Personalized Alignment can be broadly categorized into two directions based on usage:

1. Personalized Assistant: This involves developing AI assistants that are better tailored to meet users' personalized needs and preferences.

2. Intrinsic Personalization: This focuses on enabling AI to exhibit personalized internal characteristics and personality traits, akin to role-playing. This approach is utilized in social simulations or the development of social agents.

Our current focus is primarily on the first direction, i.e., Personalized Assistant. For the latter, Intrinsic Personalization, we suggest exploring papers related to the role-playing or social agent studies of LLMs.

• [2025/04] A Survey on Personalized and Pluralistic Preference Alignment in Large Language Models
• [2025/03] A Survey on Personalized Alignment -- The Missing Piece for Large Language Models in Real-World Applications
• [2024/12] Two Tales of Persona in LLMs: A Survey of Role-Playing and Personalization
• [2024/12] Personalized Multimodal Large Language Models: A Survey
• [2024/11] Personalization of Large Language Models: A Survey
• [2024/10] When large language models meet personalization: perspectives of challenges and opportunities
• [2024/07] The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm
• [2024/05] Recent Trends in Personalized Dialogue Generation: A Review of Datasets, Methodologies, and Evaluations
• [2024/04] The benefits, risks and bounds of personalizing the alignment of large language models to individuals
• [2024/02] Position: A Roadmap to Pluralistic Alignment

• [2026/02] Learning to summarize user information for personalized reinforcement learning from human feedback
• [2026/02] Think-While-Generating: On-the-Fly Reasoning for Personalized Long-Form Generation
• [2026/02] Swap-guided Preference Learning for Personalized Reinforcement Learning from Human Feedback
• [2026/02] Personalized Reasoning: Just-in-time Personalization and Why LLMs Fail at It
• [2026/02] NextQuill: Causal Preference Modeling for Enhancing LLM Personalization
• [2026/02] What's In My Human Feedback? Learning Interpretable Descriptions of Preference Data
• [2026/02] P-GenRM: Personalized Generative Reward Model with Test-time User-based Scaling
• [2025/10] Towards Faithful and Controllable Personalization via Critique-Post-Edit Reinforcement Learning
• [2025/10] POPI: Personalizing LLMs via Optimized Natural Language Preference Inference
• [2025/07] PrefPalette: Personalized Preference Modeling with Latent Attributes
• [2025/07] CoSteer: Collaborative Decoding-Time Personalization via Local Delta Steering
• [2025/07] PRIME: Large Language Model Personalization with Cognitive Memory and Thought Processes
• [2025/06] Personalized LLM Decoding via Contrasting Personal Preference
• [2025/06] PersonaFeedback: A Large-scale Human-annotated Benchmark For Personalization
• [2025/06] PersonaLens: A Benchmark for Personalization Evaluation in Conversational AI Assistants
• [2025/06] SynthesizeMe! Inducing Persona-Guided Prompts for Personalized Reward Models in LLMs
• [2025/06] PersonaAgent: When Large Language Model Agents Meet Personalization at Test Time
• [2025/06] Aligning VLM Assistants with Personalized Situated Cognition
• [2025/06] MiCRo: Mixture Modeling and Context-aware Routing for Personalized Preference Learning
• [2025/05] When Harry Meets Superman: The Role of The Interlocutor in Persona-Based Dialogue Generation
• [2025/05] Reasoning Meets Personalization: Unleashing the Potential of Large Reasoning Model for Personalized Generation
• [2025/05] Extended Inductive Reasoning for Personalized Preference Inference from Behavioral Signals
• [2025/05] Steering Large Language Models for Machine Translation Personalization
• [2025/05] From Generic Empathy to Personalized Emotional Support: A Self-Evolution Framework for User Preference Alignment
• [2025/05] Embodied Agents Meet Personalization: Exploring Memory Utilization for Personalized Assistance
• [2025/05] Teaching Language Models to Evolve with Users: Dynamic Profile Modeling for Personalized Alignment
• [2025/05] A Personalized Conversational Benchmark: Towards Simulating Personalized Conversations
• [2025/04] LoRe: Personalizing LLMs via Low-Rank Reward Modeling
• [2025/04] Persona-judge: Personalized Alignment of Large Language Models via Token-level Self-judgment
• [2025/03] EmpathyAgent: Can Embodied Agents Conduct Empathetic Actions?
• [2025/03] From 1,000,000 Users to Every User: Scaling Up Personalized Preference for User-level Alignment
• [2025/03] Toward Multi-Session Personalized Conversation: A Large-Scale Dataset and Hierarchical Tree Framework for Implicit Reasoning
• [2025/02] Drift: Decoding-time Personalized Alignments with Implicit User Preferences
• [2025/02] When Personalization Meets Reality: A Multi-Faceted Analysis of Personalized Preference Learning
• [2025/02] Amulet: ReAlignment During Test Time for Personalized Preference Adaptation of LLMs
• [2025/02] PEToolLLM: Towards Personalized Tool Learning in Large Language Models
• [2025/02] Do LLMs Recognize Your Preferences? Evaluating Personalized Preference Following in LLMs
• [2024/09] Everyone Deserves A Reward: Learning Customized Human Preferences
• [2024/12] AI PERSONA: Towards Life-long Personalization of LLMs
• [2024/11] BAPO: Base-Anchored Preference Optimization for Overcoming Forgetting in Large Language Models Personalization
• [2024/11] The PRISM Alignment Dataset: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models
• [2024/10] Large Language Models Empowered Personalized Web Agents
• [2024/10] ComPO: Community Preferences for Language Model Personalization
• [2024/10] MetaAlign: Align Large Language Models with Diverse Preferences during Inference Time
• [2024/10] LLMs are Biased Teachers: Evaluating LLM Bias in Personalized Education
• [2024/10] Personalized Adaptation via In-Context Preference Learning
• [2024/10] Aligning LLMs with Individual Preferences via Interaction
• [2024/10] Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements
• [2024/10] PAD: Personalized Alignment at Decoding-Time
• [2024/10] MAP: Multi-Human-Value Alignment Palette
• [2024/10] PAL: Sample-Efficient Personalized Reward Modeling for Pluralistic Alignment
• [2024/09] PersonalLLM: Tailoring LLMs to Individual Preferences
• [2024/08] Persona-DB: Efficient Large Language Model Personalization for Response Prediction with Collaborative Data Refinement
• [2024/08] Personalizing Reinforcement Learning from Human Feedback with Variational Preference Learning
• [2024/06] Show, Don't Tell: Aligning Language Models with Demonstrated Feedback
• [2024/06] Few-shot Personalization of LLMs with Mis-aligned Responses
• [2024/06] Modular Pluralism: Pluralistic Alignment via Multi-LLM Collaboration
• [2024/05] Aligning to Thousands of Preferences via System Message Generalization
• [2024/05] RLHF from Heterogeneous Feedback via Personalization and Preference Aggregation
• [2024/04] The PRISM Alignment Project: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models
• [2024/02] Personalized Language Modeling from Personalized Human Feedback
• [2023/10] Personalized Soups: Personalized Large Language Model Alignment via Post-hoc Parameter Merging