Original · 1957561075744010253 · Andrej Karpathy @karpathy

Ok, spent an ~hour sifting through submissions. The biggest challenge was the spam, as majority of replies are people linking to their own existing projects, not things made for the challenge. Of the ones that were:

Winner: I most enjoyed this one from @uncertainsys - OmegaQuest. He's solving Humanity's Last Exam problems with heavy AI use in the loop on video. Actually I really identified with the long pauses and general confusion in trying to use the current state of the art systems in learning something hard and new, where they are simultaneously so tantalizingly helpful at the margins, but still really poor overall, compared to an imagined human expert tutor. The "explanations" are... not. But I love the tenacity on display in working out something hard and seeing how far you can get with AI. A good reminder how it's better than what was, but also so far from what could be.
https://x.com/uncertainsys/status/1955026896795697428

Shoutout to @measure_plan for cool "visual vibe coding" projects, e.g. new musical instruments
https://x.com/measure_plan/status/1957116229379867125

A few of people commented that the challenge shouldn't have to be only for projects uniquely made for the challenge. If that were the case then shoutout to @evanliin et al. who linked to tinytpu, i really like the animated diagram, i haven't seen that before.
https://x.com/evanliin/status/1957496148874392012

Shoutout to @chrislakin for partially incepting the experiment a while ago with
https://x.com/ChrisChipMonk/status/1952079056666792340
but I basically come out agreeing with @nearcyan in his earlier comment
https://x.com/nearcyan/status/1938397408498737229 , maybe even $5K isn't :)

RT @karpathy: Ok, spent an ~hour sifting through submissions. The biggest challenge was the spam, as majority of replies are people linking…