MCNAIR(@mcnairai):In new work from @tomjiralerspong, we find that some claims in "Persona Vectors" from Anthropic likely fail to replicate on real chat data. Thomas discusses failures in relying on synthetic data for experimentation, and provides recommendations for researchers.

MCNAIR

@mcnairai

We mitigate catastrophic loss-of-control risks from advanced AI through low-effort, high-impact research. Posts may not represent the views of all staff.

Joined May 2026

11 Following 94 Followers

MCNAIR@mcnairai

2026.09.13 21:50

In new work from @tomjiralerspong, we find that some claims in "Persona Vectors" from Anthropic likely fail to replicate on real chat data. Thomas discusses failures in relying on synthetic data for experimentation, and provides recommendations for researchers.