Spoke with another engineer friend work in AV company at dinner, he was kind of surprised I had to drive my own software build. “You don’t test your software?! 😱”, I asked. “We do, but engineers never drove it. Engineers never touched the vehicles. The QA department tested

Many have asked what I meant when I said “Because we know what’s at stake.” The future… of spaceflight, of humanity, of everything beyond Earth. But on a personal level, it’s also the years of effort, the long nights, the sacrifices, and the belief that this work matters. Every

This only happens to you once.😃

Maybe pre-training is all you need. Maybe all the test time compute stuff should eventually all be learned.

Splashdown of Dragon confirmed – welcome back to Earth, Nick, Suni, Butch, and Aleks!

- Steve Jobs

next pixel prediction is in

This theme is really fascinating to me rn

infinite token prediction == consciousness?

are you an intellectually lazy, low conviction person anon

A major mistake I made in my undergrad is that I focused way too much on mathematical lens of computing - computability, decidability, asymptotic complexity etc. And too little on physical lens - energy/heat of state change, data locality, parallelism, computer architecture. The

Excited to host our first-ever AI Startup School in San Francisco next month! We're bringing together an incredible speaker lineup of the world's top AI founders and experts: Sam Altman · François Chollet · Chelsea Finn · John Jumper · Jared Kaplan · Andrej Karpathy · Fei-Fei Li · Elon Musk

Biology is by far the most advanced technology on the planet, self-assembling nanobots that can unpack a single cell into a space-faring civilization in a few billion years.

Why did Heisenberg get pulled over by the police? Officer: "Do you know how fast you were going?" Heisenberg: "No, but I know exactly where I am!"

Beff – e/acc Anything that is consistent with physics, and not mass or energy constrained by Earth, can be made

a mathematical habit I find useful for product thinking: in the limit, what does ... look like? for example, what will human-AI interaction look like in due time? then walk backwards from the limit and ask: how fast does it converge? how fast can you make it converge?

I am still struggling with the implications of Optimus watching it on YouTube, learning how to do it with RL in simulation and nailing it zero-shot. This is the most important AI breakthrough of the year, I’d say. Implications go well beyond humanoid robots.