Vincent Conitzer shows what GPT-6 Astra can't do, pushing back on loose AGI labels
conitzer · x · 2026-09-07
Duke researcher Vincent Conitzer, on his Substack "Funny AI fails," demonstrates something GPT-6 Astra cannot do — a question involving human body knowledge — arguing the "AGI" label is thrown around without definition. He adds that watching humans answer the same question (correctly) is probably even funnier.
More from Fun
- Viral post flagged as AI-written reignites debate over AI-polished prose — dbasch · 2026-09-07
- Simulated fruit fly brain gets to "play" Bad Apple in viral demo — SALM0N_SLD · 2026-09-07
- Polymarket puts just 10% odds on US enacting an AI safety bill before 2027 — Polymarket · 2026-09-07
- Kids lying about finished chores remind this parent of her AI agents — RachelVT42 · 2026-09-07
- Meme: 'You're absolutely right' — the AI sycophancy catchphrase — Malor777 · 2026-09-07
- Meme: Astra treats its token budget as a personal challenge — AlleyKatPr0 · 2026-09-07