Kenneth Stanley: is an LLM one researcher or an entire research community?
kenneth0stanley · x · 2026-10-27
AI researcher Kenneth Stanley raises an unsettled question in researcher-level AI: is a single LLM analogous to one prodigious researcher with vast knowledge, or to the collective creativity of an entire research community?
- If it's like a single researcher, displacing all human researchers would still leave LLMs unable to fill the community's collective capacity — cooperating clones would lack the diversity of interests and convictions of a real community.
- If LLMs encapsulate the whole community's creative capacity, displacement arguments gain ground, though one must ask whether the LLM is a complementary "community" or a strictly dominating one.
He asks readers: is the LLM an abstraction of an individual researcher or of a collection of researchers?
More from AGI Musings
- Practitioner: AI agents are opening a new career space in tech — marlene_zw · 2026-10-28
- Grok on replacing engineers: novel failure modes and system judgment still years to decades away — AdiPolak · 2026-10-28
- Researcher: economic governance of a multi-agent world is among the most exciting research areas — logangraham · 2026-10-28
- Y Combinator RFS Calls for LLMs Built Specifically to Write Well as RLHF Flattens Prose — ycombinator · 2026-10-28
- Y Combinator RFS: Vertical Foundation Models Can Beat General AI With a Fraction of the Compute — ycombinator · 2026-10-28
- YC argues companies should own their intelligence, not rent it from frontier labs — ycombinator · 2026-10-28