Learning How to Vote with Principles: Axiomatic Insights Into the Collective Decisions of Neural Networks

TL;DR

This paper introduces axiomatic deep voting, a framework for neural networks to learn and evaluate collective decision-making rules based on voting axioms, revealing insights into AI alignment and voting rule diversity.

Contribution

It presents a novel framework combining neural networks with voting axioms, and demonstrates how optimizing for axioms can generate new voting rules beyond traditional methods.

Findings

Neural networks often do not satisfy core voting axioms despite high accuracy.

Training with axiom-specific data does not improve axiom alignment.

Optimizing solely for axioms enables creation of new, superior voting rules.

Abstract

Can neural networks be applied in voting theory, while satisfying the need for transparency in collective decisions? We propose axiomatic deep voting: a framework to build and evaluate neural networks that aggregate preferences, using the well-established axiomatic method of voting theory. Our findings are: (1) Neural networks, despite being highly accurate, often fail to align with the core axioms of voting rules, revealing a disconnect between mimicking outcomes and reasoning. (2) Training with axiom-specific data does not enhance alignment with those axioms. (3) By solely optimizing axiom satisfaction, neural networks can synthesize new voting rules that often surpass and substantially differ from existing ones. This offers insights for both fields: For AI, important concepts like bias and value-alignment are studied in a mathematically rigorous way; for voting theory, new areas of the space of voting rules are explored.