🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

I think the golden path to scale with AI is with both great products AND better models. Elegant, easy to use yet powerful, and affordable products always are far more likely to win. Even better if they create a moated ecosystem and/or have strong network effects. iPod, iPhone,

Once again, Andrej Karpathy gives us a clear vision of the state of the art in AI

More brilliant advice from the equality brilliant Gokul Rajaram. Gokul was in my interview loop at Facebook. My wide ranging discussion with him was one of the reasons why I was so eager to work there. A rare mix of deep thinking, customer-centricity, modesty, and generosity.

Woke up to a new Lime in Paris! Looks good, seems more vandalism-resistant, has less power (longer range?), but small wheels on bumpy pavement aren’t great. e-biking is easily the best way to get around central Paris…

Diplo on the wheels of steel! I love the Paris tradition of Fete da la Musique. People playing, listening, and dancing to music in the streets and outside many of the city’s iconic buildings. This year I caught two diplo shows. I’ve seen him many times since a gig he played

Bill Gurley is the best applied economist I have ever met and he was a brilliant VC. Certainly Bill (and his Benchmark partner Peter Fenton) asked me the most probing and insightful questions I ever heard when I was raising money for my last start-up. This conversation summary

A “Matryoshka-style” LLM nests smaller models inside larger ones—like Russian dolls—so you can dial up or down reasoning power at runtime. One consistent model, with many sizes / complexities depending on the use case. Scales from on-device to edge to cloud. Maybe can also be

Had dinner with a Nobel-worthy former roommate from Stanford University. He said that the Big Beautiful Bill has so many anti-science and anti-research provisions AND Europe is so under-invested in bold science that many of the best scientists in America and Europe that he knows are

🧠 Symbolique, générative, multimodale, agentique… Comment s’y retrouver parmi toutes les formes d’#IA ? @Orange vous guide avec ce lexique clair et complet pour mieux comprendre l’IA d’aujourd’hui et de demain 👇oran.ge/4l1JuML

Another leap forward for Orange AI with our brilliant CEO Christel Heydemann and CTIO Bruno Zerbib sponsoring a company-wide Boost Camp two-day training. We now have over 75,000 employees using our internal AI toolbox Live Intelligence. With that and over 200 other AI use cases in

Good progress using an ‘ensemble’ of multiple LLMs on more complex and novel tasks as measured by the ARC AGI test. Congrats to Sakana AI in Japan. As has been suggested by Yann LeCun and François Chollet and others, I suspect that LLMs alone are not the full solution to reach

Michael Levin As AI advances, our contribution is more and more original knowledge - meaning something that can’t be inferred from what exists digitally already by reasoning. Something like the result of an experiment. Maybe it should be written more natively for AIs instead of people, eg PDF

Thankful today for the generous giants of software engineering. We all stand on their shoulders and I continued to be inspired by them. 🙏 0000 Lady Ada Lovelace 0001 Bill Atkinson Bill Atkinson 0010 Andy Hertzfeld Andy Hertzfeld 0011 Dennis Ritchie 0100 John Carmack

Scaling up RL is all the rage right now, I had a chat with a friend about it yesterday. I'm fairly certain RL will continue to yield more intermediate gains, but I also don't expect it to be the full story. RL is basically "hey this happened to go well (/poorly), let me slightly

Hosted a dinner with about 15 IT leaders around the future of AI Agents in the enterprise. Here are a few updates on the state of the world at least at some large enterprises: * For many organizations, the demand from the business for AI is continuing outstrip the ability to