mlsanityData to 14 Oct 2026

Steering Risk Preferences in Large Language Models by Aligning Behavioral and Neural Representations

arXiv 2505.11615 ↗First posted 16 May 2025cs.CL cs.AIcited by 6

On the map: Mechanistic interpretability → Activation steering methods

Abstract (arXiv)

Changing the behavior of large language models (LLMs) can be as straightforward as editing the Transformer's residual streams using appropriately constructed "steering vectors." These modifications to internal neural activations, a form of representation engineering, offer an effective and targeted means of influencing model behavior without retraining or fine-tuning the model. But how can such steering vectors be systematically identified? We propose a principled approach for uncovering steering vectors by aligning latent representations elicited through behavioral methods (specifically, Markov chain Monte Carlo with LLMs) with their neural counterparts. To evaluate this approach, we focus on extracting latent risk preferences from LLMs and steering their risk-related outputs using the aligned representations as steering vectors. We show that the resulting steering vectors successfully and reliably modulate LLM outputs in line with the targeted behavior.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability. Builds on 22 of its 44 references that are in the map. First posted on arXiv 16 May 2025.

Builds on

22 of 44 references.

Closest ancestors · 2

Same direction, newest first

  1. 2023Steering Llama 2 via Contrastive Activation Addition9 Dec
  2. 2023Steering Language Models With Activation Engineering20 Aug

Other references · 20

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  3. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,211
  4. 2019Fine-Tuning Language Models from Human Preferencescited by 2,150
  5. 2024Gemma 2: Improving Open Language Models at a Practical Sizecited by 1,998
  6. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,302
  7. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!cited by 1,154
  8. 2020Aligning AI With Shared Human Valuescited by 723
  9. 2023Getting aligned on representational alignmentcited by 143
  10. 2023Inspecting and Editing Knowledge Representations in Language Modelscited by 141
  11. 2019Bias Correction of Learned Generative Models using Likelihood-Free Importance Weightingcited by 70
  12. 2025Utility Engineering: Analyzing and Controlling Emergent Value Systems in AIscited by 67
8 more
  1. 2024Large Language Models Assume People are More Rational than We Really arecited by 54
  2. 2023Large language models predict human sensory judgments across six modalitiescited by 48
  3. 2024Incoherent Probability Judgments in Large Language Modelscited by 30
  4. 2020Gibbs Sampling with Peoplecited by 28
  5. 2024Eliciting the Priors of Large Language Models using Iterated In-Context Learningcited by 19
  6. 2024AutoElicit: Using Large Language Models for Expert Prior Elicitation in Predictive Modellingcited by 15
  7. 2024Language Models Trained to do Arithmetic Predict Human Risky and Intertemporal Choicecited by 12
  8. 2024Recovering Mental Representations from Large Language Models with Markov Chain Monte Carlocited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 5.

6 papers.

  1. 2026Can Revealed Preferences Clarify LLM Alignment and Steering?cited by 5
  2. 2026Auditing Asset-Specific Preferences in Financial Large Language Models: Evidence from Bitcoin Representations and Portfolio Allocationcited by 1
  3. 2026High-Stakes Decisions with Language Models: Insights from Emergency Triagecited by 1
  4. 2025Can Risk-taking AI-Assistants suitably represent entitiescited by 0
  5. 2026Temporal Preference Concepts and their Functions in a Large Language Modelcited by 0
  6. 2026The Internal Anatomy of Strategic Choice in Large Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization
  2. 2025Risk Profiling and Modulation for LLMs
  3. 2024Few-shot Steerable Alignment: Adapting Rewards and LLM Policies with Neural Processes
  4. 2026Where to Steer: Input-Dependent Layer Selection for Steering Improves LLM Alignment
  5. 2026On the Non-Identifiability of Steering Vectors in Large Language Models
  6. 2025Improved Representation Steering for Language Models
  7. 2026Understanding Unreliability of Steering Vectors in Language Models: Geometric Predictors and the Limits of Linear Approximations
  8. 2024Towards Inference-time Category-wise Safety Steering for Large Language Models