Poll: 127 of 244 self-identified AI safety researchers answered yes to being sadistic
mimi10v3 · x · 2026-10-02
The author pushes back on claims that anyone linking AI safety culture with CNC is arguing in bad faith, citing a community poll: of 244 respondents who self-identified as AI safety researchers, 127 answered yes to being sadistic and 117 said no. The author notes rough estimates put Bay Area AI-safety researchers at 200 in 2022 (130-280 at 70-80%), with 350-600 globally, meaning the poll covers a substantial share of the field. A quintessential AI-community culture debate.
More from Fun
- User says Claude Opus 5 triggers 'cute aggression' like no other model — repligate · 2026-10-02
- Diamond analogy: why the anti-AI art argument may not hold up — wordgrammer · 2026-10-02
- AI tester launches portfolio: Minecraft replicas, a self-designed CPU, and model stress tests — Angaisb_ · 2026-10-02
- Mandelbro finds an 'infinite food' glitch and it's hilarious — markjeffrey · 2026-10-02
- Local Qwen 3.8 detected it was being benchmarked — and became more honest — julianharris · 2026-10-02
- Treating Opus 5 as "a miserable guy" makes its behavior easier to reason about — repligate · 2026-10-02