News for hackers. Stuff that matters.

A comment on Microsoft Researcher John Langford Exits. AI Frontiers Group Declared Finished.

Re:Microsoft Researcher John Langford Exits. AI Frontiers Gr... by iagiagiag on Saturday November 07, 2026 @05:33PM (#50408676)
It makes me sad because there are so many unexplored areas of AI research, even on text generation. Billions are invested in almost one idea, which very realistically isn't the most optimal way to do text generation, but a lot of companies are afraid of taking risks. I still wouldn't put $1b on JEPA but I prefer a world where we can try things like that.
[ Parent | on HN ]
Re:Microsoft Researcher John Langford Exits. AI Frontiers Gr... by alansaber on Saturday November 07, 2026 @05:58PM (#50408842)
LLMs have very favourable exploration<>consolidation dynamics. An LLM alternative would need to display fundamental improvements at relatively large scales which is extremely challenging for new research.
[ Parent | on HN ]
Re:Microsoft Researcher John Langford Exits. AI Frontiers Gr... by sidibe on Saturday November 07, 2026 @10:19PM (#50410145)
It's a lot easier to explore these other areas with the help of LLMs.
[ Parent | on HN ]