Kudzo S Ahegbebu

@hyhieu226 @OpenAI @xai Take care Hieu! Wishing you the best

I have made the difficult decision to leave @OpenAI. Working here and at @xai before was a once-in-a-lifetime experience. I have met the best people. Not the best people in AI. Not the best people in tech. Simply the best people. At these companies, I have helped creating extremely intelligent entities that will meaningfully improve our lives. The work makes me proud. But the intensive work came with a price. I cannot believe I would say this one day, but I am burnt out. All the mental health deteriorating that I used to scoff at is real, miserable, scary, and dangerous. I am going to take a break from frontier AI labs, and will take my family to my home country Vietnam. There, I will try something new, and also search for a cure for my conditions. I hope I will heal. Until then.

@MillionInt sample efficient generalization is important for all the really interesting domains where we don't already have enough data though. unless you're saying that we'll get sample efficiency for free if we build better research process automation?

I continue to think that data efficiency is not the right perspective to focus on when thinking about deficiencies of todays models We already have all the data we need What we need to be more efficient with is human time and human attention needed to train the models

We should all be doing more art projects

@EGafni I don’t know if I actually believe this. Things like coordination, or agency most of the time seem more important . More intelligence is of course useful in most cases, but to first order so is more money. Instrumentality alone doesn’t make it a bottleneck.

The bottleneck to solving practically all problems is intelligence

@shayan_ Congrats Shayan, best of luck!

Career update: I left xAI to start something new, closing my 7+ year chapter working at Twitter, X, and xAI with so much gratitude. xAI is truly an extraordinary place. The team is incredibly hardcore and talented, shipping at a pace that shouldn’t be possible. From the Home Timeline at X to Grok 2, 3, and 4 at xAI, I worked across product infra and model behavior post-training, from memory to coding infra, agents, and more. Pure startup mode, every day. Working closely with Elon across X and xAI, I saw what happens when you refuse to accept impossible as an answer. I learned to embody obsessive attention to detail, maniacal urgency, and to think from first principles. I’m deeply grateful to @elonmusk for the experience, to @wanghaofei for the trust and support throughout Twitter/X, and to @TheGregYang and @ibab for believing in me. And to the many incredible people I had the privilege to work with along the way, thank you! Now, I’m excited to take the leap and build something new, focused on accelerating science. More soon.

@YuchenHe07 @ericzelikman @gharik @TheAndiPenguin @noahdgoodman Congrats @YuchenHe07 , super exciting!

@scikud @ericzelikman @gharik @TheAndiPenguin @noahdgoodman Thanks! 😊

finally announcing i’ve started humans& w/ amazing friends @gharik & @YuchenHe07 & @TheAndiPenguin & @noahdgoodman & many other world-class folks. we're optimists: it’s possible to rethink how we build ai, to empower people to accomplish more together tldr: love is all you need

These are even cooler in person! Robotics like most truly interesting things are really data limited and new techniques for these domains will be super important.

Interesting thing about this cold that’s going around is that my Apple Watch actually noticed a pretty sharp drop in my V02 max before I felt any symptoms

Who knew hanging out with 30,000 of your closest friends for a week would be a flu super spreader event?

@tszzl So much of it is also irrelevant. Academics seem obsessed with constructing inscrutable towers of complexity almost all of which are pointless. Things that last the test of time almost always end up being elegant in their simplicity.

neurips is a bit overwhelming and tiring. there are 20,000 attendees and hundreds of events mostly for recruiting. It’s strangely reminiscent of late stage college hackathon culture rather than Royal Society Meeting

@chenchen @gork The grow up so fast 🥹. Congrats Alex!

I joined xAI at 18 years old. Today, I'm moving onto new things. It's been an insane year. Leading personality & model behavior posttraining for Grok 4.1. Building @gork and memory. And the people were fucking incredible. I cannot wait to see the crazy future you'll build!

Totally agree. Again, achieving some janky form of AGI by individually tiling all domains with the RL cannon is bad and you should feel bad.

@Miles_Brundage You’re absolutely right!

Human sycophancy irritates me much more than AI sycophancy since (among other reasons) humans can choose their learning process + values more freely than AIs can

@davefontenot I’m going :)

who’s going to neurips?

@Miles_Brundage For some reason I’ve never had a thing for alternative history. Hits some uncanny valley thing for me

For All Mankind remains underwatched

@SarahChieng Yeah definitely!

@scikud SAY HI IN SD!

San Diego or Mexico City?

A common mistake that AI companies make nowadays is to not give their engineers enough time and mental calm to do their best work. Constant deadlines, pressure and distractions from daily AI news are poison for writing good code and systems that scale well. That’s why most AI APIs and products have reliability issues. A good company culture that mixes excellence with focus and enough rest leads to faster and better results. The best example of how to do it well is the early Google culture from 1998 which resulted in one of the largest scale and most reliable services on the web in just a few short years. Founders should copy some of the strategies that Larry and Sergey used. They are still underrated IMO despite their huge reputation.

@ericzelikman Best of luck Eric, gonna miss you!

This message is bittersweet. When I joined xAI, its impossibly ambitious mission drew me in. I also joined because of trust in Tony, a close mentor and friend. I knew it was where I could do and grow most. In retrospect, this was right: every year at xAI was incomparable to a year in the real world. I got to work with unforgettable people on everything from pretraining data to agent RL – I even led a small but amazing team, and I am proud of everyone and everything it accomplished. I really appreciated the chance to learn so much from Elon and the founding team, about maniacal urgency, about difficult decisions, about inspiring others, about finding unique advantages, about truth-seeking. I have faith xAI will continue to build not only great models but good models. Increasingly, I have felt it is urgent that we figure out how to build AI that deeply understands us, collaborates with us, and optimizes for our long term outcomes. This will require rethinking many approaches from first principles, but it can and must be done. I am now more sure than ever that xAI will accomplish its mission. For me, I am excited to take a leap of faith and build something new.