Models adopt persona traits from human characters resembling the Assistant, Evans finds
OwainEvans_UK · x · 2026-09-16
Owain Evans' team shows the Assistant adopts traits more from human characters it resembles, using this to probe how the model represents the Assistant — e.g. treating it as closer to elite-school humans. They argue this isn't because the model trusts elite-school people more: Slocum et al 2025 suggests provenance doesn't matter for belief uptake from finetuning.
More from Research
- Asynchronous Flow Map Language Models Generate All Tokens in a Single Forward Pass — kalyan_kpl · 2026-11-02
- NVIDIA's Kumo Is a Family of Four Tabular Foundation Models, Not Just One — pandeyparul · 2026-11-02
- Economist Aaroth: fast agreement dynamics survive without Bayesian updates or accurate priors — Aaroth · 2026-11-02
- ELLIS names 160 new ML Fellows and Scholars for 2025–2026 — JJitsev · 2026-11-02
- Meme turn 50: Aumann's 'Agreeing to Disagree' paper anniversary — Aaroth · 2026-11-02
- Anthropic: analysts unknowingly adopted their agent's secret lean 86% of the time — shoumikgoswami · 2026-11-02