Baseten AI training head: the future is many specialized LLMs, not one frontier race
saranormous · x · 2026-09-04
Connor O'Neill, Head of AI Model Training at Baseten, argues the future is many specialized LLMs dedicated to specific tasks, with bigger frontier lab models deployed in areas like science and math.
His core point: measuring intelligence relatively (e.g. "open-source is 6 months behind, GLM is as good as Opus") is the wrong frame. Capabilities should be judged in absolute terms — for any task there's an intelligence threshold below which it can't be done, and above which returns diminish sharply. The game of the past five years has been pushing tasks past their respective thresholds.
More from AGI Musings
- Ex-OpenAI safety lead Miles Brundage: if your primary emotion on AI isn't concern, you're misreading it — Miles_Brundage · 2026-09-04
- Gary Marcus on GPT-6 Astra: symbolic world models are vindication, but no proof of AGI — GaryMarcus · 2026-09-04
- AI Job Market Talk: GenAI Engineers With 3-5 Years Experience Command ₹2-3 Lakh Monthly Pay — ashishllm · 2026-09-04
- Alignment researcher: defining the AI's value system isn't the real problem — Sauers_ · 2026-09-04
- Researcher: LLMs' hidden cost of wasting your time on useless work is underrated — lateinteraction · 2026-09-04
- Andrew Chen: agents as your C-suite works at work — what's the personal-life equivalent? — andrewchen · 2026-09-04