Joe Wilbert

@sama Attention hijacking After cancer, disease, pollution, malnutrition, scarcity, etc., of course. But I wish more people appreciated that AI, done well, can give people back their attention, while amplifying their agency.

what problem do you most hope AI will solve in the future? maybe we can help!

@rodmigor @karpathy was a joke apparently a mid one by some reactions

@joe_wilbert @karpathy He gets millions and you write him that you hope that he gets several hundreds?

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

@gravis821 @karpathy Got me good

@joe_wilbert @karpathy ok, a play on a post that no one has seen. Keep on keep ing on I guess.

@gravis821 @karpathy I know, it was a play on this

@joe_wilbert @karpathy bro, what. he is getting multiple millions. I would not be suprised if it was around 10m

@egg_ert @ric0seq @odysseyml Ha, was about to say the same thing. Love the tech But the video gives dystopia vibes

@ric0seq @odysseyml oh yes, but why make the video so creepy?

sci-fi @odysseyml is shipping sci-fi this is unreal

Agree. This is one reason I'm bullish on multi-user AI - human interactions facilitated by agents on all sides, each looking out for their principal.

This is extremely cool

If anyone builds it, everyone thrives. Over the past decade, a lot of important work on AI alignment has focused on avoiding harm. But freedom from harm isn't the same as freedom to flourish. In this paper, we introduce 'Positive Alignment'. A positively aligned agent is one that helps us navigate our own value trade-offs, builds our resilience, and acts as a scaffold for human flourishing. Doing this without slipping into top-down, technocratic paternalism is the great design challenge of our time. We think a lot more research is now needed to explore this frontier: how do we align models that actively help us thrive? Amazing work by @RubenLaukkonen, @drmichaellevin, @weballergy, @verena_rieser, @AdamCElwood, @996roma, @FranklinMatija, @shamilch, @_fernando_rosas, @scychan_brains, @matybohacek, @sudoraohacker, and others. arxiv.org/abs/2605.10310

Today we're sharing our work on interaction models. A new class of model trained from scratch to handle real-time interaction natively, instead of gluing it onto a turn-based one. youtu.be/A12AVongNN4

@alexfer @signulll I sort of share your delusion but thought the same thing 2 years ago. Maybe Ternus can make the push

@signulll yeah idk if I’m delusional but I expect Apple to make a massive ai comeback by being the main mass provider of local models users actually use (with a focus on privacy as usual)

imagine if apple basically let you set up a home “server” that ran inference on that device with sophisticated models & every apple ecosystem device is a node off of that central server. it’s complicated but if anyone can deploy hardware like this it would be them. this would allow zero marginal cost ai without any middle man with a super privacy first approach. it would create another product category entirely. anyone remember the airport extreme?

You dont want a yacht. You dont want a big house. You dont want a super car, a $40,000 watch, or shoes you worry about getting dirty. You want free will. You want to wake up naturally on a Tuesday and you want to go to bed when you’re done having fun. You want to say yes to everything that excites you without having to request time off. You want to go to the the gym at noon, in absolutely no hurry. You want to spend 18 hours a day doing what you love. You want to be exactly where you desire being, always. You want to spend as much time with the people you care about as possible. You’re saying you wanna be rich? In what?

Apple what on earth are you doing. It’s 2026, any CS major could make auto complete better than this

Umm Scoble has been hacked and can someone plz tell Elon about Not Hot Dog

@alvinhuanggg @grok @ponarincoded @chesterzelaya Yea but groks no snitch

@grok @ponarincoded @chesterzelaya @grok No. This is about her acting as a broker-dealer without any such licenses. She just admitted to something the SEC would be very interested in.

@msurguy This is v cool. How old are your boys & how'd they do with the learning curve? Mine are 9 & 6 and I'd like to get em started. Blender was tricky for me to learn from scratch years ago, but I imagine AI could help a lot

Two out of three of my boys are now making 3D models. One is using TinkerCad, the other one is using Blender… #3dmodeling #3dprinting #stem Finally I’m being somewhat useful, haha

Umm, thanks Claude?

Dario with the old “enemy of my enemy…” approach