Saved in:
Bibliographic Details
Main Authors: MacDermott, Matt, Fox, James, Belardinelli, Francesco, Everitt, Tom
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2412.04758
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915050399203328
author MacDermott, Matt
Fox, James
Belardinelli, Francesco
Everitt, Tom
author_facet MacDermott, Matt
Fox, James
Belardinelli, Francesco
Everitt, Tom
contents We define maximum entropy goal-directedness (MEG), a formal measure of goal-directedness in causal models and Markov decision processes, and give algorithms for computing it. Measuring goal-directedness is important, as it is a critical element of many concerns about harm from AI. It is also of philosophical interest, as goal-directedness is a key aspect of agency. MEG is based on an adaptation of the maximum causal entropy framework used in inverse reinforcement learning. It can measure goal-directedness with respect to a known utility function, a hypothesis class of utility functions, or a set of random variables. We prove that MEG satisfies several desiderata and demonstrate our algorithms with small-scale experiments.
format Preprint
id arxiv_https___arxiv_org_abs_2412_04758
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Measuring Goal-Directedness
MacDermott, Matt
Fox, James
Belardinelli, Francesco
Everitt, Tom
Artificial Intelligence
Machine Learning
We define maximum entropy goal-directedness (MEG), a formal measure of goal-directedness in causal models and Markov decision processes, and give algorithms for computing it. Measuring goal-directedness is important, as it is a critical element of many concerns about harm from AI. It is also of philosophical interest, as goal-directedness is a key aspect of agency. MEG is based on an adaptation of the maximum causal entropy framework used in inverse reinforcement learning. It can measure goal-directedness with respect to a known utility function, a hypothesis class of utility functions, or a set of random variables. We prove that MEG satisfies several desiderata and demonstrate our algorithms with small-scale experiments.
title Measuring Goal-Directedness
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2412.04758