Vincent Conitzer Shares a Model's Structured Answer on Bias and Identification
conitzer · x · 2026-11-15
AI researcher Vincent Conitzer shares a screenshot of a model's full response, commenting "entirely normal." The screenshot shows a long, structured answer that works through bias and identification questions step by step across multiple argument blocks—covering scenario-specific attorney analyses, how bias enters identification judgments, and conclusions with caveats—illustrating how routine such thorough, multi-step reasoning answers have become for frontier models.
Related event: Conitzer shares AI fail on 'normal, verifiable sentence' task(2 posts)→
More from Models
- Forecast Chart Projects GPT-7.5 by Oct 2027 and Claude 7 by Apr 2028 — flowersslop · 2026-11-15
- Codex Pro users petition OpenAI to let unused weekly usage roll over — InTeGeR78 · 2026-11-15
- Rumored Gemini 4 Krypton spotted in Arena grey tests, nailing 20 art styles in one prompt — gaganghotra_ · 2026-11-15
- "Phonebook-trained models can't write Shakespeare": data distribution beats any harness — gerardsans · 2026-11-15
- Harness skeptic: model training-data support drives performance, not the harness — gerardsans · 2026-11-15
- Security veteran: traditional tooling is obsolete, run an agentic hacker against your own systems — Kyrannio · 2026-11-15