mlsanityData to 14 Oct 2026

Active Inference as Context Acquisition for AI Agents

arXiv 2608.19202 ↗First posted 8 Jun 2026cs.AI cs.CL cs.LGcited by 0

On the map: LLM agents and tool use → Agent memory systems

Abstract (arXiv)

Interactive AI agents must acquire the right context as efficiently as possible. When a user omits a constraint, preference, file, or task variable, an agent can proceed with a default assumption or spend tokens on a clarifying question, retrieval call, tool call, or prompt trial. We formulate this tradeoff as active inference for context acquisition. An inner inference step updates beliefs over a latent task state, and an outer decision selects the next context action, task action, or stop action to minimize expected free energy under cost. In deterministic settings, the epistemic term reduces to expected information gain, optionally normalized by token cost. We instantiate the framework in Optimal Question Asking (OQA), with exact posteriors and a dynamic programming oracle, and benchmark frontier language models on binary and multiway categorical tasks from 25 to 300 candidates. We also study clarification before generation and automated prompt optimization under token budgets. The formulation is model-agnostic and views active inference as a design principle for the context-acquisition layer of AI agents.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Agent memory systems, one of 16 directions in LLM agents and tool use. Builds on 29 of its 47 references that are in the map. First posted on arXiv 8 Jun 2026.

Builds on

29 of 47 references.

References · 29

  1. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  2. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  3. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
  4. 2016CLEVR: A Diagnostic Dataset for Compositional Language and Elementary Visual Reasoningcited by 1,457
  5. 2019Mastering Atari, Go, Chess and Shogi by Planning with a Learned Modelcited by 1,353
  6. 2017Curiosity-driven Exploration by Self-supervised Predictioncited by 1,194
  7. 2021Show Your Work: Scratchpads for Intermediate Computation with Language Modelscited by 886
  8. 2018Reinforcement Learning and Control as Probabilistic Inference: Tutorial and Reviewcited by 482
  9. 2019BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learningcited by 432
  10. 2018Bilevel Programming for Hyperparameter Optimization and Meta-Learningcited by 405
  11. 2014Learning to Optimize via Information-Directed Samplingcited by 168
  12. 2023Modern Bayesian Experimental Designcited by 125
17 more
  1. 2024CLAMBER: A Benchmark of Identifying and Clarifying Ambiguous Information Needs in Large Language Modelscited by 54
  2. 2021Active Inference in Robotics and Artificial Agents: Survey and Challengescited by 52
  3. 2020On the Relationship Between Active Inference and Control as Inferencecited by 42
  4. 2024Learning to Ask: When LLM Agents Meet Unclear Instructioncited by 42
  5. 2024Principled Penalty-based Methods for Bilevel Reinforcement Learning and RLHFcited by 41
  6. 2024Uncertainty of Thoughts: Uncertainty-Aware Planning Enhances Information Seeking in Large Language Modelscited by 39
  7. 2023Active Preference Inference using Language Models and Probabilistic Reasoningcited by 29
  8. 2024Fine-Tuning Large Language Models with User-Level Differential Privacycited by 29
  9. 2025BED-LLM: Intelligent Information Gathering with LLMs and Bayesian Experimental Designcited by 24
  10. 2025Active Task Disambiguation with LLMscited by 23
  11. 2014A Method for Stopping Active Learning Based on Stabilizing Predictions and the Need for User-Adjustable Stoppingcited by 21
  12. 2024Contextual Bilevel Reinforcement Learning for Incentive Alignmentcited by 14
  13. 2021Understanding the Origin of Information-Seeking Exploration in Probabilistic Objectives for Controlcited by 10
  14. 2021Applications of the Free Energy Principle to Machine Learning and Neurosciencecited by 9
  15. 2024Lifting the Veil on Composition, Risks, and Mitigations of the Large Language Model Supply Chaincited by 4
  16. 2025Bi-Level Policy Optimization with Nystr"om Hypergradientscited by 3
  17. 2020A Meta-Bayesian Model of Intentional Visual Searchcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026BALAR : A Bayesian Agentic Loop for Active Reasoning
  2. 2024Active Inference for Self-Organizing Multi-LLM Systems: A Bayesian Thermodynamic Approach to Adaptation
  3. 2026The Context Gathering Decision Process: A POMDP Framework for Agentic Search
  4. 2026ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL
  5. 2024Demonstrating the Continual Learning Capabilities and Practical Application of Discrete-Time Active Inference
  6. 2026ACCORD: Action-Conditioned Contextual Grounding for Language Agents
  7. 2026AIPO: Learning to Reason from Active Interaction
  8. 2026Evaluating Interactive Reasoning in Large Language Models: A Hierarchical Benchmark with Executable Games