# Reinforcement Learning from LLM Feedback to Counteract Goal Misgeneralization

arXiv: https://arxiv.org/abs/2401.07181 · HTML: https://mlsanity.com/paper/2401.07181/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Process reward models for reasoning, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 17 of its 30 references that are in the map.
- First posted on arXiv 14 Jan 2024.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Process reward models for reasoning](https://mlsanity.com/alignment/process-reward-models-for-reasoning/)

## Builds on (most cited first, 17 of 17)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures](https://mlsanity.com/paper/1802.01561/) (2018)
- [Leveraging Procedural Generation to Benchmark Reinforcement Learning](https://mlsanity.com/paper/1912.01588/) (2019)
- [Artificial Intelligence, Values and Alignment](https://mlsanity.com/paper/2001.09768/) (2020)
- [Scalable agent alignment via reward modeling: a research direction](https://mlsanity.com/paper/1811.07871/) (2018)
- [AI safety via debate](https://mlsanity.com/paper/1805.00899/) (2018)
- [The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models](https://mlsanity.com/paper/2201.03544/) (2022)
- [Reward learning from human preferences and demonstrations in Atari](https://mlsanity.com/paper/1811.06521/) (2018)
- [Large Language Models Sensitivity to The Order of Options in Multiple-Choice Questions](https://mlsanity.com/paper/2308.11483/) (2023)
- [Constructing Unrestricted Adversarial Examples with Generative Models](https://mlsanity.com/paper/1805.07894/) (2018)
- [Harms from Increasingly Agentic Algorithmic Systems](https://mlsanity.com/paper/2302.10329/) (2023)
- [Reinforcement Learning with a Corrupted Reward Channel](https://mlsanity.com/paper/1705.08417/) (2017)
- [Ethical-Advice Taker: Do Language Models Understand Natural Language Interventions?](https://mlsanity.com/paper/2106.01465/) (2021)
- [An overview of 11 proposals for building safe advanced AI](https://mlsanity.com/paper/2012.07532/) (2020)

## Built on by (most cited first, 5 of 5)

- [Puzzle Solving using Reasoning of Large Language Models: A Survey](https://mlsanity.com/paper/2402.11291/) (2024)
- [Large language models for artificial general intelligence (AGI): A survey of foundational principles and approaches](https://mlsanity.com/paper/2501.03151/) (2025)
- [A Survey On Enhancing Reinforcement Learning in Complex Environments: Insights from Human and LLM Feedback](https://mlsanity.com/paper/2411.13410/) (2024)
- [The Narrative Continuity Test: A Conceptual Framework for Evaluating Identity Persistence in AI Systems](https://mlsanity.com/paper/2510.24831/) (2025)
- [EAPO: Enhancing Policy Optimization with On-Demand Expert Assistance](https://mlsanity.com/paper/2509.23730/) (2025)

## Abstract (arXiv)

We introduce a method to address goal misgeneralization in reinforcement learning (RL), leveraging Large Language Model (LLM) feedback during training. Goal misgeneralization, a type of robustness failure in RL occurs when an agent retains its capabilities out-of-distribution yet pursues a proxy rather than the intended one. Our approach utilizes LLMs to analyze an RL agent's policies during training and identify potential failure scenarios. The RL agent is then deployed in these scenarios, and a reward model is learnt through the LLM preferences and feedback. This LLM-informed reward model is used to further train the RL agent on the original dataset. We apply our method to a maze navigation task, and show marked improvements in goal generalization, especially in cases where true and proxy goals are somewhat distinguishable and behavioral biases are pronounced. This study demonstrates how the LLM, despite its lack of task proficiency, can efficiently supervise RL agents, providing scalable oversight and valuable insights for enhancing goal-directed learning in RL through the use of LLMs.