2026-01-22
Sadly, it's behind a paywall, so I won't be watching the whole thing, but the gist is that @MillionInt wishes there were more diversity in approaches to scaling ML models which is totally fair! Unless you're a daily power user of these models, it's hard to tell the difference
Core Memory
Q&A with recently departed OpenAI VP of Research Jerry Tworek, who claims OpenAI's shift toward more cautious ways made high-risk, pioneering work harder
Are all the AI superpowers playing it too safe? — ∙ Paid — On January 5th, famed AI researcher Jerry Tworek stunned world+dog …