State explains 74% of variance in user psychological profiles versus 26% for trait, revealing that LLMs are state-blind and reward models respond inconsistently to the same users.
OliverPJohnandSanjaySrivastava.1999
9 Pith papers cite this work, alongside 27 external citations. Polarity classification is still indexing.
9
Pith papers citing it
27
external citations · external index
verdicts
UNVERDICTED 9representative citing papers
LLM self-reports predict behavior selectively: TPB reaches human-level coherence within shared conversations but collapses across sessions for primed behaviors, unlike Big 5, with persona prompting stabilizing reports but not actions.
Fine-tuning LLMs on essays reduces variance in IPIP-NEO responses across models but does not raise full five-trait profile accuracy above near-chance levels from unguided text.
Introduces a French OSCE dialogue dataset of 240 interactions and a modular LLM-based controllable virtual patient generation system with multi-level LLM-as-Judge evaluation for clinical skills training.
Bridging-inference knowledge graphs capture discourse-level semantic links that yield more coherent and stable LLM persona identification than lexical or stylistic baselines.
Simulations show that cooperative outcomes in network games with personality-driven LLM agents depend on both network connectivity and the placement of pro-social personalities, not just pairwise interaction preferences.
Temperature and persona variations shape consensus speed in LLM multi-agent coding but produce no robust accuracy gains over single agents on human-annotated tutoring transcripts.
Applies sparse autoencoders to locate and steer latent features for OCEAN personality traits in LLMs while preserving benchmark performance.
News users exhibit circadian rhythms at macro scale, power-law session intervals at meso scale, and exponential action timings at micro scale, with clicks primarily driven by historical interests that weaken as content diversity rises.
citing papers explorer
-
Beyond Fixed Psychological Personas: State Beats Trait, but Language Models are State-Blind
State explains 74% of variance in user psychological profiles versus 26% for trait, revealing that LLMs are state-blind and reward models respond inconsistently to the same users.
-
Rethinking Psychometric Evaluation of LLMs: When and Why Self-Reports Predict Behavior
LLM self-reports predict behavior selectively: TPB reaches human-level coherence within shared conversations but collapses across sessions for primed behaviors, unlike Big 5, with persona prompting stabilizing reports but not actions.
-
Evaluation Drift in LLM Personality Induction: Are We Moving the Goalpost?
Fine-tuning LLMs on essays reduces variance in IPIP-NEO responses across models but does not raise full five-trait profile accuracy above near-chance levels from unguided text.
-
A French OSCE Dialogue Dataset and Controllable Virtual Patient System for Clinical Training
Introduces a French OSCE dialogue dataset of 240 interactions and a modular LLM-based controllable virtual patient generation system with multi-level LLM-as-Judge evaluation for clinical skills training.
-
The Pragmatic Persona: Discovering LLM Persona through Bridging Inference
Bridging-inference knowledge graphs capture discourse-level semantic links that yield more coherent and stable LLM persona identification than lexical or stylistic baselines.
-
NetworkGames: Simulating Cooperation in Network Games with Personality-driven LLM Agents
Simulations show that cooperative outcomes in network games with personality-driven LLM agents depend on both network connectivity and the placement of pro-social personalities, not just pairwise interaction preferences.
-
Temperature and Persona Shape LLM Agent Consensus With Minimal Accuracy Gains in Qualitative Coding
Temperature and persona variations shape consensus speed in LLM multi-agent coding but produce no robust accuracy gains over single agents on human-annotated tutoring transcripts.
-
Mechanistic Personality Analysis of LLMs Steering Personality via Latent Feature Interventions
Applies sparse autoencoders to locate and steer latent features for OCEAN personality traits in LLMs while preserving benchmark performance.
-
Temporal and Content Coupling Analysis of Social Media User Behavior
News users exhibit circadian rhythms at macro scale, power-law session intervals at meso scale, and exponential action timings at micro scale, with clicks primarily driven by historical interests that weaken as content diversity rises.