AI Researchers Say Today's Systems Are Already AGI as They Pivot to Books and New Ventures
On September 6, AI researcher @avtim and Calvin McCarter discussed on social media how AI is disrupting traditional research, drawing significant attention from the community. @avtim stated flatly that he believes today's systems already constitute AGI and that no capability remains locked away from further scaling—implying there is nothing left to do in traditional algorithm research.
Confirmed
- @avtim explained that his research centered on uncovering the principles needed to develop new algorithmic capabilities, and with AGI's arrival there is nothing left to do on that front—so writing a book excites him more than doing research.
- He believes the core ideas of traditional ML research remain good abstractions that can at least help us understand why LLMs work. He also admitted he has been captivated by these ideas for a decade, and "most likely someone else would be too"—even if that person may not exist, as long as the possibility exists, he feels obligated to try.
- Calvin McCarter responded that Claude Code has "had exactly the same impact on my field." His response was to leave his job, devote his time to finishing a book he believes will hold long-term value, and pursue a long-term venture that only became possible because of AI.
Why it matters
- This is a candid statement from a veteran researcher that "AGI has arrived," backed by concrete career moves (resigning, writing a book, starting over), showing how AI coding tools are reshaping career plans for researchers in traditional fields.
- The claim that "old theories can explain why LLMs work" offers a notable perspective on bridging classical ML thinking with deep learning.
2026-09-06 ~ 2026-09-06 · 5 related posts
Primary sources
- After Claude Code upended his field, this developer pivoted to writing a book and a new AI-era venture — avt_im · 2026-09-06
- [source] Researcher says Claude Code disrupted his field, quit to finish a book — CalvinMccarter · 2026-09-06
- [source] Researcher argues today's systems are already AGI and scaling unlocks the rest — avt_im · 2026-09-06
- Researcher on writing a book: old ML theory may explain why LLMs work — avt_im · 2026-09-06
- Researcher: writing a book beats research post-AGI — 'nothing left to invent' — avt_im · 2026-09-06