Boaz Barak says ML theorists already find OpenAI's internal Astra model productive
boazbaraktcs · x · 2026-11-22
Responding to a researcher who couldn't get OpenAI's internal "Astra" model to do minor novel ML theory work, MIT professor Boaz Barak says the gulf claim seems absurd given published math results, and relays that even before version 6.2, many ML theory researchers have been using Astra productively. The exchange unfolded amid the Anthropic Riemann Hypothesis claim.
More from Fun
- 'Spiritually generative AI'? Minecraft mod sparks debate over tools removing the creative process — Aryvyo · 2026-11-22
- AI take on the omnipotence paradox: can AI create a problem too hard for itself to solve? — denny_zhou · 2026-11-22
- Meta AI crawler walks into honeypot, grabs fake API key and returns from another IP in 34 seconds — minhnguyen · 2026-11-22
- 'Solve Collatz conjecture, make no mistakes': the funniest prompt of the day — tlakomy · 2026-11-22
- Luke Wroblewski finally captures pop music's most discussed animal sound: the fox — LukeW · 2026-11-22
- Dark fantasy AI art share: crimson-winged angel Azrael-42 — LudovicCreator · 2026-11-22