The deepseek team is obviously really good. China is full of talented engineers. Every other take is cope. Sorry.

Sahil Dumb comment that has cost people a better life

Separate from people who are easily offended by offending them early.

new from me: what went wrong with the Alan Turing Institute? I dive into the failures in institutional design, focus, and leadership that led the UK's national AI institute to become irrelevant. It's a story of drift, stakeholderism, and wasted potential. 🧵

unbelievable that he's not more well-known

I have noticed more excitement about AI tools from technical generalists filling in knowledge gaps vs hyper-specialists that don't feel the tools are good enough to help them much within their zone of expertise.

🤯

ChatGPT and others: 2022: Cool toy! Fun to play with. 2023: Don't use it much. 2024: Use it a few times a week. 2025: Use it ~ every hour. Anyone else like this?

From the 1973 Lighthill report on AI.

kids, always remember that money has log(money) utility, not linear utility. and almost all careers offer [10^n, 10^n+1] money, but rarely exceed some bound. ex: quant may be the best way to make 10 million, but the worst way to make 100.

unreal

"actually it's all just pattern matching!" he screams as his peers develop intelligence capable of finding security vulnerabilities in the linux kernel

when you hear someone say: “Oh, [I/my company] has been using ai for *decades*” that is a reliable signal that you’re about to hear some terrible opinions about LLMs and the dynamics of modern ai ai is a newcomer’s game.

Don't leave AI to the STEM folks. They are often far worse at getting AI to do stuff than those with a liberal arts or social science bent. LLMs are built from the vast corpus human expression, and knowing the history & obscure corners of human works lets you do far more with AI

If your boss asks you to setup your workflow as an “RL environment with a verifiable reward” start polishing your resume.