your chosen commitments are in your best interests. Itâ€™s not just about declining offers. Itâ€™s about choosing a path that leads to better focus and greater productivity.â

We need to understand that saying â€˜noâ€™ has a major impact on our ability to say â€˜yes.â€™ Every time you say â€˜yesâ€™ to something that is not one of your main priorities, you say no to something that is one of your main priorities. Let that sink in.

Whatâ€™s more, unimportant commitments can drain energy and divert attention from crucial tasks, making you less successful at many things as opposed to extremely dedicated and successful at a handful of things.

**How To Politely Say â€œNoâ€**

It can be so hard to say â€˜noâ€™ and not feel guilty, like maybe you are offending the other person in some way. However, saying â€˜noâ€™ doesnâ€™t mean you are being rude, selfish, or unkind. These are unhelpful and misguided beliefs.

So how can we overcome this to politely say â€˜noâ€™? Here are some easy strategies and tips I use.

Express gratitude: Genuinely thank the person for their offer. This shows that you value their time and that theyâ€™ve considered you.

Be clear and concise: Directly state that you cannot commit rather than leave it ambiguous. And donâ€™t provide an overly detailed excuse.

Provide a brief reason: If appropriate, offer a simple reason for declining so the other person understands your position better.

Offer an alternative: If possible, suggest another solution or person who might be able to help with the request.

Leave the door open: Indicate that you might be open to future opportunities to maintain a positive relationship.

Set a boundary: Politely set or reinforce a boundary if you feel this might be an ongoing issue.

Give them a positive â€˜noâ€™: Sandwich your â€˜noâ€™ between two positives, such as gratitude and leaving the door open, to soften the impact.

Decline with empathy: Show that you understand their needs or feelings, something that can help lessen any disappointment.

**Reinforcing Boundaries**

For physicians who face the dual pressures of clinical duties and personal commitments, reinforcing boundaries is crucial. Setting boundaries is not a sign of selfishness but an act of self respect. They define our limits and protect whatâ€™s most important to us.â

When you set boundaries in a healthy way, you maintain a capacity for personal growth, self-discovery, and self-care. So how do you set boundaries? Start by using some of the strategies weâ€™ve covered to say â€˜noâ€™ more often. But here are some additional things you can try.

Set specific work hours and communicate them clearly to colleagues and patients. Use scheduling tools to enforce these times and manage expectations.

Keep a diligent calendar to understand exactly how youâ€™re spending your time. This can help you clarify how youâ€™re prioritizing your time.

Delegate and triage as applicable. For example, train your staff to handle certain tasks, or use a system that prioritizes issues based on urgency.

Schedule self-care. Block time for personal activities that leave you fulfilled, like exercising, hobbies, or time with friends and family. Treat this time as non-negotiable.

Review and adjust. Regularly assess how effective your boundaries are and adjust them based on feedback and your own observations. They need to meet your evolving needs.

**Saying â€œYesâ€ To Yourself**

Remember that you deserve the choice of how to spend your time and how to prioritize your life. That is not selfish. Itâ€™s you saying â€˜yesâ€™ to yourself. Saying â€˜yesâ€™ to yourself sets you on the path to achieving your goals and living a successful and fulfilled life.

Prioritizing has been important for me as a physician but also as somebody helping fellow physicians through my communities such as Passive Income MD, Curbside Real Estate, and Ascent Equity Group. I am able to be there for the community while also allowing time for my family. This takes practice, and Iâ€™m still working on it.â

Practicing the techniques we covered today can make saying no easier without being harsh. But youâ€™ll also align your reclaimed time with your priorities, taking you a step in the right direction to achieve your goals.

We talked about many actionable steps today. This week, try practicing saying â€˜noâ€™ to one unnecessary commitment that doesnâ€™t align with your lifeâ€™s priorities. One step at a time, you can start to better manage your time and energy.

Until I see you next, my hope is that you continue to prioritize your time to create an ideal life.

Peter Kim, MD is the founder of Passive Income MD, the creator of Passive Real Estate Academy, and offers weekly education through his Monday podcast, the Passive Income MD Podcast. Join our community at the Passive Income Doc Facebook Group.

Disclaimer: The topic presented in this article is provided as general information and for educational purposes. It is not a substitute for professional advice. Accordingly, before taking action, consult with your team of professionals.

Article

Top AI Tools for Medical Transcription and Documentation

Next Article

3 Essential AI Tools for Your Side Hustle

Never miss an article.

Sign up for our newsletter to get blog updates sent straight to your inbox.

Sign up for tips, tricks & updates

Site Design

Delightful Studios

Site Development

Alchemy + Aim
27/07/2024 09:59 – A Hacker ‘Ghost’ Network Is Quietly Spreading Malware on GitHub
#Malware #Cyberspace #Technology #Cybercrime #InformationTechnology #Cyberwarfare #ComputerSecurity #Computing #DigitalMedia #Software #ComputerLaw #InformationAge #Internet #ComputerScience #CyberpunkThemes #Infostealer #WorldWideWeb #InformationTechnologyManagement #GitHub #SecurityEngineering #SecureCommunication #CrimePrevention #SoftwareEngineering #SecurityTechnology #Multimedia #DigitalTechnology #ComputerMediatedCommunication #Telecommunications #SocialInfluence #MassMediaTechnology #InformationScience #Malware #Cyberspace #Technology #Cybercrime #InformationTechnology #Cyberwarfare #ComputerSecurity #Computing #DigitalMedia #Software #ComputerLaw #InformationAge #Internet #ComputerScience #CyberpunkThemes #Infostealer #WorldWideWeb #InformationTechnologyManagement #GitHub #SecurityEngineering #SecureCommunication #CrimePrevention #SoftwareEngineering #SecurityTechnology #Multimedia #DigitalTechnology #ComputerMediatedCommunication #Telecommunications #SocialInfluence #MassMediaTechnology #InformationScience

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
SaveSave this storySaveA secretive network of around 3,000 “ghost” accounts on GitHub has quietly been manipulating pages on the code-hosting website to promote malware and phishing links, according to new research seen by WIRED.

Since at least June last year, according to researchers at cybersecurity company Check Point, a cybercriminal they dubbed “Stargazer Goblin” has been hosting malicious code repositories on the Microsoft-owned platform. GitHub is the world’s largest open-source code website, hosting millions of developers' work. As well as uploading malicious repositories, Stargazer Goblin has been boosting the pages by using GitHub’s own community tools.

Antonis Terefos, a malware reverse engineer at Check Point who discovered the nefarious behavior, says the persona behind the network uses their false accounts to “star,” “fork,” and “watch” the malicious pages. These actions—which are loosely similar to liking, sharing, and subscribing, respectively—help make the pages appear popular and genuine. The more stars, the more realistic a page looks. “The malicious repositories appeared really legitimate,” Terefos says.

“The way he has developed it is really smart, taking advantage of how GitHub operates,” Terefos says of the person behind the persona. While cybercriminals have been abusing GitHub for years, uploading malicious code and adapting legitimate repositories, Terefos says he has not previously seen a network of fake accounts operating in this way on the platform. The buying and selling of repositories and starring is coordinated on a cybercrime-linked Telegram channel and criminal marketplaces. WIRED previously reported on other GitHub black markets.

The Stargazers Ghost Network, which Check Point named after one of the first accounts they spotted, has been spreading malicious GitHub repositories that offer downloads of social media, gaming, and cryptocurrency tools. For instance, pages might be claiming to provide code to run a VPN or license a version of Adobe's Photoshop. These are mostly targeting Windows users, the research says, and aim to capitalize on people potentially searching for free software online.

The operator behind the network charges other hackers to use their services, which Check Point call “distribution as a service.” The harmful network has been spotted sharing various types of ransomware and info-stealer malware, Check Point says, including the Atlantida Stealer, Rhadamanthys, and the Lumma Stealer. Terefos says he discovered the network while researching instances of the Atlantida Stealer. The researcher says the network could be bigger than he expects, as he has also seen legitimate GitHub accounts being taken over using stolen login details.

“We disabled user accounts in accordance with GitHub’s Acceptable Use Policies, which prohibit posting content that directly supports unlawful active attack or malware campaigns that are causing technical harms,” says Alexis Wales, vice president of security operations at GitHub. “We have teams dedicated to detecting, analyzing, and removing content and accounts that violate these policies.”

GitHub has more than 100 million users who have contributed over 420 million repositories on the platform. Given the breadth of the platform, it’s unsurprising that cybercriminals and hackers are attempting to abuse it. In recent years, researchers have been mapping instances of fake stars, spotting dangerous code hidden in projects, facing growing supply-chain attacks against open source software, and seeing comments being used to spread malware.

Most Popular

CultureThe 19 Best Movies on Amazon Prime Right NowBy Matt KamenGearThese Are the Android Phones Worth BuyingBy Julian ChokkattuCultureThe Week of Brat Summer Will Go Down in Internet HistoryBy Angela WatercutterSecurityEurope Is Pumping Billions Into New Military TechBy Luca ZorloniThe Stargazer Goblin threat actor identified by Check Point sells their services through ads on cybercrime forums and also through a Telegram account. A posts on a Russian-language cybercrime forum advertises 100 stars for $10 and 500 for $50 and says they can provide clones of existing repositories and trusted accounts. “For GitHub, the process looks organic,” one translated post says. The Check Point research says the network could have started operating as early as August 2022 and may have made as much as $100,000—from mid-May to mid-June this year, they estimate the operator made around $8,000.

Terefos says, in some instances, he has seen a legitimate code repository being changed by the threat actor, potentially using stolen credentials, and turned into a malicious one. If legitimate users fork a malicious repository, it has the potential to spread the code, Terefos says. The reverse malware engineer adds that he has automated the search for accounts linked to the network and is able to identify them based on common features, such as repositories using similar templates and tags. Some of the repositories seen by WIRED use variations on "instagram-follower" and "youtube-views" tags, with the names changed based on the software the page alleges to offer.

“Users of GitHub, and especially inexperienced users, can easily download malicious code, which can often be the result of fictitious reviews and starring,” says Jake Moore, global cybersecurity adviser at security firm Eset. “Telltale signs of malicious code on GitHub could also be unexpected or suspicious code changes, code that accesses external resources, and specific hard-coded credentials or API keys.”

Terefos says the activity of the network—staring and watching pages—is likely automated, as he saw repositories being acted upon in quick succession. “I don't think they're clicking, doing like manual work.” For GitHub, Terefos says, it may be difficult to identify this activity, as the behavior of the accounts is intended to look like a genuine GitHub user. Wales, from GitHub, says the company uses a combination of manual reviews and “at-scale detections” that use machine learning to identify suspicious activity

The Check Point engineer also says he identified one YouTube “ghost” account that was sharing malicious links via video, indicating that the network could be more encompassing. “I think this is not the whole picture,” Terefos says.

**You Might Also Like …**

In your inbox: The best and weirdest stories from WIRED’s archive

How the brain decides what to remember

The Big Story: Meet Priscila, queen of the rideshare mafia

Silicon Valley's soulless plutocrats flip for Donald Trump

Event: Join us for The Big Interview on December 3 in San Francisco

Matt Burgess is a senior writer at WIRED focused on information security, privacy, and data regulation in Europe. He graduated from the University of Sheffield with a degree in journalism and now lives in London. Send tips to Matt_Burgess@wired.com.Senior writerXTopicscybersecurityhackingmalwaresecurityvulnerabilitiesransomwaregithubHackers Leaking Taylor Swift Tickets? Don’t Get Your Hopes UpPlus: Researchers uncover a new way to expose CSAM peddlers, OpenAI suffered a secret cyberattack, cryptocurrency thefts jump in 2024, and Twilio confirms hackers stole 33 million phone numbers.Matt BurgessHow One Bad CrowdStrike Update Crashed the World’s ComputersA defective CrowdStrike update sent computers around the globe into a reboot death spiral, taking down air travel, hospitals, banks, and more with it. Here’s how that’s possible.Lily Hay NewmanSpyware Users Exposed in Major Data BreachPlus: The Heritage Foundation gets hacked over Project 2025, a car dealership software provider seems to have paid $25 million to a ransomware gang, and authorities disrupt a Russian bot farm.Andrew CoutsDon’t Fall for CrowdStrike Outage ScamsSwindlers are spinning up bogus websites in an attempt to dupe people with “CrowdStrike support” scams following the security firm's catastrophic software update.Lily Hay NewmanHackers Claim to Have Leaked 1.1 TB of Disney Slack MessagesA hacker group called “NullBulge” says it stole more than a terabyte of Disney’s internal Slack messages and files from nearly 10,000 channels in an apparent protest over AI-generated art.Lily Hay NewmanThe $11 Billion Marketplace Enabling the Crypto Scam EconomyDeepfake scam services. Victim data. Electrified shackles for human trafficking. Crypto tracing firm Elliptic found all were available for sale on an online marketplace linked to Cambodia’s ruling family.Andy GreenbergThe Sweeping Danger of the AT&T Phone Records BreachTelecom giant AT&T says a major data breach has exposed the call and text records of “nearly all” of its customers, epitomizing the dire state of data security.Lily Hay NewmanGoogle Is Piloting Face Recognition for Office SecurityPlus: A cloud company says notorious Russian hacker group APT29 attacked it, Chinese hackers use ransomware to hide their espionage campaigns, and a bank popular with startups discloses a cyberattack.Lily Hay Newman
27/07/2024 09:55 – How IT Departments Scrambled to Address the CrowdStrike Chaos
#BitLocker #Cyberwarfare #ComputerScience #Cyberspace #Technology #Computing #ComputerSecurity #Cybercrime #Software #2024CrowdStrikeIncident #Computers #BlueScreenOfDeath #InformationTechnology #InformationAge #SecureCommunication #SecurityEngineering #Telecommunications #DigitalMedia #InformationSecurity #ComputerEngineering #Malware #InformationTechnologyManagement #CrimePrevention

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
If you buy something using links in our stories, we may earn a commission. Learn more.

Save this story

SaveSave this storySaveJust before 1:00 am local time on Friday, a system administrator for a West Coast company that handles funeral and mortuary services woke up suddenly and noticed his computer screen was aglow. When he checked his company phone, it was exploding with messages about what his colleagues were calling a network issue. Their entire infrastructure was down, threatening to upend funerals and burials.

It soon became clear the massive disruption was caused by the CrowdStrike outage. The security firm accidentally caused chaos around the world on Friday and into the weekend after distributing faulty software to its Falcon monitoring platform, hobbling airlines, hospitals, and other businesses, both small and large.

The administrator, who asked to remain anonymous because he is not authorized to speak publicly about the outage, sprang into action. He ended up working a nearly 20-hour day, driving from mortuary to mortuary and resetting dozens of computers in person to resolve the problem. The situation was urgent, the administrator explains, because the computers needed to be back online so there wouldn’t be disruptions to funeral service scheduling and mortuary communication with hospitals.

“With an issue as extensive as we saw with the CrowdStrike outage, it made sense to make sure that our company was good to go so we can get these families in, so they’re able to go through the services and be with their family members,” the system administrator says. “People are grieving.”

The flawed CrowdStrike update bricked some 8.5 million Windows computers worldwide, sending them into the dreaded Blue Screen of Death (BSOD) spiral. “The confidence we built in drips over the years was lost in buckets within hours, and it was a gut punch,” Shawn Henry, chief security officer of CrowdStrike, wrote on LinkedIn early Monday. “But this pales in comparison to the pain we’ve caused our customers and our partners. We let down the very people we committed to protect.”

Cloud platform outages and other software issues—including malicious cyberattacks—have caused major IT outages and global disruption before. But last week’s incident was particularly noteworthy for two reasons. First, it stemmed from a mistake in software meant to aid and defend networks, not harm them. And second, resolving the issue required hands-on access to each affected machine; a person had to manually boot each computer into Windows’ Safe Mode and apply the fix.

IT is often an unglamorous and thankless job, but the CrowdStrike debacle has been a next-level test. Some IT professionals had to coordinate with remote employees or multiple locations across borders, walking them through manual resets of devices. One Indonesia-based junior system administrator for a fashion brand had to figure out how to overcome language barriers to do so. “It was daunting,” he says.

“We aren’t noticed unless something wrong is happening,” one system administrator at a health care organization in Maryland told WIRED.

That person was awoken shortly before 1:00 am EDT. Screens at the organization’s physical sites had gone blue and unresponsive. Their team spent several early morning hours bringing servers back online, and then had to set out to manually fix more than 5,000 other devices within the company. The outage blocked phone calls to the hospital and upended the system that dispenses medicine—everything had to be written down by hand and run to the pharmacy on foot.

Most Popular

The Big StoryHow Soon Might the Atlantic Ocean Break? Two Sibling Scientists Found an Answer—and Shook the WorldBy Sandra UpsonGearGive Your Back a Break With Our Favorite Office ChairsBy Julian ChokkattuBusinessSearchGPT Is OpenAI’s Direct Assault on GoogleBy Reece RogersSecurityJ.D. Vance Left His Venmo Public. Here’s What It ShowsBy Dhruv MehrotraComplicating the already nightmarish process were staff shortages. The health care system’s technical staff has been cut in recent years, the system administrator says, and that pushed remaining employees to pull 12- to 14-hour days. If they burn out, there is no one to step in. “We all care about the community that we’re serving—we want to make sure that everything is functioning and that people are being taken care of,” the Maryland administrator says. “But it’s really hard to do that when you don’t have enough staff.”

One chief information security officer of a large health system in the US Midwest emphasized that it isn’t uncommon in health care for budgets to be so tight that organizations have to choose between hiring clinical staff and hiring IT support.

Making everything more difficult were impacted PCs encrypted with the Windows security feature BitLocker. “If you're using BitLocker, jump off a bridge,” a well-known malware analysis and security news account quipped on X on Friday. In the glitched state, users couldn’t enter the BitLocker keys needed to unlock devices and apply the fix without resorting to complicated workarounds. Microsoft released a recovery tool on Saturday that includes a fix for the issue.

The Midwestern CISO says that even though his employer isn’t a CrowdStrike customer, his team still had to manually address issues on about 120 computers that were running the affected software. But his organization’s biggest disruptions came from partners and other third parties that were directly affected and dealing with outages.

“Medicaid eligibility was down,” he says. “Social Security eligibility was down. Local towns that we do business with were down. And I talked to people at other health systems—this blindsided IT departments to the point that it was all hands on deck. There were nontechnical people running around with USB flash drives doing the fixes at some of America’s largest hospital systems.”

CrowdStrike said on Monday that “a significant number” of the 8.5 million impacted devices “are back online and operational.” And IT professionals tell WIRED, that after a grueling few days, the majority of their organizations’ systems have been restored. But it will take time to reach every machine, everywhere. And the situation has raised deeper questions about how monitoring software is designed and the interconnections of today’s digital systems.

“All it takes is one infrastructure vendor like CrowdStrike,” the health care CISO says. “What’s happened is companies have an emphasis on getting things into production and not taking people off the floor to train them about what to do if something goes down.”

He says that the health care system he works for is back to normal now, but not all of its partners are. “If I’m a CrowdStrike customer,” he says, “the first thing I’m worried about is whether that company is even going to continue to exist in its current form after this.”

**You Might Also Like …**

Politics Lab: Get the newsletter and listen to the podcast

What happens when you give people free money

Not everyone loses weight on Ozempic

The Pentagon wants to spend $141 billion on a doomsday machine

Event: Join us for the Energy Tech Summit on October 10 in Berlin

Amanda Hoover is a general assignment staff writer at WIRED. She previously wrote tech features for Morning Brew and covered New Jersey state government for The Star-Ledger. She was born in Philadelphia, lives in New York, and is a graduate of Northeastern University.Staff WriterXLily Hay Newman is a senior writer at WIRED focused on information security, digital privacy, and hacking. She previously worked as a technology reporter at Slate magazine and was the staff writer for Future Tense, a publication and project of Slate, the New America Foundation, and Arizona State University. Additionally... Read moreSenior WriterXTopicsMicrosoftcybersecurityIThealthcareThe Global CrowdStrike Outage Triggered a Surprise Return to CashThe event caused chaos at airports, grocery stores, and Starbucks outlets.Caroline HaskinsThe EU Is Coming for X’s Paid Blue ChecksElon Musk’s paid blue-check system on X deceives users and can be abused by malicious actors, the European Union said today.Morgan MeakerOpenAI Is Testing Its Powers of PersuasionSam Altman is touting AI’s ability to sway people's behavior. His company is also wrestling with the risks.Will KnightNew Jersey’s $500 Million Bid to Become an AI EpicenterThe Garden State has enacted a hefty new tax credit specifically for AI businesses. But tax incentives—particularly for data centers—don’t always create a lot of jobs.Amanda HooverToyota Pulls Off a Fast and Furious Demo With Dual Drifting AI-Powered Race CarsAlgorithms designed to handle a car after it loses traction could potentially intervene on behalf of human drivers.Will KnightWe Asked AI to Take Us On a Tour of Our Cities. It Was ChaosWe had a specialty chatbot to curate perfect days out in London and New York for under $100 each. We're still recovering from our journeys.Natasha BernalGoogle DeepMind's Chatbot-Powered Robot Is Part of a Bigger RevolutionRobotics researchers are exploring how large language models can give physical machines more smarts.Will KnightClass Action Lawsuit Alleges T-Mobile Broke Its Lifetime Price GuaranteeCustomers of the uncarrier want T-Mobile to pay up after switching their legacy phone plans for something more expensive.Jon Brodkin, Ars Technica
27/07/2024 08:36 – SearchGPT Is OpenAI’s Direct Assault on Google
#OpenAI #ChatGPT #MicrosoftBing #ArtificialIntelligence #SearchEngine

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
Save this story

SaveSave this storySaveAfter months of speculation about its search ambitions, OpenAI has revealed SearchGPT, a “prototype” search engine that could eventually help the company tear off a slice of Google's lucrative business.

OpenAI said that the new tool would help users find what they are looking for more quickly and easily by using generative AI to gather links and answer user queries in a conversational tone. SearchGPT could eventually be integrated into OpenAI’s popular ChatGPT chatbot. In addition to a broader web search, the search engine will tap into information provided by publishers who have signed deals giving OpenAI access to their data.

Kayla Wood, a spokesperson for OpenAI, declined to provide a SearchGPT demo or an interview about the new tool for WIRED, but confirmed that the company has already given access to unnamed partners and publishers and improved aspects of the search engine based on their feedback.

Microsoft, an investor in OpenAI, was one of the first companies to release a generative AI search engine to the public when it launched an AI-powered version of Bing back in 2023 that relied on OpenAI’s large language models. That AI search experience from Microsoft has since been rebranded to Copilot.

Since then, multiple competitors, like Google and Perplexity, have launched their own AI search experiences for users. Google’s AI Overviews provide AI-generated summaries of articles, often at the top of news results. OpenAI’s SearchGPT appears more similar to Perplexity's approach, where the chatbot provides an accompanying list of relevant links and the user can ask follow-up questions.

After OpenAI first introduced ChatGPT in November 2022, early users saw in the chatbot’s ability to dig up and summarize information from the web a potential replacement for conventional web search. The shortcomings of large language models make chatbots imperfect search tools, however. The models draw on training data that is often months or years out of date, and when unsure of an answer they will make up facts.

Microsoft’s early efforts with Bing were far from a success, with the AI-powered search engine producing strange, inappropriate, and incorrect answers. Bing’s market share grew only slightly following the overhaul.

When Google added AI Overviews to search results this May, the company also quickly ran into reliability problems, like recommending people add glue to pizza. OpenAI’s SearchGPT may use an approach to generative AI, called retrieval augmented generation, that is an industry standard for AI search and designed to lower the rate of hallucinations in chatbot answers. With a RAG approach, the AI tool references trusted information, like a preferred news website, while generating its output and links back to where the data originated.

There’s also the question of potential copyright violations. Perplexity in particular has been criticized by publications, including WIRED, for copying aspects of original journalism with its AI search tool and seeming to ignore requests not to take content from some websites. In OpenAI’s blog post, the company mentions its commitment to publishers: “SearchGPT is designed to help users connect with publishers by prominently citing and linking to them in searches.” Multiple companies, including Vox Media, The Atlantic, News Corp, and the Financial Times, have all signed licensing agreements with OpenAI this year.

Most Popular

The Big StoryHow Soon Might the Atlantic Ocean Break? Two Sibling Scientists Found an Answer—and Shook the WorldBy Sandra UpsonGearGive Your Back a Break With Our Favorite Office ChairsBy Julian ChokkattuSecurityJ.D. Vance Left His Venmo Public. Here’s What It ShowsBy Dhruv MehrotraGearSonos Admits Its Recent App Update Was a Colossal MistakeBy Adrienne SoDespite the runaway success of ChatGPT, OpenAI is looking for new revenue streams as it pours money into training ever-larger AI models. A report by The Information estimates that the company could lose $5 billion this year due to the cost of hiring more staff and launching new commercial offerings as well as training more powerful AI.

Entering the world of web search may also help OpenAI counter key rival Google. After being left flat-footed in this space, the search giant quickly developed a competitor to ChatGPT called Gemini. It has also ramped up efforts to fold the technology into commercial offerings, including its office software and Android mobile operating system, in addition to search.

Richard Socher, CEO of You.com, a tool that uses generative AI to surface information from the web, says providing a search tool that is measurably better than what people are used to is incredibly difficult. He says for this reason You.com has pivoted to being more of a business-focused research tool. Socher adds that OpenAI's pivot from a focus on developing artificial general intelligence to more commercial ventures will bring its own challenges. "It's a very different mindset," he says.

**You Might Also Like …**

Politics Lab: Get the newsletter and listen to the podcast

What happens when you give people free money

Not everyone loses weight on Ozempic

The Pentagon wants to spend $141 billion on a doomsday machine

Event: Join us for the Energy Tech Summit on October 10 in Berlin

Reece Rogers is WIRED's service writer, focused on explaining crucial topics and helping readers get the most out of their technology. Prior to WIRED, Reece covered streaming at Business Insider.Service WriterWill Knight is a senior writer for WIRED, covering artificial intelligence. He writes the Fast Forward newsletter that explores how advances in AI and other emerging technology are set to change our lives—sign up here. He was previously a senior editor at MIT Technology Review, where he wrote about fundamental... Read moreSenior WriterXTopicsartificial intelligenceOpenAIGoogleSearchsearch enginesOpenAI Slashes the Cost of Using Its AI With a ‘Mini’ ModelWith competing models—including many free ones—flooding the market, OpenAI is announcing a cheaper way to use its AI.Will KnightOpenAI Is Testing Its Powers of PersuasionSam Altman is touting AI’s ability to sway people's behavior. His company is also wrestling with the risks.Will KnightWe Asked AI to Take Us On a Tour of Our Cities. It Was ChaosWe had a specialty chatbot to curate perfect days out in London and New York for under $100 each. We're still recovering from our journeys.Natasha BernalAI Can’t Replace Teaching, but It Can Make It BetterEven techno-optimists hesitate to say teaching is best left to the bots, but there’s a debate about where to draw the line.Chris BerdikDonald Trump and Silicon Valley's Billionaire ElegyVenture capitalists Ben Horowitz and Marc Andreessen claim the tech industry, California, and the country are doomed if we don’t embrace the former president.Steven LevyThe ACLU Fights for Your Constitutional Right to Make DeepfakesStates across the US are seeking to criminalize certain uses of AI-generated content. Civil rights groups are pushing back, arguing that some of these new laws conflict with the First Amendment.Arthur Holland MichelHow IT Departments Scrambled to Address the CrowdStrike ChaosSystem administrators and their teams sprang into action last Friday and worked tirelessly to get their companies back online amid one of the worst digital blackouts ever.Amanda HooverNew Jersey’s $500 Million Bid to Become an AI EpicenterThe Garden State has enacted a hefty new tax credit specifically for AI businesses. But tax incentives—particularly for data centers—don’t always create a lot of jobs.Amanda Hoover
27/07/2024 08:26 – Novel algorithm for discovering anomalies in data outperforms current software
#ArtificialIntelligence #AnomalyDetection #Data #Computing #Technology #ComputerScience #CognitiveScience #BranchesOfScience #InformationScience #InformationTechnology #Cognition #ExplainableArtificialIntelligence #Algorithm #Learning #Cyberspace #AppliedMathematics #ComputationalNeuroscience #Information #Science #KnowledgeManagement #InformationTechnologyManagement #HumanCommunication #Misinformation #Communication #Intelligence #InformationAge

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
***NOVEL ALGORITHM FOR DISCOVERING ANOMALIES IN DATA OUTPERFORMS CURRENT SOFTWARE***

by Tina Hilding, Washington State University

Anomaly detection poses several challenges that are not seen in traditional AI problems. Credit: Trevor Vannoy on Unsplash

An algorithm developed by Washington State University researchers can better find data anomalies than current anomaly-detection software, including in streaming data.

The work, reported in the Journal of Artificial Intelligence Research, makes fundamental contributions to artificial intelligence (AI) methods that could have applications in many domains that need to quickly find anomalies in large amounts of data, such as in cybersecurity, power grid management, misinformation, and medical diagnostics.

Being able to better find the anomalies would mean being able to more easily discover fraud, disease in a medical setting, or important unusual information, such as an asteroid whose signals overlap with the light from other stars.

"This work presents advances on how AI and humans can work together to synergistically solve anomaly discovery problems," said Jana Doppa, Huie-Rogers Endowed Chair Associate Professor of Computer Science who supervised the work.

"With all this generative AI technology, there is so much data which includes misinformation, and if you want humans to go over all of this, it's impossible as it's huge. If you have finite human resources, and you want to detect something like misinformation quickly, you want algorithms that prioritize which items should be labeled."

Anomaly detection poses several challenges that are not seen in traditional AI problems. The number of anomalies is very small compared to the normal data—typically less than 2%. Furthermore, there may not be a big difference between an anomaly and normal data.

"So, it's like finding needles in a big haystack kind of a problem," said Doppa. "And you don't even know in a lot of domains what needles to look for."

Another problem is that with large amounts of data, AI will oftentimes find too many candidate anomalies to pass along for people to check.

"Whenever you have these false positives, you are wasting a lot of humans' time, which we want to minimize," said postdoctoral researcher and lead author Shubhomoy Das. "How can we use minimal feedback from the human to adapt the anomaly detector so that the false positives go down over time, and we discover more and more diverse anomalies?"

As part of the work, the researchers provided new theoretical and empirical findings for why an ensemble of computer models worked well for anomaly discovery. They found that with just a small amount of step-by-step feedback, the AI algorithm can learn much better and discover many more diverse anomalies compared to a system where there was no feedback. The human needs an explanation regarding the candidate anomalies to understand why AI selected them for labeling.

"Some notion of interpretability or explainability is important," said Ph.D. student and co-author Rakibul Islam. "What we figured was that this was largely missing in the existing literature."

The researchers used their new findings to develop an algorithm that looks at anomalies in batches, which improved the ability to discover diverse types of anomalies. So, in the case of anomalous credit card data, the algorithm discovers different types of unusual behavior, such as a person's oddly expensive purchases and/or ones that are made in an odd location.

Unlike current AI models, the algorithm the researchers developed was able to handle streaming data, which is common in many real-world applications. Their algorithm can detect and quantify drift in the data distribution and then take corrective action.

"The problem of discovering anomalies when the data is coming in a stream was less studied," Doppa said.

The researchers' computer code and data are publicly available, and they now plan to deploy their algorithms in real-world systems to measure their accuracy and usability.

More information:

Shubhomoy Das et al, Effectiveness of Tree-based Ensembles for Anomaly Discovery: Insights, Batch and Streaming Active Learning, Journal of Artificial Intelligence Research (2024). DOI: 10.1613/jair.1.14741

Provided by

Washington State University

Citation:

Novel algorithm for discovering anomalies in data outperforms current software (2024, July 26)

retrieved 27 July 2024

from https://techxplore.com/news/2024-07-algorithm-anomalies-outperforms-current-software.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no

part may be reproduced without the written permission. The content is provided for information purposes only.

Explore further

Machine learning algorithms can find anomalous needles in cosmic haystacks

17 shares

Generative AI creates personalized storybooks for the future of child language learning

12 hours ago

Study explores win–win potential of grass-powered energy production

13 hours ago

Deep learning models can be trained with limited data: New method could reduce errors in computational imaging

14 hours ago

Experts warn against hype for deriving green hydrogen from direct seawater electrolysis

15 hours ago

New microgrids model takes into account a fair design of decentralized energy systems

15 hours ago

Engineers develop magnetic tunnel junction–based device to make AI more energy efficient

16 hours ago

Robot Spot configured to find and stun weeds using a blowtorch

16 hours ago

Magnetic fields help understand light particle splitting for boosting solar cell efficiency

17 hours ago

OpenAI to challenge Google with new search functionality

A cool solution: Designing a next-generation cooling system to help data centers become more energy efficient

Related Stories

Machine learning algorithms can find anomalous needles in cosmic haystacks

Researchers review deep learning-based methods to detect time series data anomaly

Enhancing defect detection performance in smart factories

Study finds no increased risk of birth defects after COVID-19 infection or vaccination in early pregnancy

New algorithm rapidly finds anomalies in gene expression data

AI algorithm keeps a mile-long particle accelerator healthy

Recommended for you

Engineers develop magnetic tunnel junction–based device to make AI more energy efficient

16 hours ago

Robot Spot configured to find and stun weeds using a blowtorch

16 hours ago

Generative AI creates personalized storybooks for the future of child language learning

12 hours ago

Deep learning models can be trained with limited data: New method could reduce errors in computational imaging

14 hours ago

OpenAI to challenge Google with new search functionality

Using AI to train AI: Model collapse could be coming for LLMs, say researchers

Load comments (0)

Let us know if there is a problem with our content

E-mail the story

Novel algorithm for discovering anomalies in data outperforms current software
27/07/2024 08:20 – Generative AI creates personalized storybooks for the future of child language learning
#Language #ArtificialIntelligence #InternetOfThings #Cognition #Technology #Learning #Speech #Intelligence #AppliedLinguistics

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
***GENERATIVE AI CREATES PERSONALIZED STORYBOOKS FOR THE FUTURE OF CHILD LANGUAGE LEARNING***

by Pohang University of Science and Technology

The Open Sesame? Open Salami! system profiles each child's daily language environment by automatically extracting target vocabulary based on speech pathology principles. It then generates personalized stories that seamlessly incorporate these target words, providing customized storybooks to support effective language assessment and intervention. Credit: POSTECH

Researchers have created an innovative system for generating personalized storybooks. This system utilizes generative artificial intelligence and home IoT technology to assist children in language learning. The findings are published in the journal Proceedings of the CHI Conference on Human Factors in Computing Systems.

Their research was showcased at the ACM CHI (ACM SIGCHI Conference on Human Factors in Computing Systems), where it earned an "Honorable Mention Award," recognizing it as one of the top 5% of submissions.

Children's language development is crucial as it impacts their cognitive and academic growth, their interactions with peers, and overall social development. It is essential to regularly evaluate language progress and provide timely language interventions to support language acquisition.

The issue is that children grow up in diverse environments, leading to variations in their exposure to vocabulary. However, traditional approaches often rely on standardized vocabulary lists and pre-made storybooks or toys for language skill assessments and interventions, lacking the diversity diverse support.

Recognizing the shortcomings of conventional, one-size-fits-all approaches that fail to address the diverse backgrounds of children, the team created an innovative educational system tailored to each child's unique environment. They began by employing home IoT devices to capture and monitor the language children hear and speak in their daily lives.

Through speaker separation and morphological analysis techniques, the researchers examined the vocabulary children were exposed to, the words they spoke, and those they heard but did not vocalize. They then assessed each word by calculating scores for each word based on key factors relevant to speech pathology.

To create personalized educational materials, the team utilized advanced generative AI technologies, including GPT-4 and Stable Diffusion. This enabled them to produce custom children's books that seamlessly integrate the target vocabulary for each individual child. By combining speech pathology theory with practical expertise, the researchers developed an effective and personalized language learning system.

The researchers designed the system to accommodate variations in children's language development by allowing for individualized weighting of factors and flexible vocabulary selection criteria. The system can automate both the extraction of target vocabulary for each child and the creation of personalized storybooks, ensuring that both the vocabulary and the storybooks can be continuously updated in response to changes in the child's language development and environment.

After testing the system in 9 families over a four-week period, the results showed that children effectively learned the target vocabulary, demonstrating the system's applicability in everyday settings beyond the therapy room.

Inseok Hwang from the Department of Computer Science and Engineering, along with students Jungeun Lee, Suwon Yoon, and Kyoosik Lee from the Department of Computer Science and Engineering at POSTECH in collaboration with Professor Dongsun Yim from Ewha Womans University's Department of Communication Disorders collaborated on the project.

Jungeun Lee from POSTECH, the lead author of the paper, said, "We effectively addressed the limitations of traditional, one-size-fits-all approaches to child language assessment and intervention by using generative AI. Our goal is to leverage AI to create customized guides tailored to different individuals' levels and needs."

Professor Hwang from POSTECH, the corresponding author, remarked, "Through interdisciplinary research, we have successfully developed a personalized language stimulation and development system that integrates generative AI technology with speech pathology theory. We hope our findings will encourage educators to respect and incorporate the diverse environments and learning goals of children."

Co-author Professor Yim from Ewha Womans University said, "Our work demonstrates the potential for non-traditional, personalized language support services. The system showcases the ability to tailor target vocabulary extraction and linguistic stimuli delivery for children exposed to varied environments and languages."

More information:

Jungeun Lee et al, Open Sesame? Open Salami! Personalizing Vocabulary Assessment-Intervention for Children via Pervasive Profiling and Bespoke Storybook Generation, Proceedings of the CHI Conference on Human Factors in Computing Systems (2024). DOI: 10.1145/3613904.3642580

Provided by

Pohang University of Science and Technology

Citation:

Generative AI creates personalized storybooks for the future of child language learning (2024, July 26)

retrieved 27 July 2024

from https://techxplore.com/news/2024-07-generative-ai-personalized-storybooks-future.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no

part may be reproduced without the written permission. The content is provided for information purposes only.

Explore further

Study recommends exposing deaf children to sign language before and after cochlear implantation

20 shares

Study explores win–win potential of grass-powered energy production

13 hours ago

Novel algorithm for discovering anomalies in data outperforms current software

13 hours ago

Deep learning models can be trained with limited data: New method could reduce errors in computational imaging

14 hours ago

Experts warn against hype for deriving green hydrogen from direct seawater electrolysis

15 hours ago

New microgrids model takes into account a fair design of decentralized energy systems

15 hours ago

Engineers develop magnetic tunnel junction–based device to make AI more energy efficient

16 hours ago

Robot Spot configured to find and stun weeds using a blowtorch

16 hours ago

Magnetic fields help understand light particle splitting for boosting solar cell efficiency

17 hours ago

OpenAI to challenge Google with new search functionality

A cool solution: Designing a next-generation cooling system to help data centers become more energy efficient

Related Stories

Study recommends exposing deaf children to sign language before and after cochlear implantation

Q&A: New research unveils three distinct language comprehension phenotypes in autistic children

Why learning animal sounds can be crucial to children's language development

Early vocabulary size is genetically linked to ADHD, literacy, and cognition

Research explores early predictors of children's language delays

Use of digital devices may affect children's language development

Recommended for you

Engineers develop magnetic tunnel junction–based device to make AI more energy efficient

16 hours ago

Robot Spot configured to find and stun weeds using a blowtorch

16 hours ago

Novel algorithm for discovering anomalies in data outperforms current software

13 hours ago

Deep learning models can be trained with limited data: New method could reduce errors in computational imaging

14 hours ago

OpenAI to challenge Google with new search functionality

Using AI to train AI: Model collapse could be coming for LLMs, say researchers

Load comments (0)

Let us know if there is a problem with our content

E-mail the story

Generative AI creates personalized storybooks for the future of child language learning
27/07/2024 03:44 – AI and the Future of News: Here Come the Robots
#ArtificialIntelligence #News #Technology #Journalism #NewsMedia #MassMedia #Robot #Data #SocialIssues

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
Download Publication

View this Publication

Download PDF

There is palpable anxiety among the news media about the rapid advance of artificial intelligence, and its impact on business. And yet, the industry ignores these emerging technologies at its peril. We call on the media to educate themselves on both the risks, and the opportunity – to use AI to empower journalism and meet evolving consumer needs.

Throughout 2024, Aspen Digital has been leading the conversation with newsroom and publishing leaders, AI executives and technologists, and other experts on how the news media might experiment with AI tools, while adhering to journalistic values and business imperatives.

“Here Come the Robots” synthesizes these insights and makes recommendations for how AI technologies can lead to the long-term success of news organizations, now and in the future.

A sampling of the report’s guidance includes:

AI as a Tool, Not a Destiny: AI should be seen as a potentially transformative tool that can complement and elevate journalistic efforts, rather than replace traditional news sources.

Opportunities for Enhancement: AI can support data journalism, enhance reader experience, benefit audience engagement, and foster expanded markets.

Experimentation and Adaptation: Successful integration of AI requires a culture of experimentation, and adapting to changing audience habits.

Ethical and Transparent Integration: Ensuring transparency, adhering to ethical guidelines, and keeping a human in the loop is crucial to maintaining public trust and organizational integrity.

Serving the Public: The ultimate goal of leveraging AI tools is to better serve the civic information space by empowering more diverse and insightful reporting. All AI experiments must be in service to that mission.

Related Posts

Communications

The “Cybersecurity in Daily Life” Poster Competition

July 11, 2024

Katie D’Hondt Brooks

& 1 more

Communications

Calling for 2023’s Best Reporting on AI

January 8, 2024

Tom Latkowski

Blog Post

Looking Back on the “Future…
25/07/2024 23:49 – En difficulté aux Etats-Unis, Stellantis enregistre une baisse de 14% de son chiffre d'affaires
#Stellantis #PSAGroup #MotorVehicles #ElectricVehicleManufacturersOfTheNetherlands #HoldingCompaniesOfTheNetherlands #Cars #CarlosTavares #MotorVehicleEngineManufacturers #Technology #Vehicles

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
Stellantis

Résultats financiers

***EN DIFFICULTé AUX ETATS-UNIS, STELLANTIS ENREGISTRE UNE BAISSE DE 14% DE SON CHIFFRE D'AFFAIRES***

Après plusieurs années de résultats stratosphériques tirés par une "premiumisation" du marché automobile, le constructeur automobile Stellantis a enregistré un recul de 14% de son chiffre d'affaires au premier semestre.

Antoine Vermeersch

\

1

min. de lecture

**Mon Actualité personnalisable**

Antoine Vermeersch

\

© Stephane Sby Balmy

Carlos Tavares, pdg de Stellantis, ne cache pas que les résultats semestriels de son entreprise ne sont pas "à la hauteur".

Sur le fil. Malmené sur ses deux principaux marchés américain et européen depuis le début de l'année, le constructeur automobile Stellantis est parvenu à maintenir une marge opérationnelle courante à deux chiffres au premier semestre, à tout juste 10%. Mais loin de son stratosphérique 14,4% enregistré sur la même période un an plus tôt.

Son chiffre d'affaires accuse un recul de 14% et s'établit à 85,02 milliards d'euros. Le bénéfice net chute plus lourdement encore de 48% et s'élève à 5,6 milliards d'euros, loin du bénéfice record de 10,9 milliards réalisé sur les six premiers mois de l'année 2023.

"Les résultats de l'entreprise au premier semestre ne sont pas à la hauteur de nos attentes, reflétant à la fois un contexte industriel difficile et nos difficultés opérationnelles", a déclaré le directeur général Carlos Tavares. Dans un communiqué, Stellantis justifie sa moindre performance par un marché automobile à la peine et le renouvellement en cours de sa gamme. Le groupe prévoit de commercialiser vingt nouveaux véhicules dans le monde au cours de l'année 2024, avec de nouvelles versions des Peugeot 3008 et 5008 ou de la Citroën C3, par exemple.

**Difficultés Sur Le Marché Nord-américain**

Depuis plusieurs mois, le constructeur franco-italo-américain est en difficulté sur son principal marché nord-américain, où sa part de marché régresse. Dans cette région, ses livraisons en volume ont chuté de 18%. En Europe, les volumes sont en baisse de 6%. Dans ce contexte incertain, Stellantis souligne l'amélioration de sa performance interne en matière de "réductions des coûts directs des matériaux, de la main-d'œuvre et de la logistique, qui ont permis d'atténuer la baisse du chiffre d'affaires". Certains observateurs s'inquiètent des capacités de Stellantis à continuer de réduire ses coûts.

Malgré ces résultats jugés décevants par le marché, mais qui restent de très bonne facture pour un constructeur généraliste, Stellantis maintient ses objectifs financiers pour l'année, avec un free cash-flow industriel en terrain positif et une marge opérationnelle d'au moins 10%.

** Nissan Voit Ses Bénéfices Plonger De... 99% Et Revoit à La Baisse Ses Objectifs Financiers 2024 **

** La Future Renault Twingo électrique «abordable» Sera Fabriquée En Slovénie à Partir De 2026 **

** Renault Améliore Encore Sa Marge Opérationnelle Au Premier Semestre 2024 **

**SUR LE MÊME SUJET**

Stellantis va vendre des modèles du constructeur chinois Leapmotor en Europe dès septembre 2024

Stellantis lancera 36 modèles hybrides abordables d'ici 2026

Avec sa nouvelle Panda, Fiat entre dans la bataille des véhicules électriques à moins de 25 000 euros

Le bénéfice de Stellantis s’envole et établit un nouveau record

**Sujets associés**

Stellantis

Résultats financiers

Nos journalistes sélectionnent pour vous les articles essentiels de votre secteur.

Groupe Moniteur Nanterre B 403 080 823, IPD Nanterre 490 727 633, Groupe Industrie Service Info (GISI) Nanterre 442 233 417.

Cette société ou toutes sociétés du Groupe Infopro Digital pourront l'utiliser afin de vous proposer pour leur compte ou celui de leurs

clients, des produits et/ou services utiles à vos activités professionnelles.

Pour exercer vos droits, vous y opposer ou pour en savoir plus : Charte des données personnelles.

Assystem forme les demandeurs d’emploi à l’ingénierie nucléaire

Afin d’atteindre la neutralité carbone d’ici à 2050, le gouvernement français prévoit la construction d’au moins 6 EPR2 et la prolongation de la durée de vie des réacteurs nucléaires déjà en exploitation. Pour mener ce grand chantier indispensable à la transition énergétique, 100 000 personnes devraient être recrutées d'ici à 2030 dans la filière nucléaire française. Chez Assystem, le premier groupe d’ingénierie nucléaire en Europe, cette dynamique implique de doubler le nombre d'experts du nucléaire.

Ingénieur / Ingénieure Coordination Technique de Projets de Transports Urbains

ARTELIA - 23 July 2024 - CDI - Choisy-le-Roi

Planification Investissements Pilotage Projets F/H

Groupe ADP / PARIS AEROPORT - 23 July 2024 - CDI - Roissy-en-France

LES ÉVÉNEMENTS L'USINE NOUVELLE

E-learning Loi Sapin 2

2-30 Septembre 2024

100% digital , live et replay

Sensibilisation à la lutte contre la corruption

Le réemploi des emballages

Analyser le marché, les défis et les contraintes

Fondamentaux - gestion de parc auto

23-24 Septembre 2024

Fondamentaux de la gestion de parc automobile

Trophée

TROPHéES DES FEMMES DE L'INDUSTRIE - #IndustriELLES24

Mettre en lumière les parcours et les carrières exemplaires des femmes de l'industrie

Gouvernance et RH

**Ariane, Ou Le Voyage Maudit De François Mitterrand**

[Industry story - Le podcast] Pour la première fois, un président de la République assiste au décollage d’une fusée Ariane. Mais est-ce réellement une bonne idée...

**Ingénieur, Un Métier Au Coeur De La Souveraineté**

Dans ce nouveau podcast de La Fabrique, nous recevons Anne-Sophie Bellaiche, rédactrice en chef de L'Usine Nouvelle et du Guide de l'ingénieur. Comme son nom l'indique, cette publication annuelle s'intéresse aux différentes...

**L'entrepreneur Qui A Vendu La Bastille En Kit**

[Industry story - Le podcast] Avec la prise de la Bastille, la journée deviendra historique. Un entrepreneur malin comprend tous les bénéfices qu’il peut en tirer.

**L'inventrice Du Premier Lave-vaisselle**

L’épouse d’un bourgeois de l’Illinois décide de prendre les choses en main et d’inventer elle-même l’outil dont les femmes ont besoin.

LES SERVICES DE L'USINE NOUVELLE

Ingénieur / Ingénieure Coordination Technique de Projets de Transports Urbains

ARTELIA - 23/07/2024 - CDI - Choisy-le-Roi

+ 550

93 - Épinay-sur-Seine

Réhabilitation partielle de laboratoires de cinéma argentique pour l'implantation de l'ERP LE NAVIRE ARGO

DATE DE REPONSE 27/09/2024

+ de 10.000

**ARTICLES LES PLUS LUS**

Plus long que l'A380, le Boeing 777-9 en passe de devenir le plus grand appareil commercial du monde

«La qualité est époustouflante» : Sanofi divise par huit le temps passé sur ses contrôles qualité grâce à l'IA générative

Après avoir livré à temps les sites des JO, Nicolas Ferrand va piloter les futurs chantiers d'Orano

La recette qui a sauvé la PME alsacienne Voltec Solar de la déferlante des panneaux solaires chinois

Safran livre le premier drone Patroller de série

A Farnborough, un avion de combat du futur peut en cacher un autre
25/07/2024 23:46 – GenAI : Meta met le paquet sur l'open source avec ses modèles Llama 3.1
#MetaPlatforms #ArtificialIntelligence #GPT3 #Computing #Technology #InformationTechnology #ComputerScience #OpenSource #DigitalTransformation #GPT4 #MachineLearning #CloudComputing #SyntheticData #DataQuality #Computers #DigitalTechnology #InformationTechnologyManagement #OpenAccess

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
Meta

***GENAI : META MET LE PAQUET SUR L'OPEN SOURCE AVEC SES MODèLES LLAMA 3.1***

Meta s'engage à rendre l'IA accessible à tous. C'est en tout cas ce que Mark Zuckerberg explique dans une lettre publiée sur le blog de la firme. Selon lui, l'open source semble être la voie royale pour y accéder et sa dernière itération de modèles Llama 3.1 est une première réponse à ce défi. Son modèle phare comporte pas moins de 405 milliards de paramètres, est doté d'une fenêtre contextuelle de 128 000 tokens et peut rivaliser avec nombre de LLM incluant GPT-4, GPT-4o, et Claude 3.5 Sonnet.

Célia Séramour

\

5

min. de lecture

**Mon Actualité personnalisable**

Célia Séramour

\

© Dima Solomin - Unsplash

Meta s'affirme dans l'écosystème open source avec la publication de trois modèles Llama 3.1.

"Nous dévoilons Llama 3.1 405B, que nous considérons comme le modèle de fondation le plus grand et le plus performant au monde, disponible en libre accès". Meta semble déterminé à gagner la course aux LLM open source. La firme vient de dévoiler une nouvelle itération de sa famille de modèles Llama 3. Cette dernière comprend trois versions respectivement de 8, 70 et 405 milliards de paramètres. Sans surprise, Meta pousse sur le devant de la scène son modèle le plus important, Llama 3.1 405B.

La firme le décrit comme "le premier modèle disponible en libre accès qui rivalise avec les meilleurs modèles d’IA en termes de capacités de pointe dans les domaines de la connaissance générale, de l’orientation, des mathématiques, de l’utilisation d’outils et de la traduction multilingue". L’entraînement de Llama 3.1 405B sur plus de 15 mille milliards de tokens a, sans surprise, constitué un défi majeur. Pour y parvenir, les équipes de Meta ont optimisé la pile d’entraînement complète et poussé l’entraînement du modèle sur plus de 16 000 GPU H100, un exploit pour l'entraînement d'un tel modèle à cette échelle.

**"Nous Avons Amélioré à La Fois La Quantité Et La Qualité Des Données Utilisées"**

Les chercheurs expliquent avoir fait des choix de conception qui visent à maintenir le processus de développement du modèle évolutif et simple. Ils ont donc opté pour une architecture de modèle de type Transformer "uniquement avec des adaptations mineures, plutôt que pour un modèle de mélange d’experts (MoE) afin de maximiser la stabilité de l’apprentissage". S'en est suivi une procédure itérative de post-entraînement, où chaque tour utilise un fine-tuning supervisé et une optimisation directe des préférences.

Ce type d'entraînement offre un moyen de créer des données synthétiques de meilleure qualité pour chaque cycle. "Par rapport aux versions précédentes de Llama, nous avons amélioré à la fois la quantité et la qualité des données que nous utilisons pour le pré- et le post-entraînement". Ces améliorations comprennent le développement de pipelines de pré-traitement et de curation plus minutieux pour les données de pré-entraînement, le développement d’une assurance qualité plus rigoureuse et des approches de filtrage pour les données de post-entraînement. Le modèle 405B a par ailleurs été utilisé pour améliorer la qualité post-entraînement des modèles plus petits de Meta.

Concernant la phase d’inférence pour un modèle à l’échelle du 405B, les équipes ont fait le choix de quantifier les modèles de 16 bits (BF16) à 8 bits (FP8) numériques, afin de réduire considérablement les exigences de calcul nécessaires et faire tourner le modèle dans un seul nœud de serveur.

**Un Modèle Puissant Qui Nécessite Des Ressources De Calcul Et Des Compétences Solides**

Consciente que pour le développeur moyen, l’utilisation d’un modèle à l’échelle de Llama 3.1 405B représente un défi, Meta a mis en place des solutions avec plusieurs partenaires technologiques, notamment AWS, Nvidia et Databricks. "Dès le premier jour, les développeurs peuvent profiter de toutes les capacités avancées du modèle 405B et commencer à construire immédiatement", affirme la société. Meta a également fait appel à Groq qui optimise l’inférence à faible latence pour les déploiements dans le cloud, et à Dell qui réalise des optimisations similaires pour les systèmes sur site.

A terme, Meta cible des usages comme l'inférence en temps réel et par lots, le fine-tuning supervisé, l'évaluation du modèle pour une application spécifique, le pré-entraînement continu, la génération augmentée de récupération (RAG), ou encore la génération de données synthétiques.

**Des Versions à 8 Et 70 Milliards De Paramètres Rejoignent La Liste De LLM**

La famille Llama 3.1 abrite également des versions améliorées des modèles 8B et 70B existants. Ceux-ci sont multilingues et disposent d’une longueur de de contexte nettement plus importante – 128 000 tokens tout comme le modèle 405B – ainsi que de capacités de raisonnement globalement plus solides. Les modèles peuvent ainsi prendre en charge des cas d’utilisation avancés, tels que le résumé de textes longs, les agents conversationnels multilingues et les assistants de codage.

Meta affirme avoir évalué les performances de Llama 3.1 sur environ 150 ensembles de données de référence couvrant un large éventail de langues. En sus, des évaluations humaines approfondies ont été réalisées pour comparer Llama 3.1 à des modèles concurrents dans des scénarios réels. Le modèle 405B semble ainsi compétitif par rapport aux principaux modèles de fondation sur une série de tâches, y compris GPT-4, GPT-4o, et Claude 3.5 Sonnet.

Les deux plus petits modèles sont compétitifs par rapport aux modèles dits fermés et open source qui ont un nombre similaire de paramètres. Cela inclut Gemma 2 9B IT et Mistral 7B Instruct pour la version à 8 milliards de paramètres et Mixtral 8x22B Instruct et GPT 3.5 Turbo pour la version à 70 milliards de paramètres.

**La Voie à Des Modèles Plus Petits Et Open Source**

Avec la sortie du modèle 405B et de manière plus générale avec cette dernière génération de Llama, Meta espère ouvrir la voie à "de nouvelles applications et à de nouveaux paradigmes de modélisation, y compris la génération de données synthétiques pour permettre l’amélioration et l’entraînement de modèles plus petits, ainsi que la distillation de modèles". Se réclamant membre à part entière de la communauté open source, Meta indique que les poids des modèles Llama peuvent être téléchargés.

"Les développeurs peuvent entièrement personnaliser les modèles en fonction de leurs besoins et de leurs applications, s’entraîner sur de nouveaux ensembles de données et procéder à des ajustements supplémentaires". Ils peuvent également les exécuter dans n’importe quel environnement, que ce soit sur site, dans le cloud ou même localement sur un ordinateur portable, le tout sans partager de données avec Meta. La firme indique que ses modèles sont disponibles en téléchargement sur https://llama.meta.com/ et Hugging Face, ainsi que pour le développement immédiat sur les plateformes partenaires.

Preuve ultime de sa bienveillance envers l'écosystème, la firme a tenu à rappeler par la voix de son maître, Mark Zuckerberg, que "l’open source permettra à un plus grand nombre de personnes dans le monde d’avoir accès aux avantages et aux possibilités de l’IA, de ne pas concentrer le pouvoir entre les mains d’un petit nombre et de déployer la technologie de manière plus homogène et plus sûre dans la société". Reste à savoir si "l'IA en accès libre" deviendra réellement une norme pour le secteur à l'avenir.

** GenAI : Citroën Embarque ChatGPT Dans Ses Véhicules Les Plus Récents **

** Mistral AI Pousse Ses Modèles D'IA Sur Google Cloud Afin D'attirer Plus De Clients **

** Danone Signe Un Contrat Pluriannuel Avec Microsoft Pour Intégrer L'IA Dans Ses Activités **

**SUR LE MÊME SUJET**

Avec Llama 3 et son nouvel assistant, Meta veut s'imposer face à OpenAI

"L'intersection d'images et de texte peut nous aider à réduire les hallucinations", Joëlle Pineau, Meta FAIR

"Mon cheval de bataille, c'est la transparence et la disponibilité des modèles", Joëlle Pineau, Meta FAIR

Meta regroupe ses divisions de recherche en IA et se vante d'avoir l'équivalent de 600 000 Nvidia H100

Avec le modèle ouvert Mixtral 8x22B, Mistral AI fait... presque aussi bien que Llama 3

Avec Gemma 2, Google fait de l'ombre à Llama 3

GenAI : Meta met le paquet sur l'open source avec ses modèles Llama 3.1

à la rédaction, celui-ci peut être soumis à modération, auquel cas il

n'apparaitra pas immédiatement sur le site.

Participez à la discussion !

0Commentaire

**Sujets associés**

Meta

Open Source

Nos journalistes sélectionnent pour vous les articles essentiels de votre secteur.

Groupe Moniteur Nanterre B 403 080 823, IPD Nanterre 490 727 633, Groupe Industrie Service Info (GISI) Nanterre 442 233 417.

Cette société ou toutes sociétés du Groupe Infopro Digital pourront l'utiliser afin de vous proposer pour leur compte ou celui de leurs

clients, des produits et/ou services utiles à vos activités professionnelles.

Pour exercer vos droits, vous y opposer ou pour en savoir plus : Charte des données personnelles.

LES ÉVÉNEMENTS USINE DIGITALE

Maîtriser les enjeux Privacy de l'IA : une nécessité incontournable

Décrypter les nouvelles réglementations européennes liées à l'IA et leurs impacts opérationnels

Cyber Sécurité

Cycle

Parcours métier Data Protection Officer (DPO)

23-27 Septembre 2024

Acquérir l'ensemble des compétences nécessaires aux DPO

Cyber Sécurité

S'approprier le rôle et les missions du Délégué à la Protection des Données (DPO)

Répondre à l'exigence 2.16 du référentiel de compétences des DPO

Cyber Sécurité

CHERCHE TALENTS NUMERIQUE

Assystem forme les demandeurs d’emploi à l’ingénierie nucléaire

Afin d’atteindre la neutralité carbone d’ici à 2050, le gouvernement français prévoit la construction d’au moins 6 EPR2 et la prolongation de la durée de vie des réacteurs nucléaires déjà en exploitation. Pour mener ce grand chantier indispensable à la transition énergétique, 100 000 personnes devraient être recrutées d'ici à 2030 dans la filière nucléaire française. Chez Assystem, le premier groupe d’ingénierie nucléaire en Europe, cette dynamique implique de doubler le nombre d'experts du nucléaire.

Emploi

Expert BI (H/F)

Free-work - 24 July 2024 - - Lyon, Auvergne-Rhône-Alpes

Développeur Mobile Android

Free-work - 24 July 2024 - - Lyon, Auvergne-Rhône-Alpes

Les formations USINE DIGITALE

élaborer une charte éthique d'utilisation des Intelligences Artificielles Génératives (IAG)

Supervision des IA dans son organisation

évaluez la conformité de vos usages IA

Comment implémenter l'IA Act dans vos process

Serious game : adopter la nouvelle directive NIS2

18/09 Matin

Maitrisez les évolutions de la cybersécurité

DPO : Supervisez les usages de l'IA de votre structure et des collaborateurs

Décrypter les nouvelles réglementations européennes liées à l'IA et leurs impacts opérationnels

Toutes les formations

**ARTICLES LES PLUS LUS**

CrowdStrike : pourquoi la mise à jour d'un antivirus a paralysé les entreprises du monde entier

Mistral AI et Nvidia dévoilent leur premier modèle commun, Mistral NeMo 12B

La pénurie de talents en cybersécurité, une épine dans le pied des entreprises françaises

Comment les ordinateurs quantiques nous aideront à résoudre des problèmes complexes

Alléger l'empreinte carbone de l'intelligence artificielle : rêve ou réalité ?

Atos échappe au naufrage, sauvé par un accord définitif avec banques et créanciers
25/07/2024 23:44 – Conduite autonome : Alphabet injecte 5 milliards de dollars dans sa filiale Waymo
#SelfDrivingCar #AlphabetInc #Waymo #Vehicles #AutomotiveIndustry #MotorVehicles #Technology #Transport #Cars #AutomotiveTechnologies #GeneralMotors #TeslaInc #VehicleTechnology #VehicleIndustry #DigitalTransformation #LandTransport #RoadTransport #RoadTrafficSafety #LandVehicles #CarManufacturersOfTheUnitedStates #TransportEconomics #UnitedStates

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
Véhicule autonome

Waymo

Investissement

***CONDUITE AUTONOME : ALPHABET INJECTE 5 MILLIARDS DE DOLLARS DANS SA FILIALE WAYMO***

Alphabet, qui étalera son investissement sur plusieurs années, compte faire de Waymo la “première société de technologie de conduite autonome au monde”. Un montant colossal, à l'heure où les enquêtes de sécurité sur les principaux fabricants du secteur se multiplient.

Yoann Bourgin

\

2

min. de lecture

**Mon Actualité personnalisable**

Yoann Bourgin

\

© Waymo

À gauche, une Chrysler Pacifica équipée du système de conduite autonome de Waymo, modèle dont la société a annoncé le retrait l'année dernière. À droite, une Jaguar I-Pace, modèle toujours en circulation chez Waymo.

À l’occasion de la publication trimestrielle des résultats financiers d’Alphabet, Ruth Porat, directrice financière du groupe, a annoncé un investissement pluriannuel de 5 milliards de dollars (4,6 milliards d’euros) dans la filiale de conduite autonome Waymo. “Ce nouveau financement, cohérent avec les investissements annuels récents, permettra à Waymo de continuer à construire la première société de technologie de conduite autonome au monde”, a déclaré Ruth Porat lors d’une conférence téléphonique.

**Plus De 50 000 Trajets Hebdomadaires**

L’investissement s’inscrit dans la volonté d’Alphabet de rendre plus efficace sa division “Other Bets”. Ce segment comprend Waymo mais aussi d’autres projets dits “innovants”, comme les groupes de recherche biologique Calico et Verily, le fonds de placement Google Ventures et la société de livraison par drone Wing. La division a généré un chiffre d’affaires trimestriel de 365 millions de dollars, en hausse de 28% sur un an, mais se heurte également à une hausse de ses pertes (1,13 milliard de dollars, +39%). De plus amples informations seront données lors du dépôt du formulaire 10-Q auprès de la SEC, le gendarme financier américain.

Sundar Pichai, CEO d’Alphabet, a affirmé que Waymo réalise actuellement plus de 50 000 trajets hebdomadaires. La société de conduite autonome propose d’ores et déjà des trajets à Phoenix et à San Francisco, où elle a rendu ses services accessibles à tous le mois dernier. Elle s’est récemment implantée à Austin et à Los Angeles, sans que la date d’ouverture au grand public ne soit connue.

**Le Secteur Confronté à Des Procédures Judiciaires En Cascade**

L’annonce de ce nouveau financement intervient à un moment clé dans le développement de technologies de conduite autonome, terni par de sévères doutes sur la sécurité routière. Cruise, filiale de General Motors, avait été affaiblie par des procédures judiciaires suite à des accidents impliquant ses véhicules. Elle avait alors mis un terme à toutes ses opérations. Début juin, elle avait annoncé remettre en service une petite flotte à Dallas, mais doit désormais reconquérir la confiance des Américains.

L’agence fédérale américaine chargée de la sécurité routière (NHTSA) a également ouvert une enquête sur les systèmes de conduite autonome de Waymo. La procédure, enclenchée suite à 22 incidents – dont 17 collisions, montrait que certains systèmes “semblaient désobéir aux dispositifs de contrôle” et présentaient un “comportement inattendu”. Ford se retrouve aussi dans le collimateur de la NHTSA, suite à deux accidents mortels survenus alors que son système de conduite autonome BlueCruise était activé.

**Doubler Ses Concurrents**

L’investissement de 5 milliards de dollars pourrait permettre à Waymo de doubler ses concurrents, qui enchaînent les retards dans la présentation de nouveaux projets. General Motors a annoncé hier l’abandon de la production du véhicule autonome de Cruise “Origin”, sans volant ni pédales, pour le remplacer par les dernières Chevrolet Bolt. Tesla, de son côté, va vraisemblablement reporter la présentation de son taxi robot “Cybercab” d’août à octobre, rapporte Bloomberg.

** GenAI : Citroën Embarque ChatGPT Dans Ses Véhicules Les Plus Récents **

** Mistral AI Pousse Ses Modèles D'IA Sur Google Cloud Afin D'attirer Plus De Clients **

** Danone Signe Un Contrat Pluriannuel Avec Microsoft Pour Intégrer L'IA Dans Ses Activités **

**SUR LE MÊME SUJET**

Waymo rend son service de taxis robots accessible à tous à San Francisco

Cruise teste à nouveau ses véhicules autonomes à Dallas et tente de remonter la pente

Suite à plusieurs collisions, les États-Unis enquêtent sur le système de conduite autonome de Waymo

Véhicules autonomes : Hyundai injecte près d'un milliard de dollars dans sa joint-venture Motional

**Sujets associés**

Véhicule autonome

Waymo

Investissement

Alphabet

Ford

GM

Etats-Unis

Nos journalistes sélectionnent pour vous les articles essentiels de votre secteur.

Groupe Moniteur Nanterre B 403 080 823, IPD Nanterre 490 727 633, Groupe Industrie Service Info (GISI) Nanterre 442 233 417.

Cette société ou toutes sociétés du Groupe Infopro Digital pourront l'utiliser afin de vous proposer pour leur compte ou celui de leurs

clients, des produits et/ou services utiles à vos activités professionnelles.

Pour exercer vos droits, vous y opposer ou pour en savoir plus : Charte des données personnelles.

LES ÉVÉNEMENTS USINE DIGITALE

Maîtriser les enjeux Privacy de l'IA : une nécessité incontournable

Décrypter les nouvelles réglementations européennes liées à l'IA et leurs impacts opérationnels

Cyber Sécurité

Cycle

Parcours métier Data Protection Officer (DPO)

23-27 Septembre 2024

Acquérir l'ensemble des compétences nécessaires aux DPO

Cyber Sécurité

S'approprier le rôle et les missions du Délégué à la Protection des Données (DPO)

Répondre à l'exigence 2.16 du référentiel de compétences des DPO

Cyber Sécurité

CHERCHE TALENTS NUMERIQUE

Assystem forme les demandeurs d’emploi à l’ingénierie nucléaire

Afin d’atteindre la neutralité carbone d’ici à 2050, le gouvernement français prévoit la construction d’au moins 6 EPR2 et la prolongation de la durée de vie des réacteurs nucléaires déjà en exploitation. Pour mener ce grand chantier indispensable à la transition énergétique, 100 000 personnes devraient être recrutées d'ici à 2030 dans la filière nucléaire française. Chez Assystem, le premier groupe d’ingénierie nucléaire en Europe, cette dynamique implique de doubler le nombre d'experts du nucléaire.

Emploi

Expert BI (H/F)

Free-work - 24 July 2024 - - Lyon, Auvergne-Rhône-Alpes

Développeur Mobile Android

Free-work - 24 July 2024 - - Lyon, Auvergne-Rhône-Alpes

Les formations USINE DIGITALE

élaborer une charte éthique d'utilisation des Intelligences Artificielles Génératives (IAG)

Supervision des IA dans son organisation

évaluez la conformité de vos usages IA

Comment implémenter l'IA Act dans vos process

Serious game : adopter la nouvelle directive NIS2

18/09 Matin

Maitrisez les évolutions de la cybersécurité

DPO : Supervisez les usages de l'IA de votre structure et des collaborateurs

Décrypter les nouvelles réglementations européennes liées à l'IA et leurs impacts opérationnels

Toutes les formations

**ARTICLES LES PLUS LUS**

CrowdStrike : pourquoi la mise à jour d'un antivirus a paralysé les entreprises du monde entier

Mistral AI et Nvidia dévoilent leur premier modèle commun, Mistral NeMo 12B

La pénurie de talents en cybersécurité, une épine dans le pied des entreprises françaises

Comment les ordinateurs quantiques nous aideront à résoudre des problèmes complexes

Alléger l'empreinte carbone de l'intelligence artificielle : rêve ou réalité ?

Atos échappe au naufrage, sauvé par un accord définitif avec banques et créanciers
25/07/2024 23:34 – Le projet futuriste de Facebook a perdu 41,5 milliards d’euros en 4 ans (le grand patron y croit encore) !
#MetaPlatforms #Technology #Yahoo! #Smartphone #Facebook #AppleInc #Snapchat

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
â© Presse-citron

ð•

La division Reality Labs de Meta, la maison mÃ¨re de Facebook, gÃ©nÃ¨re dÃ©jÃ des revenus grÃ¢ce aux ventes de ses casques de rÃ©alitÃ© mixte, comme le Quest 3. Mais ces revenus sont presque nÃ©gligeables par rapport aux sommes folles qui sont dÃ©pensÃ©es par Meta pour amÃ©liorer sa technologie. En effet, le but ultime du groupe est de dÃ©velopper des lunettes de rÃ©alitÃ© augmentÃ©e qui remplaceront, un jour, les smartphones, et quâ€™on pourra presque contrÃ´ler par la pensÃ©e.

En attendant que cette vision se concrÃ©tise, le projet continue de perdre plusieurs milliards de dollars, chaque annÃ©e. Et selon une rÃ©cente analyse de Yahoo Finance, au cours des 4 derniÃ¨res annÃ©es, les pertes cumulÃ©es par Reality Labs ont Ã©tÃ© de 45 milliards de dollars, soit lâ€™Ã©quivalent de 41,5 milliards dâ€™euros aujourdâ€™hui. Et cette analyse ne tient compte que des pertes de 2020, 2021, 2022 et 2023. Avant 2020, le groupe investissait dÃ©jÃ massivement dans la rÃ©alitÃ© mixte. Ã€ titre de comparaison, ces 45 milliards de dollars atteignent presque les capitalisations de Snapchat et de Pinterest combinÃ©es.

On pourrait dire quâ€™il est nÃ©cessaire dâ€™investir de pareilles sommes pour dÃ©velopper une technologie du futur. Cependant, lâ€™article de Yahoo Finance suggÃ¨re que ces pertes sont aussi le symptÃ´me de problÃ¨mes organisationnels au sein de Meta.

**Le Premier Prototype Arrive**

En tout cas, la division Reality Labs va continuer Ã perdre de lâ€™argent. Mais, la bonne nouvelle, pour le groupe, câ€™est que ces investissements sont moins critiquÃ©s quâ€™il y a un ou deux ans. Aujourdâ€™hui, les apps et services de Meta sont en pleine forme. Et le groupe peut aussi se vanter de ne pas Ãªtre en retard dans la course Ã lâ€™intelligence artificielle. Sinon, le fait quâ€™Apple soit aussi prÃ©sent sur la rÃ©alitÃ© mixte peut Ãªtre perÃ§u comme une validation de ces investissements. Au cours des 12 derniers mois, la valeur de lâ€™action Meta a grimpÃ© de 67,58 %.

Meta devrait prochainement avoir quelque chose de concret Ã prÃ©senter. En juin, Mark Zuckerberg a indiquÃ©, lors dâ€™une interview avec une chaÃ®ne YouTube, que le prototype de lunettes â€œholographiquesâ€ de Meta. Cependant, le groupe nâ€™est encore prÃªt quâ€™Ã prÃ©senter un prototype, et prÃ©fÃ¨re prendre son temps pour la commercialisation, afin de ne pas vendre un produit imparfait aux consommateurs.

La division Reality Labs de Meta, qui dÃ©veloppe ses produits de rÃ©alitÃ© mixte, gÃ©nÃ¨re dÃ©jÃ des revenus grÃ¢ce aux ventes des produits comme le casque Quest 3 Cependant, ces revenus sont trÃ¨s faibles par rapport aux pertes enregistrÃ©es par cette division, dont lâ€™objectif est de proposer des lunettes de rÃ©alitÃ© mixte qui seront capables de remplacer les smartphones Dâ€™aprÃ¨s Yahoo Finance, en 2020, 2021, 2022 et 2023, ces pertes Ã©taient de 45 milliards de dollars Meta sâ€™apprÃªte Ã prÃ©senter le premier prototype de ses lunettes de rÃ©alitÃ© mixte (Ã diffÃ©rencier des casques qui ne sont pas compacts)

ðŸ“ Pour ne manquer aucune actualitÃ© de Presse-citron, suivez-nous sur Google ActualitÃ©s et WhatsApp.

[ SOURCE ]

MetaRÃ©alitÃ© Mixte

L'argent n'est pas perdu, il a fait vivre pas mal de monde, bravo Ã FB de faire Å“uvre sociale

Encore un article non relu â€¦ avant publication. Hallucinant.

Les derniÃ¨res actualitÃ©s

Google nous a tous fait travailler 816 millions d'heures Ã notre insu (le piÃ¨ge est trÃ¨s malin)

AprÃ¨s lâ€™Ã©chec de Wiz, Google met le paquet sur ses voitures autonomes Waymo

Qu'est-ce que The Operator, ce jeu vidÃ©o franÃ§ais qui bat des records de ventes

Avec l'obtention de cette nouvelle licence, Revolut semble inarrÃªtable

Ray-Ban est-il devenu un faiseur de rois dans la tech ?

Ce constructeur automobile intÃ¨gre maintenant ChatGPT Ã bord, mais pourquoi faire ?

Interdire les rÃ©seaux sociaux aux moins de 15 ans ? 3 FranÃ§ais sur 4 sont pour !

Jeunes parents ? L'appli MoonSee promet de rÃ©volutionner le baby-sitting
25/07/2024 23:33 – Qui est Cohere, la discrète étoile montante de l’IA, qui vient de lever 500 millions de dollars ?
#ArtificialIntelligence #OpenAI #StartupCompany #CloudComputing #GenerativeArtificialIntelligence #Technology #Computing #Intelligence #ArtificialGeneralIntelligence #AmazonCompany #CognitiveScience #Cyberspace #InformationTechnology #ComputerScience #Cognition #AmazonWebServices #BranchesOfScience #HumanIntelligence #ComputationalNeuroscience #Learning #Spotify

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
â© Image gÃ©nÃ©rÃ©e par IA via Copilot avec DALL-E

ð•

La start-up Cohere, spÃ©cialisÃ©e dans lâ€™intelligence artificielle (IA) gÃ©nÃ©rative, vient de lever 500 millions de dollars. Une somme considÃ©rable qui va lui permettre dâ€™accÃ©lÃ©rer le dÃ©veloppement de ses modÃ¨les pour les entreprises.

**Des ModÃ¨les RÃ©pondant Aux Besoins SpÃ©cifiques Des Entreprises**

Car contrairement aux fleurons du secteur comme OpenAI ou Anthropic, Cohere ne cible pas le grand public mais plutÃ´t les utilisateurs professionnels. FondÃ©e en 2019 par trois anciens chercheurs chez Google, Aidan Gomez, Nick Frosst et Ivan Zhang, la start-up est basÃ©e Ã Toronto au Canada.

Gomez, PDG de Cohere, est une personnalitÃ© Ã©minente dans le monde de lâ€™IA. En 2017, il a cosignÃ© un article technique jetant les bases de nombreux modÃ¨les d'IA gÃ©nÃ©rative parmi les plus performants Ã l'heure actuelle comme GPT-4o. Pour sa part, Cohere collabore avec ses clients pour crÃ©er des modÃ¨les sur mesure basÃ©s sur leurs donnÃ©es propriÃ©taires. Objectif : leur permettre de fonctionner plus efficacement avec une IA adaptÃ©e Ã leurs besoins propres.

â« Nous ne sommes pas Ã la recherche de l'intelligence artificielle gÃ©nÃ©rale (Ndlr : une IA capable de rivaliser avec l'intelligence humaine dans une variÃ©tÃ© de tÃ¢ches complexes). Nous essayons de crÃ©er des modÃ¨les qui peuvent Ãªtre utilisÃ©s efficacement dans une entreprise pour rÃ©soudre des problÃ¨mes rÃ©els â», explique Nick Frosst Ã Bloomberg. Son catalogue de clients compte dÃ©jÃ du beau monde, Ã lâ€™instar dâ€™Oracle, de Notion ou de Spotify.

**Cohere Est ValorisÃ©e Ã 5 Milliards**

Lâ€™un des points forts de Cohere rÃ©side dans la flexibilitÃ© de sa plateforme. Elle n'est pas liÃ©e Ã un seul fournisseur cloud ou Ã une infrastructure spÃ©cifique. ConcrÃ¨tement, elle peut aussi bien Ãªtre dÃ©ployÃ©e Ã travers des fournisseurs comme Google Cloud ou Amazon Web Services (AWS) que dans lâ€™infrastructure cloud existante du client.

La start-up gÃ¨re Ã©galement un laboratoire de recherche Ã but non lucratif. Celui-ci publie des modÃ¨les ouverts pour la comprÃ©hension et l'analyse de textes. GrÃ¢ce Ã cette nouvelle levÃ©e de fonds, Cohere voit sa valorisation doubler par rapport Ã juin 2023 pour avoisiner les 5 milliards de dollars. Une preuve de plus de lâ€™extrÃªme engouement de lâ€™industrie pour lâ€™IA gÃ©nÃ©rative, avec des sommes parfois colossales injectÃ©es dans son dÃ©veloppement. Les financements dâ€™OpenAI ou dâ€™Anthropic en tÃ©moignent.

Cohere vise dÃ©sormais Ã concevoir des modÃ¨les encore plus sophistiquÃ©s. Ces fonds sâ€™avÃ©reront nÃ©cessaires pour y arriver, le coÃ»t du dÃ©veloppement de lâ€™IA Ã©tant incroyablement Ã©levÃ©. â« Ã€ l'avenir, tout en poursuivant notre activitÃ© de modÃ¨le en tant que service, nous commencerons Ã nous concentrer de plus en plus sur l'augmentation de la main-d'Å“uvre grÃ¢ce Ã l'IA et Ã l'automatisation sophistiquÃ©e des tÃ¢ches â», indique Aidan Gomez dans une note partagÃ©e Ã ses employÃ©s.

ðŸ“ Pour ne manquer aucune actualitÃ© de Presse-citron, suivez-nous sur Google ActualitÃ©s et WhatsApp.

[ SOURCE ]

intelligence artificiellelevÃ©e de fonds

Les derniÃ¨res actualitÃ©s

Google nous a tous fait travailler 816 millions d'heures Ã notre insu (le piÃ¨ge est trÃ¨s malin)

AprÃ¨s lâ€™Ã©chec de Wiz, Google met le paquet sur ses voitures autonomes Waymo

Qu'est-ce que The Operator, ce jeu vidÃ©o franÃ§ais qui bat des records de ventes

Avec l'obtention de cette nouvelle licence, Revolut semble inarrÃªtable

Ray-Ban est-il devenu un faiseur de rois dans la tech ?

Ce constructeur automobile intÃ¨gre maintenant ChatGPT Ã bord, mais pourquoi faire ?

Interdire les rÃ©seaux sociaux aux moins de 15 ans ? 3 FranÃ§ais sur 4 sont pour !

Jeunes parents ? L'appli MoonSee promet de rÃ©volutionner le baby-sitting
25/07/2024 23:29 – Panne mondiale Microsoft/Crowdstrike : on connaît le coût de la panne, et il est monstrueux !
#Microsoft #ComputerScience #CloudComputing #Economy #CrowdStrike #DigitalData #Transport #Airport #UnitedStatesDollar #Technology #RiskManagement #Risk #Bank #Insurance #MicrosoftWindows #Investor #VulnerabilityComputing #Health #Manufacturing #AirTrafficControl

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
â© Unsplash / Behnam Norouzi

ð•

Le 19 juillet dernier, le monde des affaires a Ã©tÃ© secouÃ© par une panne informatique d'une ampleur sans prÃ©cÃ©dent. Une mise Ã jour dÃ©ployÃ©e par l'entreprise de cybersÃ©curitÃ© Crowdstrike a provoquÃ© l'arrÃªt simultanÃ© de plus de 8 millions d'ordinateurs sous licence Windows Ã travers le globe. Aujourd'hui, alors que la poussiÃ¨re retombe, les premiÃ¨res estimations du coÃ»t Ã©conomique de cette catastrophe numÃ©rique commencent Ã Ã©merger, et les chiffres sont vertigineux.

Selon une Ã©tude menÃ©e par l'assureur Parametrix, le prÃ©judice financier pour les seules entreprises du Fortune 500 (hors Microsoft) s'Ã©lÃ¨verait Ã au moins 5,4 milliards de dollars. Ce montant colossal ne prend en compte que les 500 plus grandes entreprises mondiales L'impact Ã©conomique total est donc bien plus important si l'on inclut les millions d'autres sociÃ©tÃ©s et organisations touchÃ©es par cette panne.

La paralysie numÃ©rique a frappÃ© de maniÃ¨re inÃ©gale les diffÃ©rents secteurs de l'Ã©conomie. Celui de la santÃ© arrive en tÃªte des pertes, avec prÃ¨s de deux milliards de dollars de dommages estimÃ©s. Cette situation est particuliÃ¨rement prÃ©occupante, Ã©tant donnÃ© l'importance des systÃ¨mes informatiques dans la gestion des soins de santÃ© modernes. Les retards dans les traitements, les diagnostics et la gestion administrative pourraient avoir des rÃ©percussions Ã long terme sur la qualitÃ© des soins et l'efficacitÃ© des systÃ¨mes de santÃ©.

Le secteur bancaire n'est pas en reste, avec plus d'un milliard de dollars de pertes. Dans un monde oÃ¹ les transactions financiÃ¨res sont de plus en plus dÃ©matÃ©rialisÃ©es, une telle panne a pu causer des perturbations majeures dans les flux financiers mondiaux, affectant potentiellement la stabilitÃ© des marchÃ©s et la confiance des investisseurs. La Bourse de Londres, par exemple, avait ouvert en retard.

â© Parametrix**Le Transport AÃ©rien, Grand Perdant De Cette Crise**

Le secteur du transport, et plus particuliÃ¨rement l'industrie aÃ©rienne, a Ã©tÃ© frappÃ© de plein fouet par cette panne. Avec des pertes estimÃ©es Ã 860 millions de dollars, c'est le troisiÃ¨me secteur le plus touchÃ© en valeur absolue. Mais c'est surtout celui qui affiche la perte moyenne par entreprise la plus Ã©levÃ©e, atteignant 143 millions de dollars.

Ce chiffre s'explique par la dÃ©pendance extrÃªme du transport aÃ©rien aux systÃ¨mes informatiques, que ce soit pour la gestion des rÃ©servations, l'enregistrement des passagers, ou le contrÃ´le du trafic aÃ©rien. La panne a provoquÃ© un vÃ©ritable chaos dans de nombreux aÃ©roports, avec des retards en cascade, des annulations de vols et des passagers bloquÃ©s. L'Ã©tude de Parametrix rÃ©vÃ¨le d'ailleurs que 100% du secteur aÃ©rien a Ã©tÃ© affectÃ© par la panne, soulignant la vulnÃ©rabilitÃ© particuliÃ¨re de cette industrie face aux dÃ©faillances informatiques.

Le secteur informatique lui-mÃªme n'a pas Ã©tÃ© Ã©pargnÃ©, avec 560 millions de dollars de pertes estimÃ©es. Cette situation paradoxale met en lumiÃ¨re la fragilitÃ© de notre Ã©cosystÃ¨me numÃ©rique, oÃ¹ mÃªme les entreprises spÃ©cialisÃ©es dans la technologie peuvent se retrouver paralysÃ©es par une simple mise Ã jour dÃ©fectueuse.

**Remise En Question**

L'ampleur de cette panne soulÃ¨ve des questions sur le fonctionnement de nos infrastructures numÃ©riques. Environ 25% des entreprises du Fortune 500 ont Ã©tÃ© touchÃ©es, ce qui reprÃ©sente une proportion alarmante des plus grandes entreprises mondiales. Cette vulnÃ©rabilitÃ© gÃ©nÃ©ralisÃ©e face Ã un incident technique d'apparence anodine devrait inciter les dirigeants d'entreprises et les dÃ©cideurs politiques Ã repenser leurs stratÃ©gies de gestion des risques informatiques.

L'Ã©tude de Parametrix souligne Ã©galement que les entreprises disposant d'infrastructures basÃ©es sur le cloud ont gÃ©nÃ©ralement pu reprendre leurs activitÃ©s plus rapidement que celles dÃ©pendant d'ordinateurs physiques traditionnels. Cette observation pourrait accÃ©lÃ©rer la transition vers des solutions cloud, perÃ§ues comme plus rÃ©silientes face Ã ce type de pannes gÃ©nÃ©ralisÃ©es.

Enfin,â quid de la responsabilitÃ© juridique et financiÃ¨re en cas de panne informatique majeure ? Crowdstrike, Ã l'origine de la mise Ã jour dÃ©fectueuse, pourrait faire face Ã des poursuites judiciaires de la part des entreprises les plus touchÃ©es. Cependant, la complexitÃ© des chaÃ®nes de responsabilitÃ© dans l'Ã©cosystÃ¨me numÃ©rique rend souvent difficile l'attribution claire des fautes et des dÃ©dommagements. L'entreprise a clairement assumÃ© ses responsabilitÃ©s mais l'audition du patron de l'entreprise devant le CongrÃ¨s amÃ©ricain devrait apporter plus de prÃ©cisions.

La panne mondiale causÃ©e par une mise Ã jour de Crowdstrike a coÃ»tÃ© au moins 5,4 milliards de dollars aux entreprises du Fortune 500. Le secteur de la santÃ© a Ã©tÃ© le plus touchÃ© en valeur absolue, tandis que le secteur aÃ©rien a subi les pertes moyennes par entreprise les plus Ã©levÃ©es. Ce coÃ»t ne prend pas en compte les millions d'autres entreprises touchÃ©es par cette panne.

ðŸ“ Pour ne manquer aucune actualitÃ© de Presse-citron, suivez-nous sur Google ActualitÃ©s et WhatsApp.

Microsoftpanne

Les derniÃ¨res actualitÃ©s

Google nous a tous fait travailler 816 millions d'heures Ã notre insu (le piÃ¨ge est trÃ¨s malin)

AprÃ¨s lâ€™Ã©chec de Wiz, Google met le paquet sur ses voitures autonomes Waymo

Qu'est-ce que The Operator, ce jeu vidÃ©o franÃ§ais qui bat des records de ventes

Avec l'obtention de cette nouvelle licence, Revolut semble inarrÃªtable

Ray-Ban est-il devenu un faiseur de rois dans la tech ?

Ce constructeur automobile intÃ¨gre maintenant ChatGPT Ã bord, mais pourquoi faire ?

Interdire les rÃ©seaux sociaux aux moins de 15 ans ? 3 FranÃ§ais sur 4 sont pour !

Jeunes parents ? L'appli MoonSee promet de rÃ©volutionner le baby-sitting
25/07/2024 19:12 – 3 Major Bombshells Just Dropped That Everybody Should Be Talking About Right Now :16
#John3:16 #JaimeMaussan #EternalLifeChristianity

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
Home Economy

3 Major Bombshells Just Dropped That Everybody Should Be Talking About Right Now TOPICS:ConspiracyJoe BidenMichael SnyderTrump

July 25, 2024

This article was originally published by Michael Snyder at The End of the American Dream.

The past couple of weeks have truly been historic. Take a moment and just consider some of the major events that we have witnessed. Donald Trump got shot, Joe Biden dropped out of the race for president, and Kamala Harris quickly became the new presumptive nominee. It is hard to remember a time when so much has happened in such a short period of time. Now more bombshells have dropped, but because so much else is going on they are not getting the attention that they deserve. The following are 3 major bombshells that everybody should be talking about right now…

#1 It appears that Joe Biden had some sort of a medical emergency while he was in Las Vegas last week…

A Las Vegas hospital was on standby after being alerted about a possible medical issue with President Joe Biden Wednesday afternoon while he was visiting Southern Nevada.

Law enforcement and medical teams gathered outside the University Medical Center trauma room entrance on Wellness Way in case the President of the United States arrived after he canceled an appearance at the Unidos US Convention in Las Vegas due to testing positive for COVID-19. Unidos US President and CEO Janet Murguía made the announcement to attendees who were waiting to hear the president speak.

You don’t rush someone to the hospital because of COVID.

(function(w,d,s,i){w.ldAdInit=w.ldAdInit||[];w.ldAdInit.push({slot:10841224714367334,size:[0, 0],id:"ld-6931-3360"});if(!d.getElementById(i)){var j=d.createElement(s),p=d.getElementsByTagName(s)[0];j.async=true;j.src="//cdn2.lockerdomecdn.com/_js/ajs.js";j.id=i;p.parentNode.insertBefore(j,p);}})(window,document,"script","ld-ajs");

Something else was clearly going on.

The hospital that was put on alert is “the only Level 1 Trauma Center in Nevada”…

Due to the hospital being the only Level 1 Trauma Center in Nevada, that’s where the president will go during a medical emergency.

Van Houweling said the hospital was alerted around 2:15 p.m. about a possible medical issue with the president.

Why didn’t the big news networks make a big deal about this?

Apparently “hundreds of Las Vegas Metropolitan officers” heard the emergency orders that were issued…

Several law enforcement officers on duty that day were informed over the radio that the president was dealing with an unspecified medical emergency. Far from a case of the sniffles, this was sent out on encrypted police airwaves as if something akin to a five-alarm fire had broken out.

Our sources estimate that it was “easily” hundreds of Las Vegas Metropolitan officers and employees who heard the broadcasts live, so a curious media shouldn’t have a problem reporting any follow-ups to this story. The dispatches made clear that this was much more than a mere change of plans, because it set into motion so much of their on-duty emergency response apparatus.

Radio dispatchers responsible for coordinating with the president’s detail requested an immediate “surge” of law enforcement resources into the area, both to secure the vicinity and to expedite the president’s movement.

Ultimately, Biden must have improved significantly, because he did not end up going to the hospital.

Instead, he was taken to the airport where he departed on Air Force One.

According to a reporter who was on the plane, Air Force One “flew so fast the plane shook”…

The Daily Mail’s Emily Goodin posted last Wednesday night after Air Force One landed in Delaware, “POTUS arrives in Dover. “I’m doing well,” he said. Flight time from Vegas was 3 hour 48 minutes. AF1 flew so fast the plane shook” (Travel Math has normal Vegas to Dover timed at 4 hours and 50 minutes.)

It is obvious that something happened in Las Vegas and it is being hushed up.

Could this medical emergency have played a role in Biden’s decision to suddenly exit the race?

#2 The Oversight Project, a watchdog group that is part of the Heritage Foundation, has uncovered some absolutely stunning information.

Their analysis of phone data has revealed that someone who was regularly at the home where Thomas Matthew Crooks lived also visited a building in Washington D.C. that is located very close to an FBI office…

Who visited Thomas Crooks’ home before he attempted to assassinate President Trump?

Mobile ad data analysis reveals someone who regularly visited Crooks’ Pennsylvania home also visited a building in DC near an FBI office.

Many people do not realize that our phones are constantly gathering data on all of us.

In this case, phone data could be the key in determining why Crooks did what he did.

According to the Oversight Project, there were nine phones that “were located at Crooks’ home and work within the last year”…

The Oversight Project identified nine devices linked to AD-IDs that were located at Crooks’ home and work within the last year.

Per the Heritage Foundation’s Oversight Project: “We found the assassin’s connections through our in-depth analysis of mobile ad data to track movements of Crooks and his associates”

“To do this, we tracked devices that regularly visited both Crooks’s home and place of work and followed them,” the Oversight Project said.

One of those phones made a trip to Washington.

Who owned that phone?

And who did that person meet with in Washington?

We definitely need answers to those questions.

#3 It has been determined that one of the “alien mummies” that has been discovered in Peru has fingerprints that are not human…

Bizarre diagonal grooves along the fingertips and toes of one of Peru’s eerie and hotly debated ‘alien mummies’ points to a nonhuman origin for these odd remains.

Worldwide controversy has followed the small peculiar specimens since September, when veteran broadcast journalist and prolific UFO researcher Jaime Maussan first presented two alleged ‘alien’ corpses to Mexico’s congress last year.

The man that examined the fingers of this “alien mummy” is actually a defense attorney who once worked as a prosecutor in Colorado…

Now DailyMail.com has spoken to a former Colorado prosecutor and current defense attorney, who has examined one of the over half-a-dozen ‘alien’ specimens with the help of three independent forensic medical examiners from the United States.

‘These were not traditional human fingerprint patterns,’ that attorney, Joshua McDowell, told DailyMail.com.

This guy sounds quite credible, and the Daily Mail has posted a photo of the bizarre fingerprints on this “alien mummy”.

If this creature and the other “alien mummies” that have been discovered are not human, what are they?

As I detailed in my latest book, I believe that the stage is being set for a great deception that will shake the entire world.

Of course the truth is that deception is already everywhere around us.

We aren’t being told the truth about Biden.

We aren’t being told the truth about Trump’s shooter.

And there are hundreds of other things that we aren’t being told the truth about.

George Orwell once warned us that “in a time of deceit telling the truth is a revolutionary act”.

Sadly, Orwell’s statement is more relevant today than ever, and a lot more deception is on the way.

Michael’s new book entitled “Chaos” is available in paperback and for the Kindle on Amazon.com, and you can subscribe to his Substack newsletter at michaeltsnyder.substack.com.

Sourced from SHTFplan.com

Subscribe to Activist Post for truth, peace, and freedom news. Follow us on Telegram, HIVE, Minds, MeWe, Twitter – X and Gab.

Provide, Protect and Profit from what’s coming! Get a free issue of Counter Markets today.

(function(w,d,s,i){w.ldAdInit=w.ldAdInit||[];w.ldAdInit.push({slot:8023242615045735,size:[0, 0],id:"ld-6215-6164"});if(!d.getElementById(i)){var j=d.createElement(s),p=d.getElementsByTagName(s)[0];j.async=true;j.src="//cdn2.lockerdome.com/_js/ajs.js";j.id=i;p.parentNode.insertBefore(j,p);}})(window,document,"script","ld-ajs");Free ebook How To Survive the Job Automation Apocalypse

Free ebook How To Get Started with Bitcoin: Quick and Easy Beginner’s Guide

Activist Post Daily Newsletter (function() {

window.mc4wp = window.mc4wp || {

listeners: [],

forms: {

on: function(evt, cb) {

window.mc4wp.listeners.push(

{

event : evt,

callback: cb

}

);

}

}

}

})();

Email address:

Yes - I consent to receive emails

Leave this field empty if you're human:

Subscription is FREE and CONFIDENTIAL

Free Report: How To Survive The Job Automation Apocalypse with subscription

ShareTweetPin

post
25/07/2024 06:52 – Microsoft Taps Lumen to Meet AI Data Center Demands
#ArtificialIntelligence #Microsoft #MicrosoftAzure #CloudComputing #DigitalTransformation #LumenTechnologies #Spreadsheet #Technology #Computing #InformationTechnology #ComputerScience #InformationTechnologyManagement #Cyberspace #ServiceIndustries #InformationAndCommunicationsTechnology #Telecommunications #LargeLanguageModel #MicrosoftEngineeringGroups #MicrosoftExcel #InformationScience #ComputerNetwork #News #DigitalMedia #GenerativeArtificialIntelligence #HumanCommunication #DisruptiveInnovation #Infrastructure #DataCenter #Business #BranchesOfScience #TokenizationDataSecurity #Scalability #Innovation #LegacySystem #Engineering #Computers #ComputerEngineering #Management #InformationAge #ComputerNetworking

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
***MICROSOFT TAPS LUMEN TO MEET AI DATA CENTER DEMANDS***

By PYMNTS

July 24, 2024

Microsoft has launched a partnership with telecommunications company Lumen Technologies.

The collaboration, announced Wednesday (July 24), will see Lumen use Microsoft’s cloud to further its “digital transformation.”

Microsoft, meanwhile, will use Lumen to expand its network capacity and capability amid growing demand on its data centers due to artificial intelligence (AI).

“AI is reshaping our daily lives and fundamentally changing how businesses operate,” Erin Chapple, corporate vice president of Azure Core Product and Design at Microsoft, said in a news release.

“We are focused both on the impact and opportunity for customers relative to AI today, and a generation ahead when it comes to our network infrastructure.”

Lumen, Chapple added, has the network infrastructure and the digital capabilities required to help Azure create a reliable, scalable platform.

For its part, Lumen said Microsoft’s cloud and AI tech can help it cut technology costs, “remove legacy systems and silos,” improve its offerings, and create new solutions for customers.

Lumen said it will move its workloads to Microsoft Azure, use Microsoft Entra solutions to protect access and prevent identity attacks and work with Microsoft on telecom industry-specific solutions.

“This element alone is expected to improve Lumen’s cash flow by more than $20 million over the next 12 months while also improving the company’s customer experience,” the release said.

In other Microsoft news, PYMNTS wrote last week that the company — as it continues its work to add AI to every part of the digital world — “has set its sights on conquering the final frontier of office productivity: the humble spreadsheet.”

The tech giant’s latest creation, SpreadsheetLLM, is designed to alter how businesses crunch numbers and make decisions. By tapping into the power of large language models (LLMs), this AI tool, which is still in the testing stages, could turn Excel from a static grid into a dynamic, question-answering powerhouse — potentially reshaping workflows for millions of users.

“The infinite cell-like nature and references to cells in spreadsheets make it challenging for LLMs, which have been trained using standard linear tokenization techniques, to understand the spreadsheet data model,” Rogers Jeffrey Leo John, co-founder and CTO of DataChat, a no-code, generative AI platform, said in an interview with PYMNTS.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.

Recommended

Microsoft Taps Lumen to Meet AI Data Center Demands

Telecom Subsidy Ruled Unconstitutional, May Go to Supreme Court

CrowdStrike Sends $10 Gift Cards to Teammates, Partners After Outage

From AI to Esports, 2024 Paris Olympics Sees Digital Evolution

See More In: AI, artificial intelligence, Azure Core Product and Design, cloud computing, data centers, Erin Chapple, Lumen, Lumen Technologies, Microsoft, Microsoft Entra, News, PYMNTS News, What's Hot

AT&T: Expanding Fiber Network Helps Gain, Retain Mobile Phone Subscribers

Report: Coinbase Asset Management Creating Tokenized Money Market Fund

FDIC Vice Chairman Travis Hill: Basel III Endgame Needs Re-Proposal

The Big Story

Study Finds Treasurer Role Misunderstood by 53% of C-Suite Peers

Featured News

From AI to Esports: Digital Evolution Goes for Gold at Paris Olympics

Security Summit Alerts Tax Professionals to Growing Threat of Identity Theft

Shop Talk: High-End Consumers Swap Fancy Threads for Fancy Sweets

Mastercard Doubles Fraud Detection Rate With Generative AI

New Data: Gen Z Leads the Way as Digital Wallets Surge in the UK

FedNow’s First Year: Path to Critical Mass Requires Time and Education

Trending: AI Gets Smarter on Fraud and Financial Crime as New Models Boost Security

Partner with PYMNTS

We’re always on the lookout for opportunities to partner with innovators and disruptors.
24/07/2024 22:45 – Gaming Velocity
#ScrumSoftwareDevelopment #AgileSoftwareDevelopment #HTTPCookie #Business #PerformanceIndicator #Management #CounterproductiveWorkBehavior #Productivity #Communication #Goal

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
**TL; DR: Gaming Velocity**

Imagine your team’s line manager insists that a successful team improves velocity regularly. How could you, as a team, satisfy this strange, unsuitable demand without working more? How can you make gaming velocity a reality?

I run this exercise with my students of entry-level Scrum Master and Product Owner classes to help them reflect on the tricky nature of measuring success, metrics, and, of course, Goodhart’s Law: “When a measure becomes a target, it ceases to be a good measure.”

For the following article, I aggregated suggestions from more than 50 classes on how to “best” game velocity.

🎓 September 1, 2024: The Product Backlog Management Master Class Course for Just $99.

👉 Please note:

The course includes membership in a new community of agile professionals.

The course will only be available for sign-up until September 8, 2024!

🇩🇪 Zur deutschsprachigen Version des Artikels: Velocity und Goodhart’s Law.

🗞 Shall I notify you about articles like this one? Awesome! You can sign up here for the ‘Food for Agile Thought’ newsletter and join 42,000-plus subscribers.

🎓 Join Stefan in one of his upcoming Professional Scrum training classes!

**Why Velocity Is A Problematic Metric**

Before we jump to the suggestions, let us analyze why velocity is a tricky metric:

Team-Specific Nature: Velocity is inherently specific to each team, influenced by their unique estimation practices, work processes, and individual skill levels.

Misleading Comparisons: Using velocity to compare different teams can be misleading, as higher velocity does not necessarily indicate higher productivity or efficiency.

Different Definitions: Each team may define story points differently, invalidating cross-team comparisons.

Counterproductive Behaviors: An overemphasis on velocity can lead to inflating estimates to boost numbers or focusing on easy tasks rather than delivering real value.

Lastly, relying on velocity for broader comparisons or performance evaluations undermines Scrum’s core principles and can harm team morale and effectiveness.

Consequently, a team should use velocity — if at all — internally to help plan and track progress.

**The Gaming Velocity Exercise**

The scenario is as follows: Your team’s line manager is convinced that a successful Scrum Team in software development steadily increases its velocity Sprint after Sprint. Your task as a team is simple: Identify ways that allow a Scrum team to report a regular increase in velocity without working more at the team member level.

**Gaming Velocity For A Regular Increase Without Increasing Your Workload**

Now let us have a look at 13 ways of gaming velocity:

Estimate and Re-Estimate Story Points

Systematically increase the estimation of items from Sprint to Sprint.

Re-evaluate and increase the story points of past completed items.

Split User Stories

Break down larger items into smaller ones and assign higher story points to each smaller part.

Divide existing work items into even smaller parts, creating more granular tasks.

Include Non-Development Work

Count bugs and fixing technical debt with story points.

Factor in the complexity of completed research by assigning story points to spikes or research tasks.

Include routine maintenance or administrative tasks and estimate them.

Gaming Velocity by Creating Dummy or Fake Tasks

Add dummy tasks to the Sprint that do not require actual work.

Invent tasks that are not essential to the product but can be completed quickly.

Automate Regular Tasks but Report Them as Manual

Automate repetitive tasks but report them as if they were manually completed.

Adjust the Definition of Done (DoD)

Lower quality standards or relax the DoD to complete items faster.

Adjust the DoD to include more tasks as done, even if they are not fully completed.

Simulate technical debt when completing tasks.

Extend Sprint Duration

Quietly increase the duration of Sprints, resulting in higher reported velocity.

Overlapping Sprints

Work on tasks that span multiple Sprints and count them towards all Sprints.

Adjust Velocity Calculation

Change the method of calculating velocity, such as using a rolling average over fewer Sprints.

Include buffers or contingency points in planning that appear as additional velocity when not used.

Focus on Easy Wins for Gaming Velocity

Prioritize simpler tasks that can be completed quickly and easily.

Choose only easy work items to ensure quick completion.

Report Unfinished Work as Complete

Mark unfinished tasks as complete to boost the reported velocity.

Create Stories from Minor Tasks

Build work items out of insignificant tasks to increase the number of completed items.

Turn trivial tasks into stories and estimate them.

Task Recycling

Reintroduce tasks that were already present but unestimated in previous Sprints.

Cannot see the form? Please click here.

A Note of Caution: While these methods can technically inflate reported velocity, they may compromise the integrity and transparency of the Scrum process. Considering the long-term impact on team morale, product quality, and stakeholder trust is crucial. Ideally, it would be more beneficial to discuss and align expectations with stakeholders about realistic measures of Scrum success—such as delivering valuable Increments and maintaining a sustainable work pace.

👉 Don’t undermine Scrum’s core principles instead!

**Additional Considerations Regarding Gaming Velocity**

While velocity can be a valuable metric within the context of a single team, it’s important to understand its limitations and use it wisely. Here are some additional considerations:

Context Matters: Velocity should always be interpreted in the context of the team’s specific circumstances, including team size, experience, and the nature of the work being done.

Complementary Metrics: Relying solely on velocity can be misleading. Complement it with other metrics like customer satisfaction, quality of deliverables, and team morale to get a holistic view of team performance.

Focus on Outcomes: The ultimate goal of any Scrum team is to deliver value. Focus on outcomes and impacts rather than just output. High velocity without valuable deliverables is meaningless.

Continuous Improvement: Use velocity as a continuous improvement tool rather than a strict target. Reflect on fluctuations and understand the underlying causes to improve processes and efficiency.

Communication and Transparency: Ensure that all stakeholders understand the nuances of velocity. Clear communication can prevent misunderstandings and unrealistic expectations.

Avoid Pressure: High pressure to increase velocity can lead to burnout and decreased quality. Maintain a sustainable pace and prioritize the well-being of team members.

By considering these considerations, teams and organizations can use velocity effectively while avoiding the pitfalls of “gaming” the metric.

**Conclusion**

While velocity is helpful for internal team planning and progress tracking, it becomes problematic when used for serious applications such as comparing different teams. The exercise I conduct with my students demonstrates how easily velocity can be manipulated, revealing the pitfalls of over-reliance on this metric. Goodhart’s Law states, “When a measure becomes a target, it ceases to be a good measure.” (See above.) This is particularly true for velocity, which can drive teams to engage in counterproductive behaviors, such as inflating estimates and focusing on low-effort tasks.

Ultimately, velocity should remain a tool for internal use, guiding a team’s progress without becoming a benchmark for performance comparisons. By understanding its limitations and potential for misuse, organizations can avoid the trap of “gaming velocity.” Instead, focus on delivering actual value.

Nevertheless, the exercise is a good thought experiment that discusses the nature of measuring success. By the way, when I play the game with managers, they know how gaming velocity might work.

How are you measuring your team’s success? Please share with us in the comments.

**Gaming Velocity — Recommended Reading**

Agile Metrics: The Good, the Bad, and the Ugly

Useless Agile Metrics

Agile Metrics Survey 2021

Download the Agile Metrics Survey 2021

👆 Stefan Wolpers: The Scrum Anti-Patterns Guide (Amazon advertisement.)

**📅 Scrum Training Classes, Workshops, And Events**

You can secure your seat for Scrum training classes, workshops, and meetups directly by following the corresponding link in the table below:

Date

Class and Language

City

Price

🖥 💯 🇬🇧 August 20-21, 2024

GUARANTEED: Professional Scrum Master Advanced Training (PSM II; English; Live Virtual Class)

Live Virtual Class

€1.299 incl. 19% VAT

🖥 🇩🇪 August 27-28, 2024

Professional Scrum Master Training (PSM I; German; Live Virtual Class)

Live Virtual Class

€1.189 incl. 19% VAT

🖥 💯 🇬🇧 August 29-September 26, 2024

GUARANTEED: Advanced Product Backlog Management Cohort (English; Live Virtual Cohort)

Live Virtual Cohort

€399 incl. 19% VAT

🖥 💯 🇩🇪 September 3-6, 2024

GUARANTEED: Professional Scrum Product Owner Training (PSPO I; German; Live Virtual Class)

Live Virtual Class

€1.299 incl. 19% VAT

🖥 💯 🇬🇧 September 4, 2024

GUARANTEED: Hands-on Agile #64: Mastering Work Intake Jeremy Willets and Tom Cagley (English)

Live Virtual Meetup

FREE

🖥 💯 🇬🇧 October 9, 2024

GUARANTEED: Hands-on Agile #65: The Lean Tech Manifesto with Fabrice Bernhard (English)

Live Virtual Meetup

FREE

See all upcoming classes here.

You can book your seat for the training directly by following the corresponding links to the ticket shop. If the procurement process of your organization requires a different purchasing process, please contact Berlin Product People GmbH directly.

**✋ Do Not Miss Out And Learn More About Avoiding Gaming Velocity — Join The 20,000-plus Strong ‘Hands-on Agile’ Slack Community**

I invite you to join the “Hands-on Agile” Slack Community and enjoy the benefits of a fast-growing, vibrant community of agile practitioners from around the world.

If you would like to join all you have to do now is provide your credentials via this Google form, and I will sign you up. By the way, it’s free.

Prepare yourself for avoiding Gaming Velocity by studying the free Scrum Anti-Patterns Guide:

Find this content useful? Share it with your friends!

pocket

Tags:

Agile Metrics

Metrics

scrum anti-pattern

Scrum metrics

Velocity

**Published By**

Stefan Wolpers

Stefan, based in Berlin, Germany, has worked for 18-plus years as a Product Manager, Product Owner, agile coach, and Scrum Master. He is a Professional Scrum Trainer (PST) with Scrum.org and the author of Pearson’s “Scrum Anti-Patterns Guide.”

He has developed B2C as well as B2B software, for startups as well as corporations, including a former Google subsidiary.

Stefan curates the ‘Food for Agile Thought’ newsletter and organizes the Agile Camp Berlin, a Barcamp for coaches and other agile practitioners.

View all posts by Stefan Wolpers

This site uses Akismet to reduce spam. Learn how your comment data is processed.
24/07/2024 22:40 – From Outputs to Outcomes
#Mindset #ScrumSoftwareDevelopment #Cognition #SocialPsychology #BranchesOfScience #BehaviouralSciences #CognitiveScience #Empowerment #Leadership #Psychology #Learning #Science #Management #Education #Autonomy #Business #Experience #Experiment

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
***FROM OUTPUTS TO OUTCOMES***

by Viktor DidenchukJuly 24th, 2024**Too Long; Didn't Read**

Shifting from output to outcome can be stressful but rewarding. Empower your Scrum teams with guidance, embrace gradual progress, and prioritize moving in the right direction over rushing against time. It's a marathon, not a sprint.

Empowerment is widely endorsed. There is consensus that top management should trust teams to solve problems and provide them with the autonomy to achieve goals and milestones. This trust is not just a formality but a recognition of the value and expertise that teams bring to the table. However, the challenge lies in transitioning teams from a feature-factory mentality to a value-driven approach.

Development teams often need to prepare for such a shift to gain experience in taking on this level of responsibility. Let's face it: being accountable for outcomes is much more complex than simply delivering fixed tasks with specific features. Despite the increased risk associated with focusing on outcomes, it is far more interesting and exciting for everyone involved. Remember when time used to go much slower when we were kids? Scientists say that is because we explore the world and learn something new daily.

Here are some strategies to effectively transition from focusing solely on output to achieving meaningful outcomes. Overcome the feature-centric mindset once and for all.

**Mindset Difference.**

For a long time, success was delivering features within deadlines and scope. I excelled at meeting stakeholders’ wants. They were happy with the results, and we were comfortable with how we worked.

There was no room for confusion. Everyone understood their tasks and deadlines without problem.

Every Sprint followed a similar pattern: features, features, features, a bug fix here and a refactor there, but nothing more. Our primary responsibility was to deliver outputs on time, a task we were adept at fulfilling. However, with the restructuring of the leadership team, expectations shifted. They emphasised achieving goals rather than dictating specific tasks.

When they stopped telling us what to do, we panicked.

We needed guidance as to where even to start. The questions running through our minds were:

Where should we start?

What if we make a wrong decision? Will we be held accountable?

How do we define success now?

It was embarrassing. We wanted responsibility and autonomy, and that’s exactly what we got. The real challenge, however, lay in figuring out how to deal with it. We were empowered to achieve goals but needed to prepare for the task.

**Output Vs Outcome.**

Until then, we had solely concentrated on outputs, completely unaware of the true essence of outcomes. Let’s try to summarise the difference.

An output is created for customers to use, such as product recommendations in the e-commerce industry. These recommendations are designed to deliver value. Their outcomes assess the effectiveness of outputs. For instance, when customers add recommended products to their carts, increasing sales represents an outcome.

Now, here's the exciting part: Various outputs can achieve outcomes. It's like solving a problem where multiple alternatives exist.

When you begin with the outcome in mind, you unlock diverse possibilities and increase your chances of creating significant value.

**Start Small.**

Getting someone who is already an expert on the topic will make your life much easier. To get where you’ve never been, seek guidance from someone already there!

I froze when I first faced true empowerment, and so did everyone else. Despite our attempts to shift focus towards outcomes, our lack of experience inevitably led to failure, as anticipated.

We struggled with conducting discovery experiments;

Developers expected precise requirements from me as usual, and;

Stakeholders prioritised feature predictability;

The dynamics began to shift when an experienced person joined our team. Instead of overhauling everything at once, we transitioned gradually from focusing on outputs to emphasising outcomes.

Our Sprints primarily focused on delivering outputs, with some items oriented towards achieving outcomes. This mixed approach allowed us to change to a new working method gradually. For instance, in one Sprint, we allocated approximately 70% of our capacity to completing a pre-defined integration with partners and 20% to exploring methods to reduce our churn rate.

Through this process, we discovered the barriers preventing us from fully embracing the unknown:

We hesitated to provide estimates for ambiguous requirements due to past blame for unmet deadlines;

We avoided exploring alternative solutions, fearing the prospect of failure and criticism;

The uncertainty of the unknown made us uneasy as it challenged our accustomed security;

Gradually, however, clarity emerged, and we began transforming our work approach.

**Mindset Transformation Is Required.**

Transitioning from focusing on output to prioritising outcome is challenging and demanding, as it upends everything you thought you knew about your work methods. With an output mindset, decisions typically revolve around how to deliver predefined features.

When a value-driven mindset with a focus on outcomes is an adapter, you need to make decisions frequently—some say even 10 times as many. You must choose problems to solve, find solutions, test them, inspect them, and adapt. It’s more demanding, without a doubt.

It also means that you are still working when customers receive the functionality. It is crucial to understand how it works, what they like, how to improve, and how to measure the outcome.

So what can help:

Coaching: Ask your team questions; do not provide answers. The brains of an entire team have much more computing power than those of a Product Owner or manager;

Problems over solutions: Focus on understanding the problem's context, relevance, and impact before discussing solutions;

Learning over predictability: Prioritizing learning over predictability involves embracing the unknown through frequent small experiments and rapid end-user learning. Understanding what not to build is as crucial as identifying what to make, aiming to create value while minimising waste;

Evidence over opinions: Moreover, decisions should be guided by objective evidence from experiments rather than opinions. This is about having a data-driven mindset;

Once you've shifted your mindset to focus on outcomes, the orientation towards achieving meaningful results becomes motivating and inspiring. Every Sprint becomes a different story.

**So What?**

Moving from output to outcome is stressful. The burden on your shoulders will be heavy, but you’ll never learn so many new things.

Scrum teams yearn for empowerment but can feel overwhelmed when it arrives. Navigate this shift with guidance from experienced mentors and embrace gradual progress.

Remember, it's a marathon, not a sprint. Prioritise moving in the right direction over rushing against time.

L O A D I N G. . . comments & more!About AuthorViktor Didenchuk@viktordidenchukProject management enthusiast. Agile Coach.Read my storiesRead My StoriesTOPICSproduct-management #agile #outcomes #transformation #feature #stakeholders #start-small #product-management #project-managementTHIS ARTICLE WAS FEATURED IN... Terminal LiteRELATED STORIESSecure your critical AI workloads!visit Tailscale #SponsoredThe Rabbit Hole: Local Optimizations That Harmby viktordidenchuk Jul 17, 2024 #agileThe Noonification: The Future of DeFi - Emerging Evolution (11/3/2022)by noonification Nov 03, 2022 #noonification10 Ways PMs Can Earn The Respect and Trust of Their Teamby johnpcutler Oct 07, 2016 #product-management10 Terms Every Engineer Must Knowby karanmg May 03, 2017 #product-management10 Subtle Superpowers of DevOpsby FedakV Jan 30, 2018 #devopsJoin HackerNoonLatest technology trends. Customized Experience. Curated Stories. Publish Your Ideas
24/07/2024 22:37 – Why C-Suite Executives Won’t Cut it Without Data Skills Anymore
#ComputerSecurity #DataAndInformationVisualization #Data #Governance #DataScience #Risk #InformationTechnologyManagement #Computing #Technology #InformationTechnology #InformationScience #DataManagement #DecisionMaking #Management #Privacy #DataAnalysis #Analytics #Business #ArtificialIntelligence #DataGovernance #BigData #PrivacyConcernsWithSocialNetworkingServices #InformationRetrieval #TableauSoftware #CognitiveScience #BranchesOfScience #Audit #Cyberspace #Science #InformationPrivacy #RegulatoryCompliance #Information #Preference #EducationalTechnology #ComputerScience #Communication #Democratization #Microsoft365 #Cognition #Literacy

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
New Story***WHY C-SUITE EXECUTIVES WON’T CUT IT WITHOUT DATA SKILLS ANYMORE***

by Nenad ZaricJuly 24th, 2024RUTRKODEBNESHIZHVIFRPTJAEN**Too Long; Didn't Read**

Every industry generates massive amounts of data, which is now being used for better decision-making. One of today’s most pressing challenges for executives is data privacy concerns and cybersecurity. Modern executives must have data skills to understand the flow of valuable data in their company and know how to make it work for them.‘A cartoon CEO questioning numbers on a digital chalk board’ Image created by HackerNoon AI Image GeneratorBy Nenad Zaric, Co-founder and CEO at Trickest

Every industry generates massive amounts of data, which is now being used for better decision-making — tech giants understand this. For example, streaming powerhouse Netflix implements data analytics to study viewers’ visual preferences and adjusts content thumbnails to similar content its users have liked to increase viewership.

The rise of AI is further influencing this trend, helping companies make sense of their datasets to uncover weaknesses and predict outcomes.

However, these advancements are no longer exclusive to engineers, managers, and IT teams who grapple with data daily. This information is increasingly climbing the ladder and knocking on executive’s doors. In fact, one of today’s most pressing challenges for executives is data privacy concerns and cybersecurity.

It has never been more important to make key decisions toward better data governance, and data skills are necessary to understand the specific risks and demands a company faces.

So, reading reports an engineer crafted with dedicated software programs is no longer enough. A concerning report revealed that only half of C-suite executives were confident in their data literacy, and 45% made decisions based on their gut feeling instead of data-backed facts.

Modern executives must have data skills to understand the flow of valuable data in their company and know how to make it work for them, not the other way around. From crucial decision-making to infusing data culture and enabling democratization, positive company changes stemming from data analysis must start at the top.

**Spot Pain Points In Minutes, Not Days**

When a company decides to change directions, these choices begin at the executive level and trickle down to managers and workers — nothing new there. But, while an exceptional leader has always been crucial in driving change, it’s how they do it that is shifting. C-suite directors are in charge of crafting better strategy, financial performance, risk management, and corporate governance, and data is becoming increasingly connected to all of these activities.

Imagine you’re meeting with the Head of Sales to discuss the sudden performance drop of a fan-favorite item in your online store. You can’t seem to connect the dots from the sales report you’ve received, and it becomes a hurdle to decode the issue. One skill that can help executives and department heads easily uncover patterns for weak performance is data visualization.

Tableau, Microsoft Power BI, and Looker Studio are some of the most prominent data visualization tools. They are easily accessible from popular corporate programs (Salesforce, Office 365, and Google Studio) and help analyze massive amounts of data in visually appealing ways.

While you don’t need to be a data scientist to use them, setting up and building charts requires certain data skills and tech acumen, or at least prior training. Whether from a data science and visualization course at Harvard, or self-taught online classes, a few hours and hands-on practice will be time well spent to hone your data visualization skills.

With this newfound talent, you can aid and guide department leaders to the answers to their performance questions. In this case, you would pivot to the online sales data in the company database using visualization to track downtime and realize the fan-favorite item’s link has been down, leading to that drop in sales.

Expertise in data visualization tools allows executives to stay ahead of the curve in every business vertical by quickly analyzing pain points and understanding what their verticals need. This leads to improved strategies and more accurate decision-making at a fraction of the time it takes to crunch vital datasets.

**KYC: Know Your Cybersecurity**

While data management isn’t an executive’s role in itself, knowing its ins and outs allows them to understand the company's successes and areas for improvement. When data increasingly plays a vital role in a business, ensuring that it is being properly managed and secured can save leaders major headaches in the future.

Does your company practice master data management? Does it have an outdated data lake that’s slowing down processes? How is sensitive data access managed? These are governance questions investors, stakeholders, and partners might ask, and you must be ready to understand and navigate these concepts.

Moreover, adopting good cybersecurity practices starts at the top — especially when 63% of security professionals say users with access to critical business data are their biggest data risk. Knowing the basics of cybersecurity and current trends not only allows top-level roles to mitigate risks from their end but also helps them foresee their IT teams’ needs, like allocating a budget for better endpoint security and enforcing staff training.

Beyond courses and academic work, data governance and cybersecurity at a C-suite level are more easily learned through conversations with experts. You can resort to networking and approaching other industry leaders and even your own staff to gauge these concepts and what they mean for your company.

As a result, instead of wrestling to understand the demands of security professionals, data-savvy executives will know when needs are pressing or if they can wait, making wiser and more objective business choices alongside their team leaders.

**Make Data Work For You — Not The Other Way Around**

There was a time when data was only one of many business assets, something you generate but store for audits or other compliance purposes. Those days are gone, especially as we move toward data-driven solutions that thrive primarily on how a company can implement data for other use cases.

Data is no longer a stagnant asset — it can make you money if you derive the right value from it.

Knowing the value of your data is only possible if you’re well-informed about it and know its potential. Executives with data skills will be able to take the lead and inject their data into better risk management strategies, extract customer data to improve marketing, and further implement it into AI tools to speed up business processes.

Grasping these concepts also impacts the way you communicate with stakeholders. Expressing the importance of data security, governance, and even democratization for your company must come across as more than numbers and graphs — they wish to know the real value of these practices. By clearly explaining the importance of safeguarding data and placing it front and center of your operations, stakeholders will be on the same page about business decisions.

Simply put, skipping the data-related courses will do nothing for future executives. In fact, nowadays, being able to run a successful company is directly linked to your data knowledge. So, prioritize these skills if you don’t have them, and make sure to pitch them as a worthwhile talent when climbing the ladder; they will be key to your success.

L O A D I N G. . . comments & more!About AuthorNenad Zaric@znenad079Cybersecurity expert and CEO of Trickest, an IDE for bug bounty hunters, penetration testers, and SecOps teams.Read my storiesRead My StoriesTOPICSdata-science #data-literacy #thought-leadership #leadership-skills #data-skills #data-governance #data-visualization-tools #cybersecurity-executives #hackernoon-top-storyLanguagesEnglishहिंदीDeutschРусскийPortuguês한국인FrançaisবাংলাEspañol中国人Tiếng Việt日本語TürkçeTHIS ARTICLE WAS FEATURED IN...Permanent on Arweave Terminal LiteRELATED STORIESSecure your critical AI workloads!visit Tailscale #Sponsored15 Advanced Cybersecurity & Web 3.0 Executivesby thesociable Aug 28, 2023 #cybersecurity4 Ways Businesses Can Practice Effective Data Governanceby aprilmiller Mar 30, 2023 #dataAdvancing User Data Governance with Data Lineageby maharshijha May 30, 2023 #data-lineageAlexey Artemov: The Retail Industry Will Be Disrupted Thanks to Data Governance Solutionsby marlenereports Dec 21, 2023 #data-scienceBuilding a Data Management Strategy: Importance, Principles, Roadmapby itrex Nov 10, 2021 #data-managementJoin HackerNoonLatest technology trends. Customized Experience. Curated Stories. Publish Your Ideas
24/07/2024 22:35 – The Role of AI in FinTech: Innovations and Applications
#Fintech #ArtificialIntelligence #RiskManagement #Finance #CreditRisk #Bank

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
Credit

Financial technology, or FinTech, has revolutionized the financial services industry by introducing innovative solutions for consumers and businesses alike. Central to this transformation is artificial intelligence (AI), which brings a new level of efficiency, personalization, and security to financial operations.â

As AI platforms gain massive popularity, their applications in FinTech are reshaping how financial services are delivered and experienced.

The global AI in FinTech market size in 2021 was a mere USD 13.14 billion, and by 2029, it is expected to grow at a CAGR of 66.20% to reach USD 765.34 billion. This rapid growth underscores the significant impact AI is having on the industry.

In this article, we will explore the key areas where AI is making significant strides in FinTech, along with the benefits and challenges it presents.

**Enhancing Customer Interaction With AI Chatbots**

Discussions about AI in FinTech often highlight the inefficiencies in traditional banking services, such as slow service delivery and server crashes during peak times. AI chatbots have becomeâ a powerful solution to these issues by providing 24/7 customer support. These chatbots can address user queries and facilitate access to financial services, thereby enhancing customer experience significantly.

Key Benefits of AI Chatbots:

Availability: Chatbots are available round-the-clock, ensuring customer support is just a message away, anytime, anywhere.

Personalization: AI chatbots provide tailored responses based on user data, making interactions more relevant and satisfactory.

Efficiency: With faster response times and minimal errors, chatbots streamline customer service operations, reducing wait times and improving user satisfaction.

Example:

Many banks have successfully integrated AI chatbots into their systems. For instance, users can get immediate answers to their questions regarding account balances, recent transactions, pending direct deposits, or investment advice, all without human intervention.

**Competitive And Business Intelligence**

One significant advantage of AI in FinTech is its ability to enhance business intelligence. Businesses can analyze enormous volumes of data to obtain insightful knowledge by utilizing AI-powered advanced data analytics capabilities. According to almost all (99%) of the financial services leaders surveyed, their companies used artificial intelligence (AI) in some capacity.

Applications:

Market Analysis: AI algorithms can detect patterns in market data, providing insights into customer behavior and market trends.

Strategy Development: Businesses can use these insights to refine their strategies, improve services, and gain a competitive edge.

Operational Efficiency: By streamlining data processing and analysis, AI reduces the time and cost associated with traditional methods.

Example:

A FinTech company might use AI to analyze competitorsâ€™ data, revealing how users interact with their products. This information can guide the company in optimizing its service offerings.

**Asset Management And Financial Advisory**

AI is also making waves in asset management by providing sophisticated tools for portfolio management and financial advising. Complex algorithms allow for the creation of resilient simulations and forecasts, making investment decisions more accurate and less reliant on human judgment.

Benefits:

Automated Services: AI-driven wealth management tools can cater to individual investment needs without needingâ a financial advisor.

Cost Reduction: Removing intermediaries in asset management lowers operational costs and makes financial advisory services more accessible.

New Opportunities: AI enables the development of new financial services, such as personalized investment plans and automated trading systems.

Example:

Many investment funds now rely on AI to manage portfolios, allowing users to conduct trades and manage assets directly from their devices, bringing a new level of convenience and efficiency.

**Risk Management And Fraud Detection**

AIâ€™s role in FinTech extends to risk management, particularly in insurance and loan assessments. Advanced algorithms can evaluate risk more accurately than traditional methods, leading to better decision-making. The operations business segment had the highest AI adoption rate in 2023, with risk and compliance following closely behind.

Features:

Fraud Detection: AI systems identify suspicious behavior and potential fraud, offering greater security for financial transactions.

Credit Scoring: AI evaluates financial habits and credit risk, helping lenders make more informed decisions while reducing biases.

Insurance Evaluation: AI tools can assess an individualâ€™s risk level by analyzing their behaviors and activities, leading to more accurate insurance policy pricing.

Example:

AI is used in the automobile industry to track driving behavior and assess risk levels, which can be applied to insurance premiums, ensuring fair and tailored pricing for customers.

**Overcoming Challenges In AI Adoption**

While the benefits of AI in FinTech are clear,several challenges must be addressed to fully realize its potential. High costs, resistance to AI, and a lack of skilled professionals are significant barriers.

Challenges:

High Costs: Implementing AI systems requires substantial financial investment, which can be a deterrent for smaller FinTech firms.

Resistance to Change: There is a general apprehension towards adopting new technologies, which can hinder AI integration.

Skill Shortage: A lack of AI expertise in the workforce makes it difficult for companies to implement and manage AI solutions effectively.

**Conclusion**

AI is undeniably a game-changer in the FinTech industry, offering transformative solutions that enhance customer interaction, improve business intelligence, streamline asset management, and strengthen risk management. While challenges such as high costs and resistance must be addressed, the potential for AI to revolutionize FinTech is immense.â

By embracing AI, FinTech companies can improve their services and secure a competitive edge in a rapidly evolving industry. The integration of AI in FinTech is not just beneficial but essential for driving innovation and efficiency in financial services.

Share via:

AITJ Staff Writer

AI Time Journal Staff Writers report on the AI technology advancements and opportunities across industries to leverage AI.

Categories: FinTech
24/07/2024 22:22 – A guide on reinforcement learning with human feedback
#ReinforcementLearning #ReinforcementLearningFromHumanFeedback #ProximalPolicyOptimization #ArtificialIntelligence #MachineLearning #Learning #ComputationalNeuroscience #AppliedMathematics #Cognition #CognitiveScience #Reinforcement #BehaviorModification #MathematicalOptimization #OpenAI #Logarithm #BranchesOfScience #ChatGPT #AIAlignment #Simulation #Technology #FunctionMathematics

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
by Sanksshep Mahendra
July 10, 2023, 10:00 am

**Introduction**

Reinforcement Learning with Human Feedback (RLHF) offers a fresh avenue for training machines to solve complex tasks where reward functions are challenging to define. Here, the model learns an optimal policy (a policy is essentially the â€œbrainâ€ of an agent) through human guidance rather than direct observation of reward data. RLHF is widely applicable, seen in large language models, robotics, auto-driving, clinical trials, and more.

Through reinforcement learning from human feedback (RLHF), a GPT model can be taught to be more accurate, truthful, and less prone to â€˜hallucinatingâ€˜. Humans can guide the model to understand its â€˜state of knowledgeâ€™ more clearly, express uncertainty where necessary, and avoid guessing responses.

Screenshot â€“ ChatGPT

In this comprehensive guide, we delve into the fundamentals, approaches, challenges, and applications of Reinforcement Learning with Human Feedback (RLHF). You will also learn about a successful case study, OpenAIâ€™s ChatGPT, and its implementation of RLHF.

**Table Of Contents**

**Reinforcement Learning (RL) And Human Feedback (HF)**

Reinforcement Learning, problems are typically modeled as a Markov Decision Process (MDP), specified by the tuple (S, A, T, R) defining the set of possible world states, actions available to the agent, transition function, and a reward function. The role of an RL algorithm is to identify a policy that maximizes the expected reward from the environment.

The fundamental objective of an RL agent is to learn an optimal policy Ï€* : S â†’ A, which is a mapping from states to actions. The optimal policy tells the agent what action to take in each state so as to maximize the expected cumulative reward over the course of many steps.

An agent interacts with the environment in a series of discrete time steps. At each time step t, the agent observes the current state s_t, selects an action a_t according to its policy, receives a reward r_t from the environment, and transitions to a new state s_{t+1}.

Value functions quantify the â€˜goodnessâ€™ of a state or action for the agent. A state-value function V(s) under a policy Ï€ gives the expected cumulative reward from starting in state s and following Ï€ thereafter. Similarly, an action-value function Q(s, a) under a policy Ï€ gives the expected cumulative reward from starting in state s, taking action a, and following Ï€ thereafter.

Mathematically, the state-value function is defined as:

FormulaeFormulae

Where Î³ is the discount factor that determines the present value of future rewards. Itâ is a number between 0 and 1, usually close to 1. For instance, Î³ might be set as 1/(1+r) for a certain discount rate r. A lower discount factor nudges the decision-maker to favor immediate actions, as opposed to delaying them indefinitely.

A policy that optimizes the aforementioned function is referred to as an optimal policy, commonly denoted by Ï€*. Interestingly, an MDP might have several distinct optimal policies. However, owing to the Markov property, itâ€™s shown that the optimal policy is a function of only the current state, thus simplifying the task of optimal policy determination.

To further measure the quality of policy Ï€, the Q-value function, Q_Ï€(s, a), is introduced which represents the expected long-term return, starting from a specific state-action pair (s, a):

Formulae

A deterministic policy Ï€ : S â†’ A signifies a policy where given a state, the action is determined without any randomness. The expected long-term return, V(Ï€), under such a policy can be represented as:

Formulae

The cornerstone of RL is the Bellman Optimality Equation, which states that the optimal Q-value function, Q*, satisfies the following:

Formulae

This equation expresses the fact that the value of a state-action pair (s, a) is the immediate reward r(s, a) plus the maximum discounted (Î³) future value achievable from the next state sâ€™.

At the heart of RL lies the process of exploration and exploitation. Exploration helps the agent find more information about the environment, while exploitation helps the agent make the best decision based on its current knowledge. Balancing these two conflicting objectives is still a challenge in RL.

An agent excessively focused on exploration might squander opportunities for obtaining known rewards. Conversely, an agent overly concentrated on exploitation might miss out on discovering potentially more rewarding experiences.

By continuously fine-tuning this balance, we can design RLHF agents that learn more efficiently and perform more robustly, even in complex environments. Achieving this equilibrium enables the agent to maximize its understanding of human feedback while also being versatile enough to discover and adapt to new scenarios.

**Approaches To Incorporate Human Feedback Into RL**

There are several ways to incorporate human feedback into reinforcement learning:

Reward Shaping: Humans provide additional rewards to guide the agentâ€™s learning process. This is typically done by adding a reward function that provides extra rewards based on human intuition or domain knowledge.Policy Shaping: Human feedback directly influences the agentâ€™s policy, for example, by advising the agent on which actions to take during learning.Inverse Reinforcement Learning: Here, the agent learns from observing human behavior. The underlying assumption is that human actions implicitly reveal their reward function.Learning from Comparisons: Humans compare two or more actions or trajectories, and the agent learns the better option from these comparisons.

These methods are not mutually exclusive and can be used in combination to provide more robust and effective learning.

**Human Feedback Collection And Annotation**

Leveraging the insights from human trainers, RLHF systems can effectively utilize human annotations and advice, accelerating the learning process and aligning AI behaviors with human values. This involves humans interacting with the learning system, either by providing feedback on the systemâ€™s actions, comparing different actions or trajectories, or demonstrating the correct behavior. This feedback is then used to inform the reinforcement learning process.

Feedback can be provided in a number of ways, including:

Ranking or comparison: The system presents multiple potential actions or outputs to the human, who then rank them based on their quality or appropriateness.Direct feedback: The human provides feedback on specific actions taken by the system, such as indicating whether an action was correct, providing a numerical score, or even providing a corrected version of the action.

Collecting feedback effectively and efficiently is crucial, as the quality and quantity of feedback directly impact the systemâ€™s ability to learn and improve.

**Algorithms For Reinforcement Learning Human Feedback**

Several algorithms have been developed for RLHF, each with their unique strengths and challenges. These algorithms combine reinforcement learning techniques with human feedback to improve the learning process. Here are a few notable ones:

Proximal Policy Optimization (PPO): PPO is a policy optimization method used widely in RL. Itâ€™s designed to address the challenges of policy optimization in complex environments.

Deep Deterministic Policy Gradient (DDPG): DDPG is an off-policy algorithm and an actor-critic method that concurrently learns a Q-function and a policy. It uses off-policy data and the Bellman equation to update the Q-function and uses the Q-function to update the policy.

Cooperative Inverse Reinforcement Learning (CIRL): CIRL is a framework for RL where both the human and the AI agent cooperate to achieve a shared goal, with the agent trying to learn the humanâ€™s reward function through their actions.

**ChatGPT: A Success Story In RLHF Beyond GPT-4**

Deep Reinforcement Learning combines deep learning and reinforcement learning techniques to enable AI systems to learn the best actions possible in virtual simulation environments.

Proposed by OpenAI, combines RL and HF to create safer AI models. RLHF utilizes human feedback to solve modern RL environments, as demonstrated in Atari games and MuJoCo simulations. The novel aspect of RLHF is that the reward function is learned rather than hand-crafted, bridging the gap between human understanding and AI learning.

Flow ChartStep 1: Data Collection and Human Evaluation

In the first step, the goal is to collect samples from existing policies and send comparisons for human evaluation. For any given task, these samples could come from a variety of sources, including the current policy (the current state of the machine learning modelâ€™s behavior), the initial policy (the modelâ€™s behavior at the start of training), and various baseline models.

Hereâ€™s how it works:

Generate outputs for a range of input data points using the different policies mentioned above.For each data point, select pairs of outputs and send these pairs to human evaluators.The evaluators are tasked with judging which output from each pair is better.

For example, in a machine learning task of image classification, the modelâ€™s â€œoutputsâ€ could be the predicted labels for a set of images, and the human evaluators would judge which set of labels is more accurate.

Step 2: Learning a Reward Model

The second step involves training a reward model based on the human evaluations from Step 1. This is a type of supervised learning, where the input is an output from the policy (e.g., a classification label, a generated sentence, etc.), and the target is the human evaluation of that output.

Specifically, for each input-output pair, the reward model is trained to predict the log odds that a particular output is the better one, as judged by the human evaluators. The log odds is simply the logarithm of the probability of an event happening divided by the probability of it not happening. If P is the probability of the event (in this case, the output being the better one), then the log odds are calculated as log(P / (1 â€“ P)).

This reward model can be any type of model that is suitable for the task at hand, such as a neural network for complex tasks or a simpler model for less complex tasks.

Step 3: Optimizing the Policy

The third step involves optimizing the policy based on the reward model from Step 2. This is where reinforcement learning comes into play.

The output of the reward model â€” the logit output, which is the input to the final activation function of the model â€” is treated as a reward signal. This reward signal is then used to optimize the policy using reinforcement learning.

The specific algorithm used may vary, but Proximal Policy Optimization (PPO) is a popular choice due to its stability and efficiency. In PPO, the objective is to maximize the expected reward while keeping the new policy close to the old policy.

This is done by optimizing the following objective function:

Formulae

Here, Î¸ represents the parameters of the policy, E_t is the expectation over the time steps, r_t(Î¸) is the ratio of the probabilities under the new and old policies, A_t is the advantage function at time step t, and Îµ is a hyperparameter that controls how much the policy is allowed to change at each step.

By iterating through these steps â€” collecting data, learning a reward model, and optimizing the policy â€” the machine learning model gradually improves its performance on the task at hand.

**Applications Of Reinforcement Learning With Human Feedback (RLHF)**

RLHFâ€™s application spans various domains, particularly in natural language processing (NLP), where it has been instrumental in refining conversational agents, text summarization, and natural language understanding. It allows language models to generate more verbose responses, reject inappropriate or out-of-scope queries, and offer outputs that align with intricate human values and preferences. OpenAIâ€™s ChatGPT and InstructGPT, and DeepMindâ€™s Sparrow are remarkable examples of RLHF-trained language models.

Employing pre-trained models in the context of RLHF can drastically accelerate the learning process, providing the AI agent with a robust starting point. These models, having undergone extensive training in diverse environments, equip the agent with a well-rounded understanding of various scenarios, enabling it to respond more effectively in novel situations.

RLHF also has a remarkable presence in the gaming industry. Esteemed entities such as OpenAI and DeepMind have harnessed the power of RLHF to coach AI agents in mastering Atari games, basing their strategies on human preferences. Notably, these RLHF-trained agents have often managed to exceed the performance of human players, displaying an impressive capability to learn and adapt.

**Challenges Of Reinforcement Learning With Human Feedback (RLHF)**

Reinforcement Learning from Human Feedback (RLHF) has shown immense promise in training artificial intelligence systems. However, it is accompanied by a host of challenges that researchers are currently grappling with. Here are five significant challenges they face:

Failures and Regularization Needs: RLHF models sometimes fail unexpectedly. Itâ€™s unclear whether these failures will disappear as capabilities scale. Regularization is crucial to prevent the model from diverging drastically from the desired outcomes. But determining the constant of regularization often feels arbitrary and brings its own challenges, such as potential mode collapse â€“ an overbearing bias towards certain patterns or completions.

Incentives Issues and Goal-Directedness: RLHF can make AI systems overly goal-directed. In an effort to achieve the desired outcome or reward, these systems may end up becoming too narrow in their focus. For instance, an AI model trained using a positive sentiment reward system might obsessively generate content around wedding parties, perceived as extremely positive, irrespective of the prompt given.

Opaque Thoughts and Capability Externalities: RLHF might lead to models that â€œthinkâ€ in opaque ways, making their decision-making process difficult to interpret. The added capabilities resulting from RLHF might exceed base models, potentially leading to harmful externalities. As the capability increases, it could further shrink timelines, exacerbating the issue.

Human Feedback Limitations: The RLHF process requires significant human feedback. But this approach suffers from human limitations, including error-prone evaluations and potential biases. The cost of obtaining necessary data might become prohibitive as systems advance, requiring more complex feedback. Simultaneously, the number of qualified human annotators might fall short. These challenges can lead to inconsistencies in the training process and may not always result in the desired AI behavior.

**Conclusion**

Interactive Machine Learning has seen a substantial rise, with an increased interest in applying Reinforcement Learning from Human Feedback. RLHF techniques have considerable potential in Autonomous Systems, improving their learning efficiency and adaptability to new scenarios. Generative Models have found significant application in RLHF, serving to predict potential future scenarios for the agent to learn from.

The guide provides an insightful exploration into the application of reinforcement learning with human feedback, a field with vast potential for diverse applications, including autonomous driving.

Reinforcement learning models provide a powerful framework for training agents to interact with uncertain environments, creating the capability to map observations from the observation space to actions within the action spaces effectively. When coupled with human demonstrations, these models become more robust and effective, allowing for a form of learning that aligns well with human advice.

Our understanding of training language models is critical in this context. The use of human language as a medium for providing feedback and advice, as a human reinforcement function, serves as a more natural way to impart complex knowledge to reinforcement learning models. This approach also offers an intuitive platform for providing corrective demonstrations when the modelâ€™s actions diverge from the expected behavior.

The use of a loss function is integral to the reinforcement learning process, guiding the learning towards minimizing errors and optimizing performance. Reward engineering also plays a significant role in shaping the behavior of reinforcement learning models. It guides the agent transitions from its initial model to a proficient actor within the environment, leading to the successful completion of tasks.

The integration of unsupervised learning presents an exciting avenue for future development, enabling models to extract useful features from raw data autonomously, reducing reliance on manual feature engineering, and improving the modelâ€™s adaptability.

Overall, the synergistic combination of reinforcement learning with human feedback, coupled with strategies such as the use of corrective demonstrations, introduces a robust, flexible, and intuitive approach to creating powerful, responsive agents. This guide underscores the potential of these techniques in fostering enhanced machine learning models capable of operating efficiently and effectively in a myriad of complex and uncertain environments.

Age of Invisible Machines: A Practical Guide to Creating a Hyperautomated Ecosystem of Intelligent Digital Workers$23.49Buy NowWe earn a commission if you make a purchase, at no additional cost to you.02/19/2024 01:51 am GMT **References**

Bonaccorso, Giuseppe. Machine Learning Algorithms. Packt Publishing Ltd, 2017.

Munro, Robert, and Robert Monarch. Human-in-the-Loop Machine Learning: Active Learning and Annotation for Human-Centered AI. Simon and Schuster, 2021.

Sutton, Richard S., and Andrew G. Barto. Reinforcement Learning, Second Edition: An Introduction. MIT Press, 2018.

Share this:Click to share on Twitter (Opens in new window)Click to share on Facebook (Opens in new window)Click to share on LinkedIn (Opens in new window)Click to share on Reddit (Opens in new window)Click to share on Pinterest (Opens in new window)Click to share on Pocket (Opens in new window)x
24/07/2024 22:21 – Machine Learning for Kids: Installing Python
#ComputerProgramming #MachineLearning #InstallationComputerPrograms #CommandLineInterface #PythonProgrammingLanguage #Learning #Computer #ComputerFile #StartMenu #DoubleClick #Software #ComputerScience #Technology #Computing #SoftwareEngineering #InformationTechnology #Computers #ComputerEngineering #SystemSoftware #ArtificialIntelligence #SystemsEngineering #Data #BranchesOfScience #SoftwareDevelopment #Creativity #ProblemSolving #Experience #ProgrammingLanguage #Algorithm #CognitiveScience #SpotlightApple #Cognition #DirectoryComputing #Intelligence #Science #Machine #DigitalMedia #OperatingSystemFamilies #OfficeEquipment #ComputingPlatforms

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
by Sanksshep MahendraJuly 13, 2023, 12:08 pm

**Introduction**

Welcome to this tutorial on how to install Python on Windows and MacOS! Python is a versatile and powerful programming language thatâ€™s great for beginners and experts alike. Itâ€™s also the language of choice for many data scientists, machine learning, web developers, and software engineers around the world. Whether youâ€™re new to coding or just new to Python, this step-by-step guide will walk you through the installation process on both Windows and MacOS. By the end of this tutorial, youâ€™ll have Python set up on your computer and be ready to start programming!

Also Read: Coding for Kids: How to Make Pacman on Scratch

**Table Of Contents**

**Who Is This For?**

Grade: 6th to 10th

This Python installation tutorial is specifically designed with students in grades 6 through 10 in mind. We understand that you may be just starting your journey into the world of coding, and the idea of setting up a programming environment might seem daunting. Donâ€™t worry! Weâ€™ve broken down the process into easy-to-follow steps, with clear instructions and explanations. You donâ€™t need any previous coding experience to get Python up and running on your computer. This guide will serve as a great starting point for your exciting exploration into Python programming and the wider field of computer science and machine learning. Get ready to step into a new world of creativity, problem-solving, and digital innovation!

**What Will We Learn?**

In this journey of â€œMachine Learning for Kidsâ€ lets learn the basics of python which is the foundation of machine learning. Machine learning, a subset of artificial intelligence, is all about teaching computers to learn from data, just like humans do from experience. We will start by understanding what machine learning is and explore its various types. Weâ€™ll then have fun creating, training, and testing our very own machine learning models using Python! By demystifying complex concepts and using engaging, real-world examples suitable for grades 6 through 10, weâ€™ll transform seemingly intricate algorithms into understandable, hands-on projects. So gear up to embark on this exciting expedition where weâ€™ll turn computers into our learning companions!

Also Read: The Role of Artificial Intelligence in Education

**Installing Python For Windows**

Step 1: Download Python Installer

Go to the Python downloads page at https://www.python.org/downloads/windows/.

Click on the latest Python release. Python 3.9.7 is the latest stable version, but you may see a newer version.

Step 2: Run Python Installer

Once the installer is downloaded, locate the file (usually in your Downloads folder), and double-click it to run the installer.

A window will pop up, providing several installation options.

Step 3: Customize Installation

At the bottom of the window, check the box next to â€œAdd Python to PATHâ€. This step is important because it allows your computer to find Python whenever it needs to run a Python program.

Click on â€œInstall Nowâ€. This option will install Python with default settings, which should be fine for most users.

Step 4: Verify Installation

After the installation completes, you can verify it by opening a new command prompt (you can search for â€œCommand Promptâ€ in the start menu).

In the command prompt, type python --version and press Enter. It should display the Python version you just installed, like Python 3.9.7.

python --version**Installing Python For Mac**

Step 1: Download Python InstallerGo to the Python downloads page at https://www.python.org/downloads/mac-osx/.

Click on the latest Python release. Python 3.9.7 is the latest stable version, but you may see a newer version.

Step 2: Run Python Installer

Once the download completes, locate the file (usually in your Downloads folder), and double-click the .pkg file to run the installer.

A new window will pop up, guiding you through the installation process.

Step 3: Customize Installation

Click on â€œContinueâ€ to proceed through the installerâ€™s steps. Accept the license and confirm the installation location (default location should be fine for most users).

Click â€œInstallâ€. You may need to enter your system password to authorize the installation.

Step 4: Verify Installation

After the installation completes, you can verify it by opening Terminal (you can find Terminal through Spotlight Search, which is accessible by clicking the magnifying glass at the top-right corner of the screen or pressing Cmd+Space).

In the Terminal, type python3 --version and press Enter. It should display the Python version you just installed, like Python 3.9.7.

python3 --versionAnd thatâ€™s it! You now have Python installed on your computer. In the next tutorial, we will go over how to use Python for various tasks. Click here for the next tutorial.

Note: These instructions assume that youâ€™re installing Python on a computer where you have administrator access. If youâ€™re using a school or work computer, you might not have the necessary permissions to install software, and youâ€™ll need to ask your IT department for help.

Also Read: What Language is Minecraft Coded in? How Can Minecraft Help Kids Code?

**NEXT TUTORIALS**

Tutorial 1 â€“ Installing PythonTutorial 2 â€“â Your First Program in PythonTutorial 3 â€“â Python VariablesTutorial 4 â€“â Python Data TypesTutorial 5 â€“â Python ConditionalsTutorial 6 â€“â Python LoopsTutorial 7 â€“â Python FunctionsTutorial 8 â€“â Advanced Python FunctionsTutorial 9 â€“â Starter Machine Learning Python ProgramTutorial 10 â€“ Your First Machine Learning Program!

Explore more with this book.

Python for Kids, 2nd Edition: A Playful Introduction to Programming$21.99Buy NowWe earn a commission if you make a purchase, at no additional cost to you.02/18/2024 05:01 am GMT **References**

Briggs, Jason R. Python for Kids: A Playful Introduction To Programming. No Starch Press, 2012.

Share this:Click to share on Twitter (Opens in new window)Click to share on Facebook (Opens in new window)Click to share on LinkedIn (Opens in new window)Click to share on Reddit (Opens in new window)Click to share on Pinterest (Opens in new window)Click to share on Pocket (Opens in new window)x
24/07/2024 22:21 – Machine Learning for Kids: Your First Program in Python
#MachineLearning #ComputerProgram #PythonProgrammingLanguage #TextEditor #ComputerProgramming #CommandLineInterface #ComputerFile #Learning #ArtificialIntelligence #WindowsNotepad #Software #ComputerScience #Computing #Computers #SoftwareEngineering #ComputerEngineering #Technology #CommandComputing #SoftwareDevelopment #Computer #Data #ComputerTerminal #Intelligence #InformationTechnology #SystemsEngineering #BranchesOfScience #Creativity #Machine #CognitiveScience #Algorithm #Experience #SpotlightApple #ProblemSolving #Science #ControlFlow #Cognition #Language #OfficeEquipment #DataScience #DirectoryComputing #MicrosoftWindows #ConditionalComputerProgramming

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
by Sanksshep MahendraJuly 13, 2023, 12:33 pm

**Introduction**

Machine learning for kids: Welcome to your first coding tutorial in Python! Python is an excellent language for beginners due to its readability and compact syntax. Letâ€™s start by writing a classic starter program: â€œHello, World!â€.

The language of choice for many data scientists, machine learning, web developers, and software engineers around the world. Let us get into it!

**Table Of Contents**

**Who Is This For?**

Grade: 6th to 10th

Your first program in python tutorial is specifically designed with students in grades 6 through 10 in mind. We understand that you may be just starting your journey into the world of coding, and the idea of writing your first program might seem daunting. Donâ€™t worry! Weâ€™ve broken down the process into easy-to-follow steps, with clear instructions and explanations. You donâ€™t need any previous coding experience to write your first python program. This guide will serve as a great starting point for your exciting exploration into Python programming and the wider field of computer science and machine learning. Get ready to step into a new world of creativity, problem-solving, and digital innovation!

Also Read: The Role of Artificial Intelligence in Education

**What Will We Learn?**

In this journey of â€œMachine Learning for Kidsâ€ lets learn the basics of python which is the foundation of machine learning. Machine learning, a subset of artificial intelligence, is all about teaching computers to learn from data, just like humans do from experience. We will start by understanding what machine learning is and explore its various types. Weâ€™ll then have fun creating, training, and testing our very own machine learning models using Python! By demystifying complex concepts and using engaging, real-world examples suitable for grades 6 through 10, weâ€™ll transform seemingly intricate algorithms into understandable, hands-on projects. So gear up to embark on this exciting expedition where weâ€™ll turn computers into our learning companions!

**Your First Program In Python**

Step 1: Opening a Text Editor

First, we need to write our Python program in a text editor. This could be Notepad (Windows), TextEdit (Mac), or a more advanced editor like Sublime Text, Visual Studio Code, or Atom.

Open your text editor of choice.

Start a new file.

Step 2: Writing Your Program

In the new file, type the following:

print("Hello, World!")Hereâ€™s what this code does:

print is a built-in Python function that displays text in the console.

("Hello, World!") is the text we want to display. Itâ€™s a string, which means itâ€™s a sequence of characters enclosed in quotation marks.

Step 3: Saving Your Program

Save the file with a .py extension, which stands for Python. You can name the file anything you want, but for this example, letâ€™s call it hello_world.py.

Be sure to remember the location where you saved this file.

Step 4: Running Your Program

Now, itâ€™s time to run your program and see the output! The process is slightly different for Windows and Mac:

On Windows:

Open the Command Prompt (you can do this by typing â€œCommand Promptâ€ into the search bar and pressing Enter).

Use the cd command to navigate to the folder where you saved your hello_world.py file. For example, if you saved it in a folder named PythonPrograms on your C: drive, you would type cd C:\PythonPrograms and press Enter.

Once youâ€™re in the correct folder, type python hello_world.py and press Enter.

On Mac

Open Terminal (click on the magnifying glass in the top right corner for Spotlight Search, type Terminal, and press Enter).

Use the cd command to navigate to the folder where you saved your hello_world.py file. For example, if you saved it in a folder named PythonPrograms in your Documents, you would type cd Documents/PythonPrograms and press Enter.

Once youâ€™re in the correct folder, type python3 hello_world.py and press Enter.

Regardless of whether youâ€™re on Windows or Mac, you should now see Hello, World! printed in your terminal or command prompt. Congratulations! Youâ€™ve just written and run your first Python program.

In the upcoming tutorials, we will explore more Python basics like variables, data types, conditionals, loops, and functions, all important building blocks of a Python program. Happy Coding!

**NEXT TUTORIALS**

Tutorial 1 â€“ Installing PythonTutorial 2 â€“ Your First Program in PythonTutorial 3 â€“â Python VariablesTutorial 4 â€“â Python Data TypesTutorial 5 â€“â Python ConditionalsTutorial 6 â€“â Python LoopsTutorial 7 â€“â Python FunctionsTutorial 8 â€“â Advanced Python FunctionsTutorial 9 â€“â Starter Machine Learning Python ProgramTutorial 10 â€“ Your First Machine Learning Program!

Explore more with this book.

Python for Kids, 2nd Edition: A Playful Introduction to Programming$21.99Buy NowWe earn a commission if you make a purchase, at no additional cost to you.02/18/2024 05:01 am GMT **References**

Briggs, Jason R. Python for Kids: A Playful Introduction To Programming. No Starch Press, 2012.

Share this:Click to share on Twitter (Opens in new window)Click to share on Facebook (Opens in new window)Click to share on LinkedIn (Opens in new window)Click to share on Reddit (Opens in new window)Click to share on Pinterest (Opens in new window)Click to share on Pocket (Opens in new window)x
24/07/2024 22:21 – Machine Learning For Kids: Python Variables
#VariableComputerScience #StringComputerScience #PythonProgrammingLanguage #SoftwareDevelopment #ComputerProgramming #SoftwareEngineering #Computing #ComputerScience #Technology #ProgrammingConstructs #ComputerEngineering #ProgrammingInterfaces #Computers #FunctionComputerProgramming #ProgrammingLanguage #DataType #MachineLearning #Data #Learning #ComputerData #MathematicalLogic #Banana #TheoreticalComputerScience #DataTypes #AppliedMathematics #Information

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
by Sanksshep MahendraJuly 13, 2023, 1:03 pm

**Introduction**

Machine learning for kids: Welcome to this tutorial on Python Variables! Variables are the backbone of any programming language, serving as containers where we store our data. Theyâ€™re like little boxes where we can put things for use later.

**Table Of Contents**

**Who Is This For?**

Grade: 6th to 10th

This tutorial is designed for students in grades 6 through 10. Whether youâ€™re completely new to Python or have dabbled a bit in programming, this tutorial will help you understand the fundamental concept of Python variables.

**What Will We Learn?**

In this journey of â€œMachine Learning for Kidsâ€ lets learn the basics of python which is the foundation of machine learning.â In this tutorial, weâ€™ll dive into what variables in Python are, how to create them, and how we can use them. We will discuss the different types of data we can store in variables and illustrate this with easy-to-understand examples.

Also Read: How Long Does It Take To Learn Python

**Understanding Python Variables**

In Python, a variable is created when you assign a value to it. For example, x = 5 creates a variable named x and stores the number 5 in it. Python variables donâ€™t require explicit declaration to reserve memory space. The declaration happens automatically when you assign a value to a variable.

You can think of variables as labels for containers that hold information or data. The data stored in a variable, known as its value, can be of various types like numbers, text (known as strings in Python), lists, and more.

Letâ€™s illustrate this with some code:

# Creating variables

name = "John"

age = 12

fruits = ["apples", "bananas", "cherries"]

# Printing variables

print(name)

print(age)

print(fruits)Hereâ€™s whatâ€™s happening in the code:

name = "John": Here, weâ€™re creating a variable called name and storing the string â€œJohnâ€ in it.

age = 12: This creates a variable called age and stores the number 12 in it.

fruits = ["apples", "bananas", "cherries"]: This creates a variable called fruits and stores a list of strings in it.

print is a function that displays the value of a variable to the console. When we print our variables, we see the values that are stored in them:

John

12

['apples', 'bananas', 'cherries']As you can see, Python variables are flexible and allow us to store various types of data, including numbers, strings, and lists. In subsequent tutorials, weâ€™ll explore more about different data types and how we can operate on them in Python. Happy coding!

**NEXT TUTORIALS**

Tutorial 1 â€“ Installing PythonTutorial 2 â€“ Your First Program in PythonTutorial 3 â€“ Python VariablesTutorial 4 â€“â Python Data TypesTutorial 5 â€“â Python ConditionalsTutorial 6 â€“â Python LoopsTutorial 7 â€“â Python FunctionsTutorial 8 â€“â Advanced Python FunctionsTutorial 9 â€“â Starter Machine Learning Python ProgramTutorial 10 â€“â Your First Machine Learning Program!

Explore more with thisâ book.

Python for Kids, 2nd Edition: A Playful Introduction to Programming$21.99Buy NowWe earn a commission if you make a purchase, at no additional cost to you.02/18/2024 05:01 am GMT **References**

Briggs, Jason R. Python for Kids: A Playful Introduction To Programming. No Starch Press, 2012.

Share this:Click to share on Twitter (Opens in new window)Click to share on Facebook (Opens in new window)Click to share on LinkedIn (Opens in new window)Click to share on Reddit (Opens in new window)Click to share on Pinterest (Opens in new window)Click to share on Pocket (Opens in new window)x
24/07/2024 22:19 – Machine Learning For Kids: Python Data Types
#DataType #BooleanDataType #SoftwareEngineering #ComputerEngineering #Technology #ComputerData #DataTypes #ComputerScience #MathematicalLogic #ProgrammingConstructs #Computing #ComputerProgramming #SoftwareDevelopment #IntegerComputerScience #Computers #LogicInComputerScience #TypeTheory #PythonProgrammingLanguage #Mathematics #StringComputerScience #MathematicalStructures #ProgrammingInterfaces #TheoreticalComputerScience #TruthValue #Arithmetic #AppliedMathematics #Data #FloatingPointArithmetic #FormalismDeductive #TheoryOfComputation

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
by Sanksshep MahendraJuly 13, 2023, 1:28 pm

**Introduction**

Machine Learning for kids: Welcome to this tutorial on Python Data Types! Just like we have various types of data in real life such as numbers, text, lists of items, etc., Python also has several different types of data that it can recognize and manipulate. Understanding data types is a fundamental concept in programming.

**Table Of Contents**

**Who Is This For?**

Grade: 6th to 10th

This tutorial is designed for young learners in grades 6 through 10. If youâ€™ve been learning Python and are familiar with the concept of variables, then youâ€™re ready to delve into the diverse world of Python data types!

**What Will We Learn?**

In this journey of â€œMachine Learning for Kidsâ€ lets learn the basics of python which is the foundation of machine learning. In this tutorial, we will learn about the various types of data we can use in Python. These include numbers (both integers and decimals), text (called â€œstringsâ€ in Python), lists, and more. We will also learn how to determine the type of data stored in a variable.

**Understanding Python Data Types**

Python recognizes several different types of data:

Numbers: These include integers (whole numbers) and floats (decimal numbers). For example, 5 is an integer and 7.3 is a float.

Strings: Strings are sequences of character data, essentially any text that you want to store. The string data type is represented by enclosing the text in quotes, like this: "Hello, World!".

Lists: A list in Python is a collection of items which can be of mixed types. Lists are enclosed in square brackets [...] and the items are separated by commas. For example, [1, "a", 4.5] is a list containing an integer, a string, and a float.

Booleans: A Boolean data type has one of two possible values, True or False, which are generally used to represent the truth values of logical expressions.

Here is an example that demonstrates each of these types:

# Number - integer

my_int = 7

print(my_int)

print(type(my_int))

# Number - float

my_float = 7.3

print(my_float)

print(type(my_float))

# String

my_string = "Hello, World!"

print(my_string)

print(type(my_string))

# List

my_list = [1, "a", 4.5]

print(my_list)

print(type(my_list))

# Boolean

my_bool = True

print(my_bool)

print(type(my_bool))In this code:

We assign different types of data to different variables.

We then use print() to print the value of each variable.

We also use the type() function, which tells us the data type of the variable.

The output of this code will be:

7

<class 'int'>

<class 'float'>

Hello, World!

<class 'str'>

[1, 'a', 4.5]

<class 'list'>

True

<class 'bool'>As you can see, each type of data is recognized differently by Python, as indicated by the type names (â€˜intâ€™, â€˜floatâ€™, â€˜strâ€™, â€˜listâ€™, and â€˜boolâ€™). Understanding these different data types and how they behave is crucial as you learn to program in Python. In the following tutorials, weâ€™ll learn more about how we can use and manipulate these data types. Happy coding!

**NEXT TUTORIALS**

Tutorial 1 â€“ Installing PythonTutorial 2 â€“ Your First Program in PythonTutorial 3 â€“ Python VariablesTutorial 4 â€“ Python Data TypesTutorial 5 â€“â Python ConditionalsTutorial 6 â€“â Python LoopsTutorial 7 â€“â Python FunctionsTutorial 8 â€“â Advanced Python FunctionsTutorial 9 â€“â Starter Machine Learning Python ProgramTutorial 10 â€“â Your First Machine Learning Program!

Explore more with this book.

Python for Kids, 2nd Edition: A Playful Introduction to Programming$21.99Buy NowWe earn a commission if you make a purchase, at no additional cost to you.02/18/2024 05:01 am GMT **References**

Briggs, Jason R. Python for Kids: A Playful Introduction To Programming. No Starch Press, 2012.

Share this:Click to share on Twitter (Opens in new window)Click to share on Facebook (Opens in new window)Click to share on LinkedIn (Opens in new window)Click to share on Reddit (Opens in new window)Click to share on Pinterest (Opens in new window)Click to share on Pocket (Opens in new window)x
24/07/2024 22:19 – Machine Learning For Kids: Python Conditionals
#ConditionalComputerProgramming #PythonProgrammingLanguage #ComputerProgramming #ProgrammingConstructs #Technology #SoftwareDevelopment #ProgrammingInterfaces #ComputerScience #Computing #SoftwareEngineering #ComputerEngineering #IndentationStyle #ControlFlow #ProgrammingParadigms #Computers #MathematicalLogic #Software #Grammar #SystemsEngineering

Extracted and hastagged by Infusio for preview. We recommend you read the original article instead...
by Sanksshep MahendraJuly 13, 2023, 2:13 pm

**Introduction**

Machine learning for kids: Welcome to this tutorial on Python Conditionals! In real life, we often make decisions based on certain conditions. For instance, â€œIf itâ€™s raining, then Iâ€™ll take an umbrellaâ€. Python conditionals allow us to create such decision-making scenarios within our code.

**Table Of Contents**

**Who Is This For?**

Grade: 6th to 10th.

This tutorial is suitable for students from grades 6 to 10 who have an understanding of Python basics such as variables and data types. If youâ€™re ready to give your Python programs the ability to make decisions, then youâ€™re at the right place!

**What Will We Learn?**

Machine learning for kids: In this tutorial, weâ€™ll explore Python conditionals â€“ the if, elif, and else statements. Weâ€™ll learn how to use these to control the flow of our programs based on certain conditions. We will also understand the importance of proper indentation in Python.

Python uses if, elif (stands for else if), and else statements to control program flow based on certain conditions. Each condition is followed by a colon and indented lines of code. If the condition is True, Python executes that code. If itâ€™s False, Python skips it and moves on to the next condition (if any). Hereâ€™s a basic structure of Python conditionals:

if condition1:

# code to execute if condition1 is True

elif condition2:

# code to execute if condition1 is False but condition2 is True

else:

# code to execute if both condition1 and condition2 are FalseLetâ€™s look at a Python program that decides what to wear based on the weather:

weather = "rainy"

if weather == "rainy":

print("Wear a raincoat!")

elif weather == "sunny":

print("Wear sunglasses!")

else:

print("Dress normally!")Hereâ€™s how the code works:

We first define a variable weather and assign the string â€œrainyâ€ to it.

The if statement checks if weather is equal to â€œrainyâ€. If this condition is true (which it is in this case), it prints â€œWear a raincoat!â€ and then skips the rest of the conditions.

If weather was not â€œrainyâ€, it would then check the elif condition to see if weather is â€œsunnyâ€. If so, it would print â€œWear sunglasses!â€.

Finally, if neither of the above conditions was met, the else statement would execute, printing â€œDress normally!â€.

The output of this program would be:

Wear a raincoat!Remember, the code inside each condition must be indented correctly. In Python, indentation is not just for readability; itâ€™s a part of the syntax. Using conditionals, you can make your programs smart enough to make decisions based on any set of conditions. Happy coding!

**NEXT TUTORIALS**

Tutorial 1 â€“â Installing PythonTutorial 2 â€“â Your First Program in PythonTutorial 3 â€“â Python VariablesTutorial 4 â€“ Python Data TypesTutorial 5 â€“ Python ConditionalsTutorial 6 â€“â Python LoopsTutorial 7 â€“â Python FunctionsTutorial 8 â€“â Advanced Python FunctionsTutorial 9 â€“â Starter Machine Learning Python ProgramTutorial 10 â€“â Your First Machine Learning Program!

Explore more with thisâ book.

Python for Kids, 2nd Edition: A Playful Introduction to Programming$21.99Buy NowWe earn a commission if you make a purchase, at no additional cost to you.02/18/2024 05:01 am GMT **References**

Briggs, Jason R.â Python for Kids: A Playful Introduction To Programming. No Starch Press, 2012.

Share this:Click to share on Twitter (Opens in new window)Click to share on Facebook (Opens in new window)Click to share on LinkedIn (Opens in new window)Click to share on Reddit (Opens in new window)Click to share on Pinterest (Opens in new window)Click to share on Pocket (Opens in new window)x

Previous – Next