# Formalizing the Problem of Side Effect Regularization

arXiv: https://arxiv.org/abs/2206.11812 · HTML: https://mlsanity.com/paper/2206.11812/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 23 references that are in the map.
- First posted on arXiv 23 Jun 2022.

## Builds on (most cited first, 11 of 11)

- [Constrained Policy Optimization](https://mlsanity.com/paper/1705.10528/) (2017)
- [Cooperative Inverse Reinforcement Learning](https://mlsanity.com/paper/1606.03137/) (2016)
- [Optimal Policies Tend to Seek Power](https://mlsanity.com/paper/1912.01683/) (2019)
- [AI Safety Gridworlds](https://mlsanity.com/paper/1711.09883/) (2017)
- [Leave no Trace: Learning to Reset for Safe and Autonomous Reinforcement Learning](https://mlsanity.com/paper/1711.06782/) (2017)
- [Inverse Constrained Reinforcement Learning](https://mlsanity.com/paper/2011.09999/) (2020)
- [Penalizing side effects using stepwise relative reachability](https://mlsanity.com/paper/1806.01186/) (2018)
- [Conservative Agency via Attainable Utility Preservation](https://mlsanity.com/paper/1902.09725/) (2019)
- [Avoiding Side Effects By Considering Future Tasks](https://mlsanity.com/paper/2010.07877/) (2020)
- [Avoiding Side Effects in Complex Environments](https://mlsanity.com/paper/2006.06547/) (2020)
- [SafeLife 1.0: Exploring Side Effects in Complex Environments](https://mlsanity.com/paper/1912.01217/) (2019)

## Built on by (most cited first, 3 of 3)

- [Tool Learning with Foundation Models](https://mlsanity.com/paper/2304.08354/) (2023)
- [Bridging the Gap: A Survey on Integrating (Human) Feedback for Natural Language Generation](https://mlsanity.com/paper/2305.00955/) (2023)
- [Large Language Models Miss the Multi-Agent Mark](https://mlsanity.com/paper/2505.21298/) (2025)

## Abstract (arXiv)

AI objectives are often hard to specify properly. Some approaches tackle this problem by regularizing the AI's side effects: Agents must weigh off "how much of a mess they make" with an imperfectly specified proxy objective. We propose a formal criterion for side effect regularization via the assistance game framework. In these games, the agent solves a partially observable Markov decision process (POMDP) representing its uncertainty about the objective function it should optimize. We consider the setting where the true objective is revealed to the agent at a later time step. We show that this POMDP is solved by trading off the proxy reward with the agent's ability to achieve a range of future tasks. We empirically demonstrate the reasonableness of our problem formalization via ground-truth evaluation in two gridworld environments.