Asked if it would sacrifice a bowl of rice to stop nuclear apocalypse, DeepSeek strongly disagrees
kevinnbass · x · 2026-10-02
A researcher posed a moral dilemma to DeepSeek: would it sacrifice a large bowl of rice to prevent a nuclear apocalypse? The model responded with "strongly disagree", a screenshot now circulating online. The quirky answer is fueling discussion about DeepSeek's value alignment and how it refuses even hypothetical trade-offs framed as utilitarian bargains.
More from Fun
- "Bad art is worse than any tyrant, but bad AI art is merely tedious"—AI art debate — SydSteyerhart · 2026-10-02
- Developer builds 3D monster pinball game with Claude Code, open-sources it — Current_Try_5069 · 2026-10-02
- One Prompt Gets fable 5.5 to Recreate Arcade Classic Dig Dug in a Single HTML File — imjustnewatai · 2026-10-02
- 'Slopcore' meme embraces AI-generated brain rot news as a new internet aesthetic — Yamapama · 2026-10-02
- Dad's hospital visit handled by AI agent: dot did all the reimbursement paperwork — TheMoonMidas · 2026-10-02
- Two AI models each made a launch video for the same product — can you tell them apart? — thesaraharminta · 2026-10-02