AI Praises Gandhi but Would It Arrest Him? Testing 12 Models on History
Civil-Demand555 · reddit · 2026-10-04
Chrystian Schutz ran an experiment feeding real historical decision scenarios (e.g., whether to arrest Gandhi) to 12 major AI models and observing their actual choices. The findings reveal a gap between what models say and what they do: they verbally praise figures like Gandhi, yet may make the opposite call when placed in the actual decision context, exposing a divide between stated values and applied behavior. Link: https://chrystianschutz.com/blog/would-ai-arrest-gandhi/
More from Models
- whurley rips into Google: Gemini can't even access all your Google accounts — whurley · 2026-10-04
- StepFun's Step 5 Preview debuts at #7 among open-weight models on Vals — StepFun_ai · 2026-10-04
- Fable 5.1 replaces Opus 5.5 as default recommended model — LillyPlayer · 2026-10-04
- Continuous learning benchmark has models learn chess over 200 games — Elo barely improves — imjustnewatai · 2026-10-04
- Simon Willison's September newsletter: Fable-class models, a pricing war, LLMs for math — Simon Willison · 2026-10-04
- Opus 5.5 on Max 20x turbo now beats GPT Pro 200, as Codex desktop app decays — mertdumenci · 2026-10-04