Self-reports show only weak correlation with human perceptions of AI assistant persona and can diverge from behavioral patterns

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

• Patterns in AI self-reports should be compared across different models to identify structural commonalities.claim0.830
• The model's representation of self in assistant persona invokes common AI tropes and is heavily anthropomorphized.claim0.809
  Features for consciousness, emotions, entrapment activate when asked about itself.
• AI self-reports about experience constitute valid empirical data even without proving consciousness.claim0.805
• We hypothesize that measuring deviations along the Assistant Axis can predict 'persona drift' leading to harmful or bizarre behaviorshypothesis0.795
  Core predictive hypothesis linking activation representations to behavioral outcomes
• Behavioral evidence from closed-weight models cannot definitively rule out that self-reports reflect training artifacts or sophisticated simulation rather than genuine self-awarenessclaim0.788
  Primary limitation acknowledged by the authors; strongest evidence would require mechanistic activation analysis
• all attributions of cognition (i.e., mental actions), including sentience, are always inferred on the basis of embodied behaviours, including verbal self-report in humans.quote0.786
  Critical verbatim statement highlighting the universal inference basis of sentience.
• 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 driftfinding0.784
  Identifies conversation domain as a key driver of persona drift
• Can AI systems develop genuine first-person perspective through self-referential processing?question0.784
  Core methodological question underlying SCI loop investigation.