mlsanityData to 14 Oct 2026

Steerable Chatbots: Exploring Personalization Control Interfaces via LLM Activation Steering

arXiv 2505.04260 ↗First posted 7 May 2025cs.HC cs.AIcited by 1

On the map: Mechanistic interpretability → Activation steering methods

Abstract (arXiv)

Personalizing LLM responses typically requires users to articulate their preferences through prompting, which can be burdensome at cold start and difficult to articulate in natural language. We introduce an alternative paradigm, steerable chatbots: rather than asking users to describe what they want, let them directly manipulate it via a linear factor. We implement this through activation steering, leveraging a linear scalar to control how strongly a preference is expressed in the LLM's output. We first assess the computational viability of activation steering as a method to control granular preference expression, then we explore how the factor can be exposed to users. We prototype three activation steering interface designs that vary on the axes of agency (user-led vs. system-driven) and fluidity (static vs. adaptive). A within-subjects user study (n=14) in cold-start personalization tasks shows the potential for steerable chatbots to align better with underlying user preferences than prompting alone, while revealing heterogeneous values around control, persistence, and transparency in LLM personalization.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability. Builds on 47 of its 99 references that are in the map. First posted on arXiv 7 May 2025.

Builds on

47 of 99 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025Personalized Text Generation with Contrastive Activation Steering7 Mar
  2. 2024Improving Instruction-Following in Language Models through Activation Steering15 Oct
  3. 2024Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization28 May
  4. 2023Steering Llama 2 via Contrastive Activation Addition9 Dec
  5. 2022Extracting Latent Steering Vectors from Pretrained Language Models10 May

Other references · 42

  1. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  2. 2024Gemma 2: Improving Open Language Models at a Practical Sizecited by 1,998
  3. 2024A Survey on LLM-as-a-Judgecited by 1,423
  4. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,302
  5. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  6. 2023OpenAssistant Conversations -- Democratizing Large Language Model Alignmentcited by 721
  7. 2020TweetEval: Unified Benchmark and Comparative Evaluation for Tweet Classificationcited by 417
  8. 2023Personalized Soups: Personalized Large Language Model Alignment via Post-hoc Parameter Mergingcited by 280
  9. 2024Personal LLM Agents: Insights and Survey about the Capability, Efficiency and Securitycited by 260
  10. 2023Rewarded soups: towards Pareto-optimal alignment by interpolating weights fine-tuned on diverse rewardscited by 243
  11. 2025AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoderscited by 218
  12. 2024From Persona to Personalization: A Survey on Role-Playing Language Agentscited by 212
30 more
  1. 2024Programming Refusal with Conditional Activation Steeringcited by 206
  2. 2019Gradio: Hassle-Free Sharing and Testing of ML Models in the Wildcited by 178
  3. 2024Rewards-in-Context: Multi-objective Alignment of Foundation Models with Dynamic Preference Adjustmentcited by 175
  4. 2024Personalizing Reinforcement Learning from Human Feedback with Variational Preference Learningcited by 155
  5. 2023SteerLM: Attribute Conditioned SFT as an (User-Steerable) Alternative to RLHFcited by 111
  6. 2023A Confederacy of Models: a Comprehensive Evaluation of LLMs on Creative Writingcited by 99
  7. 2024Decoding-Time Language Model Alignment with Multiple Objectivescited by 94
  8. 2023Pearl: Personalizing Large Language Model Writing Assistants with Generation-Calibrated Retrieverscited by 92
  9. 2024Stable LM 2 1.6B Technical Reportcited by 90
  10. 2018The Geometry of Culture: Analyzing Meaning through Word Embeddingscited by 87
  11. 2023The Metacognitive Demands and Opportunities of Generative AIcited by 87
  12. 2023Luminate: Structured Generation and Exploration of Design Space with Large Language Models for Human-AI Co-Creationcited by 64
  13. 2024Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineeringcited by 63
  14. 2023Prompting AI Art: An Investigation into the Creative Skill of Prompt Engineeringcited by 43
  15. 2024A Language Model's Guide Through Latent Spacecited by 43
  16. 2024Conditional Language Policy: A General Framework for Steerable Multi-Objective Finetuningcited by 43
  17. 2023PromptPaint: Steering Text-to-Image Generation Through Paint Medium-like Interactionscited by 34
  18. 2024Aligning Language Models with Demonstrated Feedbackcited by 34
  19. 2018SemAxis: A Lightweight Framework to Characterize Domain-Specific Word Semantics Beyond Sentimentcited by 26
  20. 2023Understanding Users' Dissatisfaction with ChatGPT Responses: Types, Resolving Tactics, and the Effect of Knowledge Levelcited by 26
  21. 2024Context Steering: Controllable Personalization at Inference Timecited by 22
  22. 2024Improving Context-Aware Preference Modeling for Language Modelscited by 22
  23. 2024Implicit Personalization in Language Models: A Systematic Studycited by 18
  24. 2024Identifying and Manipulating Personality Traits in LLMs Through Activation Engineeringcited by 14
  25. 2022GANSlider: How Users Control Generative Models for Images using Multiple Sliders with and without Feedforward Informationcited by 13
  26. 2024Textoshop: Interactions Inspired by Drawing Software to Facilitate Text Editingcited by 12
  27. 2023MIRACLE: Towards Personalized Dialogue Generation with Latent-Space Multiple Personal Attribute Controlcited by 11
  28. 2025Code Shaping: Iterative Code Editing with Free-form AI-Interpreted Sketchingcited by 7
  29. 2024DrawTalking: Building Interactive Worlds by Sketching and Speakingcited by 5
  30. 2024Navigating the Unknown: A Chat-Based Collaborative Interface for Personalized Exploratory Taskscited by 4

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025ThoughtProbe: Classifier-Guided LLM Thought Space Exploration via Probing Representationscited by 3

Similar, not cited

Close in wording, no citation either way

  1. 2025SteerX: Disentangled Steering for LLM Personalization
  2. 2025CUPID: Evaluating Personalized and Contextualized Alignment of LLMs from Interactions
  3. 2026Towards Natural Personalization: Evaluating Long-Horizon Preference Following in Personalized User-LLM Interactions
  4. 2025SynthesizeMe! Inducing Persona-Guided Prompts for Personalized Reward Models in LLMs
  5. 2026Synthetic Interaction Data for Scalable Personalization in Large Language Models
  6. 2026FleetPilot: Fine-Grained Steering Empowers Real-World LLM Personalization
  7. 2024PersonalLLM: Tailoring LLMs to Individual Preferences
  8. 2026Preference Heads in Large Language Models: A Mechanistic Framework for Interpretable Personalization