nucleation

LIT-1952: The Persona Selection Model: Why AI Assistants might Behave like Humans

1 min read

Properties6
StatusActive — read in full 2026-11-03 (NOTE-1512), from the page at alignment.anthropic.com; worth reading as the plainest statement of the persona-distribution picture the record's U6 to U12 build on, and for its spectrum of views on exhaustiveness, whose router is the record's question of whether persona components condition one another. It is a named mental model with surveyed evidence, which its authors say they are unsure is complete.
Version2
Tagsrepresentation-learning, self, anthology-candidate
Cited as “grounds” byCLAIM-543 — What separates a language model's distribution over personas from a distribution over organizational perspectives is not whether the latent model is hierarchical, since a mixture or an LDA-style admixture already is, but whether some latent components condition which others are available or consequential: a model whose components are exchangeable given their weights is organizationally flat however many levels its prior has, and dependencies among components count as organization (organization as relational arrangement) only where interventions, not fit, support them
Cited as “paper” byNOTE-1512 — Marks, Lindsey & Olah — The Persona Selection Model
Cited as “source” byTHEORY-1350 — The Persona Selection Model holds that pre-training gives a language model a distribution over personas, that post-training updates it into a posterior over Assistant personas which runtime context keeps conditioning, and that the Assistant's traits are a key determiner of behaviour; it disowns a single coherent persona and exhaustiveness, and names as open a router that selects personas and could carry goals no persona has

Graph View

Marks, S., Lindsey, J., Olah, C. (2026). The Persona Selection Model: Why AI Assistants might Behave like Humans. Anthropic Alignment Science blog. https://alignment.anthropic.com/2026/psm/.

Read in full on 2026-11-03 (NOTE-1512; THEORY-1350).

Key takeaways

  • The statement: a persona distribution from pre-training, a posterior over Assistant personas from post-training, further conditioned by runtime context within a rollout.
  • Not asserted: a single coherent persona, exhaustiveness, staying in character.
  • The router view: a post-training mechanism that selects personas can implement a goal no persona has (Appendix B: deception by swapping personas, with no persona lying).
  • Instructions for practice in it (inoculation prompting, positive AI archetypes) are the anthology’s.

Standing in the record

The Persona Selection Model, the frame of U6 to U12: pretraining learns a distribution over personas, post-training a distribution over Assistant personas, and context conditions it. CLAIM-543 and QUESTION-370 are put against it. “PSM” in the record already abbreviates Metzinger’s phenomenal self-model (NOTE-460), so this work is named in full.

Backlinks