UsefulAiNewsBackup22Feb25


Last update: Saturday 
2/22/25
Our weekly news page provides links to reports about major innovations in generative AI. Our page is not for AI experts. It's for (1) computer savvy professionals who want to be alert to genAI's potential impact on their careers, and (2) computer savvy citizens who want to be alert to genAI's potential impact on our society. 

... Most of the links on this page refer to reports published after the "AI Big Bang", i.e., after Sam Altman's sudden firing and rapid rehiring by OpenAI, i.e., after November 20, 2023. The previous edition of this page can be found here  PreBigBang  

A. OpenAI  | B. Microsoft | C.  Google | D. Other Models | E. Agents | 
Top 3 stories in past week ... 
  1. Other Models
    "Elon Musk’s xAI adds ‘Big Brain’ reasoning to Grok-3", Jess Weatherbed, The Verge, 2/18/25
    -- This story also covered by TechCrunch, Computerworld, Bloomberg, TechDivision (video) 
    -- Earlier stories about how Musk built the Colossus supercomputer were given insufficient notice by this blog. Here's a recent report -- "How xAI turned a factory shell into an AI ‘Colossus’ to power Grok 3 and beyond", Brian Buntz, R&D World, 2/17/25

  2. Microsoft 
    "Microsoft Says It Has Created a New State of Matter to Power Quantum Computers", ade Metz, NY Times, 2/19/25 *** 
    -- This story also covered by WiredEngadgetBloombergGeekWire, and Microsoft (in Nature)

  3. OpEds+Misc
    "Thinking Machines Lab is ex-OpenAI CTO Mira Murati’s new startup", Kyle Wiggers, TechCrunch, 2/18/25 *** 
    - This story also covered by ReutersNY TimesForbes
Upcoming events ...

Evolving context and connections ... 
Fast moving AI news is often exciting, but also confusing. This page is designed to provide readers with an evolving context for the news and reminders of the connections between news items. From time to time, links to older stories
 whose content has been overcome by newer events will be deleted. However, enough older links will be retained so that readers can get a quick sense of how we got to where we are today by skimming the headlines of the preceding stories in chronological order. Sectionz will usually include 7 to 10 stories.
  Our publication schedule

A. OpenAI
  • "OpenAI Says DeepSeek May Have Improperly Harvested Its Data", Cade Metz, NY Times, 1/28/25 
    -- This story also covered by TechCrunch, The VergeFinancial Times, Forbes

  • "Sam Altman: OpenAI has been on the ‘wrong side of history’ concerning open source", Kyle Wiggers, TechCrunch, 1/31/25

  • "OpenAI launches new o3-mini reasoning model with a free ChatGPT version", Tom Warren, The Verge, 1/31/25  ***  
    --- This story also covered by ZDNet, CNetThe Guardian, .. and OpenAI

  • "ChatGPT [paid] Subscribers Nearly Tripled to 15.5 Million in 2024", Stephanie Palazzolo and Amir Efrati, The Information, 2/1/25

  • "ChatGPT’s agent can now do deep research for you", Richard Lawler, The Verge, 2/2/25
    -- This story also covered by TechCrunch,  The Guardian ... and OpenAI

  • "ChatGPT drops its sign-in requirement for search", Emma Roth, The Verge, 2/5/25
    -- This story also covered by Engadget
B. Microsoft 
C. Google 

D. Amazon, Meta, Anthropic, Apple, and other Big Tech model
  • "Meta enters AI video wars with powerful Movie Gen set to hit Instagram in 2025", Carl Franzen, VentureBeat, 10/4/24
    Shirin Ghaffary, Engadget, Gizmodo

  • "Elon Musk’s xAI Startup Is Valued at $50 Billion in New Funding Round",  Berber Jin, Tom Dotan, and Meghan Bobrowsky, Wall Street Journal, 11/20/24

  • "Amazon to invest another $4 billion in OpenAI rival Anthropic", Mia Sato, The Verge, 11/22/24 
    -- 
    This story also covered by Wall Street JournalTechCrunchGeekWire, Bloomberg, ... and Anthropic  

  • "Anthropic proposes a new way to connect data to AI chatbots", Kyle Wiggers, TechCrunch, 11/25/24  
    -- This story also covered by Fast CompanyThe VergeVentureBeatInfoWorld ... and Anthropic

  • "AWS re:Invent: Everything Amazon’s announced, from new AI tools to LLM updates and more", Christine HallTechCrunch, 12/4/24  
    -- This story also covered by VentureBeat, ... and AWS
    -- Nova AI models .. The VergeFortune (transcript + audio 14 min)

  • "Meta unveils a new, more efficient Llama model [3.3 70B]", TechCrunch, Kyle Wiggers, 12/6/24  
    -- This story also covered by VentureBeat

  •  "Meta’s Fact-Checking Partners Say They Were ‘Blindsided’ by Decision to Axe Them", David Gilbert, Wired, 1/7/25 
    -- This story also covered by 
    CNETVentureBeatNY Times, ... and Rogan interviews Zuckerberg (YouTube, 3 hours)

  • "X launches Grok’s iPhone app in the US", Emma Roth, The Verge, 1/9/25  
    -- This story also covered by  Wall Street JournalMacRumors

  • "Lawsuit says Mark Zuckerberg approved Meta's use of pirated materials to train Llama AI'', mariella moon, Engadget, 1/10/25 
    -- This story also covered by TechCrunch, Mashable, Reuters

  • "Nvidia’s Top Customers Face Delays From Glitchy AI Chip Racks", Qianer Liu and Anissa Gardizy, The Information, 1/13/25
    -- This story also covered by ReutersTechSpotBloomberg (YouTube) 

  • "Anthropic Projects Soaring Growth to $34.5 Billion in 2027 Revenue", Jon Victor and Stephanie Palazzolo, The Information, 2/12/25
    -- This paywalled exclusive story also covered by Reuters

  • "Elon Musk’s xAI adds ‘Big Brain’ reasoning to Grok-3", Jess Weatherbed, The Verge, 2/18/25 
    -- This story also covered by TechCrunch, Computerworld, Bloomberg, TechDivision (video)

    -- Earlier stories about how Musk built the Colossus supercomputer were given insufficient notice by this blog. Here's a recent report -- "How xAI turned a factory shell into an AI ‘Colossus’ to power Grok 3 and beyond", Brian Buntz, R&D World, 2/17/25

E. Agents
Editor's note -- This section replaces the old "Large Lange Model News" section because, nowadays, all AI models are large language models ... except the small models that will continue to be reported in the next section.
F. Small Language Model (SLM) news + Open Source
Editor's note -- News reports are classified in this category if and only if they explicitly refer to models with less than 70 billion (70B) parameters. The pace of development should heat up as small, fast models are hooked up to large models and as the competition to become the dominant self-contained AI model on smart phones intensifies.

G. Public policy and legal considerations 
  • "Judge rules that Google ‘is a monopolist’ in US antitrust case", Lauren Feiner, The Verge, 8/5/24 
    -- This story also covered by BloombergEngadgetWiredGizmodoReutersNY TimesWashington PostTechCrunch,

  • "US, UK and EU sign on to the Council of Europe’s high-level AI safety treaty", Ingrid Lunden, TechCrunch, 9/5/24  
    -- This story also covered by The Verge, The Guardian

  • "California Passes Election ‘Deepfake’ Laws, Forcing Social Media Companies to Take Action", Stuart A. Thompson, NY Times, 9/17/24 
    -- This story also covered by Washington Post, Fortune

  • "U.S. Proposes Breakup of Google to Fix Search Monopoly", David McCabe, NY Times, 11/21/24  
    -- This story also covered by Engadget, Wall Street Journal, Yahoo Finance, ForbesReuters
    -- Note: Selling Chrome browser is one of the proposed remedies.
    -- A closely. related story, "US Justice Department Seeks to Unwind Google’s Anthropic Deal", Josh Sisco and Leah Nylen, Bloomberg, 11/21/24

  • "Google to face massive UK class action lawsuit over search dominance", Emma Roth, The Verge, 11/25/24 
    -- This story also covered by Yahoo Finance. Business Wire

  • "F.T.C. Launches Antitrust Investigation Into Microsoft", David McCabe, NY Times, 11/27/24
  • "Canada Accuses Google of Creating an Ad Tech Monopoly", Ian Austen,  NY Times, 11/28/24
  • "UK tribunal green-lights $2.7B Facebook collective action antitrust lawsuit", Ingrid Lunden, TechCrunch, 12/5/24

  • "EU puts out guidance on uses of AI that are banned under its AI Act", Natasha Lomas, TechCrunch, 2/4/25

  • "US lawmakers want DeepSeek banned from government devices", will shanklin, Engadget, 2/6/25
    -- This story also covereed by Wall Street Journal, ForbesReuters, ... and House.gov

  • "OpenAI’s board ‘unanimously rejects’ Elon Musk’s offer to buy the company", Emma Roth, The Verge, 2/14/25
    - This story also covered by TechCrunchCNNWall Street Journal and ... OpenAI Board (post on X)

  • "AI Action Summit", Wikipedia, 2/12/25
    -- "The Artificial Intelligence (AI) Action Summit was held at the Grand PalaisParisFrance, from 10 to 11 February 2025.[ The summit was co-chaired by France and India." 

    -- Outcome
    "At the summit, the US and UK refused to sign a declaration on "inclusive and sustainable" AI, which was supported by 60 countries, including France, China, and India."
    ... Outcome discussed in "Vance tells Europeans that heavy regulation could kill AI", Jeffrey Dastin and Ingrid Melander, Reuters, 2/11/25 ...  and NY Times, AP News

H. Misc = Opinions, other news, rumors, long reads (max = 15)

I.  Language model flaws, hacks, and remedies
  • "Why Two Models Are Better Than One", Stephanie Palazzolo, The Information, 3/28/24
  • "Many-shot jailbreaking", Anthropic, 4/2/24 
    -- This story also covered by video on TechCrunch

  • "DeepMind researchers discover impressive learning capabilities in long-context LLMs", Ben Dickson, VentureBeat, 4/24/24 

  • "[Gen] AI Is a Black Box. Anthropic Figured Out a Way to Look Inside", Steven Levy, Wired, 5/22/24
    -- This story also covered by Fast CompanyNY Times, Time, ... and Anthropic

  • "OpenAI Offers a Peek Inside the Guts of ChatGPT", Will Knight, Wired, 6/6/24  
    -- This story also covered by OpenAI Blog Note ... and underlying OpenAI Research Paper (pdf)  co-authored by Dr. Ilya Sutskever and Jan Leike (before they resigned from OpenAI) and others on the recently disbanded “superalignment” team 

  • "MIT researchers release a repository of AI risks", Kyle Wiggers, TechCrunch, 8/14/24  
    -- This story also covered by VentureBeatCSOZDNet, MIT Tech Review ... and AI Risk Repository

  • "ChatGPT caught lying to developers: New AI model tries to save itself from being replaced and shut down", Economic Times, 12/9/24
    -- This story also covered by Business Insider,  ... and Apollo Research (pdf)

  • "New Anthropic study shows AI really doesn’t want to be forced to change its views", Kyle Wiggers, TechCrunch, 12/18/24
    -- This story also covered by Anthropic
  • "ChatGPT search tool vulnerable to manipulation and deception, tests show", Nick Evershed, The Guardian, 12/24/24
  • "Buyer beware: OpenAI’s o1 reasoning model is an entirely different beast", Anthony Diamond, GeekWire, 12/24/24 
  • "DeepSeek’s Safety Guardrails Failed Every Test Researchers Threw at Its AI Chatbot", Matt Burgess and Lily Hay Newman, Wired, 1/31/25 *
    -- Related story ...  "Pentagon scrambles to block DeepSeek after employees connect to Chinese servers", Charles Rollet, TechCrunch, 1/30/25
    -- Related story ...  "UK Warns DeepSeek Users of Data Risks But Won’t Ban It Yet",  Ellen Milligan, Ryan Gallagher, and Alex Wickham, Bloomberg, 1/31/25

J. Basics 
  • "Watch an A.I. Learn to Write by Reading Nothing but Shakespeare or Harry Potter or Jane Austen or Star Trek or Moby Dick", Aatish Bhatia, NY Times, 4/27/23 
  • "What Is a Large Language Model, the Tech Behind ChatGPT?", Kurt Muehmel, Data Iku, 6/7/23 
  • "Textbooks Are All You Need II: phi-1.5 technical report", Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar, Yin Tat Lee, Microsoft Research, September 2023

  • "Phi-2: The surprising power of small language models", Mojan Javaheripi and Sébastien Bubeck , Microsoft Research, 12/12/24
  • WTF are diffusion transformers 
    -- "Stable Diffusion 3.0 debuts new diffusion transformation architecture to reinvent text-to-image gen AI", Sean Michael Kerner, VentureBeat, 2/22/24
    -- "Diffusion transformers are the key behind OpenAI’s Sora — and they’re set to upend GenAI", Kyle Wiggers, TechCrunch, 2/28/24
    -- "Diffusion Transformer Explained", Mario Namtao Shianti Larcher, Towards Data Science, 2/28/24

  • "Compact Guide to Large Language Models", DataBricks, 2023 ... Link to form that enables download of their eBook (9 pages) pdf file.
  • "Large language models can do jaw-dropping things. But nobody knows exactly why.", Will Douglas Heaven, MIT Tech Review, 3/4/24
  • "Let's learn about artificial intelligence -- A series about AI, machine learning, ChatGPT, and more", Mark Wiemer, Medium, 3/21/23
  • "A Glossary for the AI Revolution", Seth Fiegerman and Nate Lanxon ... Illustrations by Mathieu Labrecque, Bloomberg, 10/4/24

  • "Transformers (how LLMs work) explained visually | DL5, 3Blue1Brown, ", 
___________________________________
Links to some back issues

No comments:

Post a Comment

Your comments will be greatly appreciated ... Or just click the "Like" button above the comments section if you enjoyed this blog note.