Persona induction in LLMs produces stable task-dependent cognitive performance shifts aligned with human patterns at 73.68%, and Dynamic Persona Routing outperforms static personas without retraining.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Systematic Analysis of the Impact of Persona Steering on LLM Capabilities
Persona induction in LLMs produces stable task-dependent cognitive performance shifts aligned with human patterns at 73.68%, and Dynamic Persona Routing outperforms static personas without retraining.