Crossed random-effects models on LLM word ratings show 16.9% variance from genuine stimulus-specific individuality, exceeding null models and forming coherent per-model fingerprints.
hub
arXiv preprint arXiv:2303.13988 , year=
11 Pith papers cite this work. Polarity classification is still indexing.
11
Pith papers citing it
hub tools
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
LLMs perform in-context learning as trajectories through a structured low-dimensional conceptual belief space, with the structure visible in both behavior and internal representations and causally manipulable via interventions.
Steering Dark Triad features in an LLM increases exploitative and aggressive behavior while leaving strategic deception and cognitive empathy unchanged, indicating dissociable antisocial pathways.
Expanded recall in LLM agents erodes cooperative intent in multi-agent social dilemmas, observed in 18 of 28 model-game settings.
TEA Nets extracts agents, events, and targets from text to reveal emotional and semantic patterns in conspiracy theories and psychotherapy transcripts from humans and LLMs.
LLMs diverge from human goal selection in self-directed learning by exploiting single solutions with low variability across instances.
Lexical frequency is a stronger predictor of metaphor novelty than LM surprisal, with the surprisal-novelty link peaking early in training before declining as surprisal becomes more aligned with frequency.
ELDER-SIM builds personality-stable elderly digital twins via LLM orchestration with OCEAN traits, Beck CBT diagrams, long-term memory, and LoRA fine-tuning on CHARLS data, validated by Cronbach's alpha 0.70-0.94 and ICC 0.85-0.96.
NormCoRe is a replication-by-translation framework that maps human subject studies onto multi-agent AI environments, showing AI normative judgments on fairness differ from human baselines and vary with model choice and persona language.
STAMP-LLM is a two-phase psychometric protocol for designing and applying bias measures to LLMs, illustrated with one explicit and two implicit racial bias tests.
EU consumer law needs adaptation to accommodate AI agents acting as autonomous purchasing decision-makers.
citing papers explorer
-
Machine individuality: Separating genuine idiosyncrasy from response bias in large language models
Crossed random-effects models on LLM word ratings show 16.9% variance from genuine stimulus-specific individuality, exceeding null models and forming coherent per-model fingerprints.
-
Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space
LLMs perform in-context learning as trajectories through a structured low-dimensional conceptual belief space, with the structure visible in both behavior and internal representations and causally manipulable via interventions.
-
Exploitation Without Deception: Dark Triad Feature Steering Reveals Separable Antisocial Circuits in Language Models
Steering Dark Triad features in an LLM increases exploitative and aggressive behavior while leaving strategic deception and cognitive empathy unchanged, indicating dissociable antisocial pathways.
-
The Memory Curse: How Expanded Recall Erodes Cooperative Intent in LLM Agents
Expanded recall in LLM agents erodes cooperative intent in multi-agent social dilemmas, observed in 18 of 28 model-game settings.
-
The TEA Nets framework combines AI and cognitive network science to model targets, events and actors in text
TEA Nets extracts agents, events, and targets from text to reveal emotional and semantic patterns in conspiracy theories and psychotherapy transcripts from humans and LLMs.
-
Language Model Goal Selection Differs from Humans' in a Self-Directed Learning Task
LLMs diverge from human goal selection in self-directed learning by exploiting single solutions with low variability across instances.
-
The Frequency Confound in Language-Model Surprisal and Metaphor Novelty
Lexical frequency is a stronger predictor of metaphor novelty than LM surprisal, with the surprisal-novelty link peaking early in training before declining as surprisal becomes more aligned with frequency.
-
Elder-Sim: A Psychometrically Validated Platform for Personality-Stable Elderly Digital Twins
ELDER-SIM builds personality-stable elderly digital twins via LLM orchestration with OCEAN traits, Beck CBT diagrams, long-term memory, and LoRA fine-tuning on CHARLS data, validated by Cronbach's alpha 0.70-0.94 and ICC 0.85-0.96.
-
Normative Common Ground Replication (NormCoRe): Replication-by-Translation for Studying Norms in Multi-Agent AI
NormCoRe is a replication-by-translation framework that maps human subject studies onto multi-agent AI environments, showing AI normative judgments on fairness differ from human baselines and vary with model choice and persona language.
-
Designing Psychometric Bias Measures for ChatBots: An Application to Racial Bias Measurement
STAMP-LLM is a two-phase psychometric protocol for designing and applying bias measures to LLMs, illustrated with one explicit and two implicit racial bias tests.
-
Consumer Law for AI Agents
EU consumer law needs adaptation to accommodate AI agents acting as autonomous purchasing decision-makers.