Measuring Error Alignment for Decision-Making Systems

TL;DR

This paper introduces two new behavioral metrics to assess how well AI systems align with human decision-making, aiming to improve trustworthiness without costly internal state analysis.

Contribution

It proposes novel, cost-effective behavioral alignment metrics that correlate with internal representational similarity, advancing methods for AI value alignment.

Findings

Metrics correlate with representational alignment measures

Metrics are effective across multiple domains

Provide a new approach for value alignment

Abstract

Given that AI systems are set to play a pivotal role in future decision-making processes, their trustworthiness and reliability are of critical concern. Due to their scale and complexity, modern AI systems resist direct interpretation, and alternative ways are needed to establish trust in those systems, and determine how well they align with human values. We argue that good measures of the information processing similarities between AI and humans, may be able to achieve these same ends. While Representational alignment (RA) approaches measure similarity between the internal states of two systems, the associated data can be expensive and difficult to collect for human systems. In contrast, Behavioural alignment (BA) comparisons are cheaper and easier, but questions remain as to their sensitivity and reliability. We propose two new behavioural alignment metrics misclassification agreement which measures the similarity between the errors of two systems on the same instances, and class-level error similarity which measures the similarity between the error distributions of two systems. We show that our metrics correlate well with RA metrics, and provide complementary information to another BA metric, within a range of domains, and set the scene for a new approach to value alignment.