Theo's claim that LLMs were 'discovered' by a long autocomplete run draws ridicule
nptacek · x · 2026-11-16
A small AI-circle feud: Theo claimed LLMs were discovered when an autocomplete training run went on too long — as if someone left autocomplete running over a weekend and came back to ChatGPT. Critics push back that LLMs exist because of the Transformer architecture, which nobody stumbled upon by accident, calling it a confused analogy to accidental discoveries like penicillin.
More from Fun
- Remember when AI looked like this and could do nothing else? — w1kke · 2026-11-16
- AI Twitter has more fake insiders than real ones, and nobody knows anything — VraserX · 2026-11-16
- Dev builds custom standalone Artificial Analysis alternative with Pareto frontiers and dark mode — teortaxesTex · 2026-11-16
- Gemini leaks its own internal instruction "do not use python" into a Google AI Overview — bclavie · 2026-11-16
- "No One Worry": AI Safety Anxiety Meets Deadpan Denial — iruletheworldmo · 2026-11-16
- Gary Marcus shuts down claim that GPT-6 could make GTA 6 in an afternoon — GaryMarcus · 2026-11-16