'Nowhere near AGI': koltregaskes calls year-end AGI claims gibberish, sees 1-2 years minimum
koltregaskes · x · 2026-10-06
koltregaskes re-upped his AGI skepticism, quoting ChrisGPT's question of whether Anthropic avoids AGI claims due to high standards or fear of looking like they've lost the frontier pace. From daily use of Dots he argues AGI remains far: the model still errs, forgets, loses context, gets distracted, hallucinates, and is slow and unreliable. He calls any lab claiming AGI by year-end 'talking absolute gibberish' and maintains his long-held timeline of at least one to two more years.
More from AGI Musings
- Yoav Goldberg: the current AI era truly began with the GPT-3 paper, and Alec Radford deserves the credit — yoavgo · 2026-10-06
- Pope insists algorithms lack humanity while Anthropic reportedly lobbies Vatican on AI consciousness — dotey · 2026-10-06
- Founder: the presentation deck is quietly disappearing in the AI era — aloncarmel · 2026-10-06
- Researcher Yoav Goldberg: OpenAI wouldn't have been founded without AlexNet — yoavgo · 2026-10-06
- LeCun amplifies NYT op-ed arguing six charts show just how much the economy needs AI — ylecun · 2026-10-06
- Exit interviews: why safety researchers quit OpenAI, Anthropic, and DeepMind — Tinac4 · 2026-10-06