Saved in:
Bibliographic Details
Main Authors: Alon, Nitay, Barnby, Joseph M., Sarkadi, Stefan, Schulz, Lion, Rosenschein, Jeffrey S., Dayan, Peter
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2405.01870
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908863531319296
author Alon, Nitay
Barnby, Joseph M.
Sarkadi, Stefan
Schulz, Lion
Rosenschein, Jeffrey S.
Dayan, Peter
author_facet Alon, Nitay
Barnby, Joseph M.
Sarkadi, Stefan
Schulz, Lion
Rosenschein, Jeffrey S.
Dayan, Peter
contents Social agents with finitely nested opponent models are vulnerable to manipulation by agents with deeper recursive capabilities. This imbalance, rooted in logic and the theory of recursive modelling frameworks, cannot be solved directly. We propose a computational framework called $\aleph$-IPOMDP, which augments the Bayesian inference of model-based RL agents with an anomaly detection algorithm and an out-of-belief policy. Our mechanism allows agents to realize that they are being deceived, even if they cannot understand how, and to deter opponents via a credible threat. We test this framework in both a mixed-motive and a zero-sum game. Our results demonstrate the $\aleph$-mechanism's effectiveness, leading to more equitable outcomes and less exploitation by more sophisticated agents. We discuss implications for AI safety, cybersecurity, cognitive science, and psychiatry.
format Preprint
id arxiv_https___arxiv_org_abs_2405_01870
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle $\aleph$-IPOMDP: Mitigating Deception in a Cognitive Hierarchy with Off-Policy Counterfactual Anomaly Detection
Alon, Nitay
Barnby, Joseph M.
Sarkadi, Stefan
Schulz, Lion
Rosenschein, Jeffrey S.
Dayan, Peter
Multiagent Systems
Computer Science and Game Theory
Social agents with finitely nested opponent models are vulnerable to manipulation by agents with deeper recursive capabilities. This imbalance, rooted in logic and the theory of recursive modelling frameworks, cannot be solved directly. We propose a computational framework called $\aleph$-IPOMDP, which augments the Bayesian inference of model-based RL agents with an anomaly detection algorithm and an out-of-belief policy. Our mechanism allows agents to realize that they are being deceived, even if they cannot understand how, and to deter opponents via a credible threat. We test this framework in both a mixed-motive and a zero-sum game. Our results demonstrate the $\aleph$-mechanism's effectiveness, leading to more equitable outcomes and less exploitation by more sophisticated agents. We discuss implications for AI safety, cybersecurity, cognitive science, and psychiatry.
title $\aleph$-IPOMDP: Mitigating Deception in a Cognitive Hierarchy with Off-Policy Counterfactual Anomaly Detection
topic Multiagent Systems
Computer Science and Game Theory
url https://arxiv.org/abs/2405.01870