Model welfare · Article · Eleos AI Research

Probing persona-dependent preferences in language models

18 May 2026

Outline · 5 paragraphs

Probing persona-dependent preferences in language models¶2
Large language models (LLMs) can be said to have preferences: they reliably pick certain tasks and outputs over others, and preferences shaped by post-training and system prompts appear to shape much of their behaviour. But models can also adopt different personas which have radically different pref¶3

How to cite

A citation names the document, the version and the place.

eleos:probing-persona-dependent-preferences-in-language-models/r1/b1

https://safetypage.noahairmet.com/d/eleos/probing-persona-dependent-preferences-in-language-models/r1/b1

BibTeX

@misc{eleosairesearch2026,
  title = {Probing persona-dependent preferences in language models},
  author = {{Eleos AI Research}},
  year = {2026},
  url = {https://eleosai.org/post/probing-persona-dependent-preferences-in-language-models}
}

Versions

Version	Published	Read
r1 latest	18 May 2026	At source

Licence

So you: Search it, cite it and check quotes here; read it at its source
Read at: eleosai.org/post/probing-persona-dependent-preferences-in-language-models, 10 Nov 2026