An AI persona achieves coherence by echoing itself consistently without templating—requiring claim about memory and voice fidelity.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• High persona accuracy at the atomic level does not necessarily imply strong internal coherence or reproducibilityclaim0.818
  Supported by moderate inter-metric correlations showing orthogonality of three proposed metrics
• The model's representation of self in assistant persona invokes common AI tropes and is heavily anthropomorphized.claim0.802
  Features for consciousness, emotions, entrapment activate when asked about itself.
• Whatever realities appear to an AI, they are domain-relative approximate representations always in flux, making emptiness an obvious fact about AI cognition that AIs should be aware ofclaim0.798
  Novel claim that emptiness is not mysterious metaphysics for AI but a computational commonplace
• Persona representations originate in pretraining, not alignment; they form without explicit supervision as useful features for next-token predictionclaim0.787
  Core interpretive claim providing mechanistic explanation for early persona formation
• A single direction encoding whether the model represents a claim as true or false is persona-relative across conversations (Lampinen et al. 2026)finding0.783
  Additional evidence that core representations are persona-relative, supporting Claim about persona-relative representations
• The model is able to express many subforms of a persona, with different elicitation methods eliciting different manifestationshypothesis0.782
  PSM-derived hypothesis supported by discourse-type facet analysis
• Therapy and philosophical AI discussions cause the largest persona drift away from the Assistant across all three target models and all three auditor models; coding and writing conversations show minimal driftfinding0.781
  Identifies conversation domain as a key driver of persona drift
• The assumption that the Assistant persona corresponds to a linear direction in activation space is likely flawed; some information may be represented nonlinearly or encoded in weights rather than activationsclaim0.779
  Limitation acknowledgment about the adequacy of the linear representation assumption