mlsanityData to 14 Oct 2026

Scaling Empowered Agents

arXiv 2610.13002 ↗First posted 9 Oct 2026cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Empowerment measures the capacity of an agent to actively control its environment. In this work, we explain the connection between skill learning and empowerment. While skill learning is not empowerment, we introduce a simple scalable recipe for empowerment maximization: empowerment corresponds to the number of executable skills from a state, so maximizing empowerment corresponds to reaching states that are good starting points for many skills. Our recipe applies to arbitrary underlying skill learning methods. Whereas most prior work on empowerment looks at tabular settings, this recipe enables us to study empowerment at a scale where we can realize theoretically-predicted but empirically-elusive properties of empowerment (e.g., gathering resources, environment modification, survival). Experiments show that empowerment maximization unlocks interesting emergent properties, such as building a house in Crafter. Website and code at hrefhttps://empowerment-geometry.github.io/scaling-empowerment/this https url.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 30 of its 43 references that are in the map. First posted on arXiv 9 Oct 2026.

Builds on

30 of 43 references.

References · 30

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2018Representation Learning with Contrastive Predictive Codingcited by 7,777
  3. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,885
  4. 2017Curiosity-driven Exploration by Self-supervised Predictioncited by 1,194
  5. 2017Hindsight Experience Replaycited by 1,136
  6. 2018Exploration by Random Network Distillationcited by 993
  7. 2018Diversity is All You Need: Learning Skills without a Reward Functioncited by 752
  8. 2016Unifying Count-Based Exploration and Intrinsic Motivationcited by 639
  9. 2019Dynamics-Aware Unsupervised Discovery of Skillscited by 313
  10. 2016Variational Intrinsic Controlcited by 267
  11. 2024OGBench: Benchmarking Offline Goal-Conditioned RLcited by 238
  12. 2015Variational Information Maximisation for Intrinsically Motivated Reinforcement Learningcited by 144
18 more
  1. 2018Noise Contrastive Estimation and Negative Sampling for Conditional Models: Consistency and Statistical Efficiencycited by 105
  2. 2024Craftax: A Lightning-Fast Benchmark for Open-Ended Reinforcement Learningcited by 88
  3. 2023METRA: Scalable Unsupervised RL with Metric-Aware Abstractioncited by 86
  4. 2024Simplifying Deep Temporal Difference Learningcited by 78
  5. 2013Empowerment -- an Introductioncited by 68
  6. 2024Can a MISL Fly? Analysis and Ingredients for Mutual Information Skill Learningcited by 21
  7. 2017Unsupervised Real-Time Control through Variational Empowermentcited by 18
  8. 2025Plasticity as the Mirror of Empowermentcited by 12
  9. 2014Changing the Environment Based on Empowerment as Intrinsic Motivationcited by 10
  10. 2020Efficient Empowerment Estimation for Unsupervised Stabilizationcited by 9
  11. 2025Intrinsically-Motivated Humans and Agents in Open-World Explorationcited by 8
  12. 2025BuilderBench: The Building Blocks of Intelligent Agentscited by 6
  13. 2025Estimating the Empowerment of Language Model Agentscited by 3
  14. 2025Empowerment Gain and Causal Model Construction: Children and adults are sensitive to controllability and variability in their causal interventionscited by 3
  15. 2026Temporal Representations for Exploration: Learning Complex Exploratory Behavior without Extrinsic Rewardscited by 2
  16. 2026Learning Generalizable Skill Policy with Data-Efficient Unsupervised RLcited by 1
  17. 2026What Is Worth Representing? Representational Empowerment for Continual Model Constructioncited by 1
  18. 2026The Geometry of Empowermentcited by 1

Built on by

Not cited yet.