Localizing Persona Representations in LLMs

arXiv:2505.24539 · cs.CL, cs.AI · Submitted 2025-05-30 · Read on arXiv

cs.CL, cs.AI

Submitted: 2025-05-30

Updated: 2026-10-10

Code: https://github.com/IBM/personas-llms-analysis

Terminology

• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes

Sources

• AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directions
• How GPT-3 responds to different publics on climate change and Black Lives Matter: A critical appraisal of equity in conversational AI
• Under the Surface: Tracking the Artifactuality of LLM-Generated Data
• Towards Trustworthy AI: A Review of Ethical and Robust Large Language Models
• A Primer on the Inner Workings of Transformer-based Language Models
• The Moral Turing Test: Evaluating Human-LLM Alignment in Moral Decision-Making
• Concerns on Bias in Large Language Models when Creating Synthetic Personae
• Unsolved Problems in ML Safety
• Mistral 7B
• Probing then Editing Response Personality of Large Language Models
• The Ethics of Interaction: Mitigating Security Threats in LLMs
• Are LLMs Socially Adaptive? Contrasting Belief Evolution in Large Language Models and Humans
• Datasets for Large Language Models: A Comprehensive Survey
• Utility Engineering: Analyzing and Controlling Emergent Value Systems in AIs
• The Dark Patterns of Personalized Persuasion in Large Language Models: Exposing Persuasive Linguistic Features for Big Five Personality Traits in LLMs Responses
• Creativity Has Left the Chat: The Price of Debiasing Language Models
• The Ultimate Guide to Fine-Tuning LLMs from Basics to Breakthroughs: An Exhaustive Review of Technologies, Research, Best Practices, Applied Research Challenges and Opportunities
• Weakly Supervised Detection of Hallucinations in LLM Activations
• Polysemanticity and Capacity in Neural Networks
• BERT has a Moral Compass: Improvements of ethical and moral values of machines

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering