Apolitical Intelligence? Auditing Delphi's responses on controversial political issues in the US

TL;DR

This paper audits the political neutrality of the Delphi language model on controversial US issues, revealing significant bias and calibration issues, and discusses the implications for societal roles of AI.

Contribution

It provides a novel analysis of Delphi's responses across political groups, highlighting biases and challenging notions of neutrality in AI models.

Findings

Delphi shows political skew in responses.

Calibration of confidence levels is poor.

Bias varies across political subgroups.

Abstract

As generative language models are deployed in ever-wider contexts, concerns about their political values have come to the forefront with critique from all parts of the political spectrum that the models are biased and lack neutrality. However, the question of what neutrality is and whether it is desirable remains underexplored. In this paper, I examine neutrality through an audit of Delphi [arXiv:2110.07574], a large language model designed for crowdsourced ethics. I analyse how Delphi responds to politically controversial questions compared to different US political subgroups. I find that Delphi is poorly calibrated with respect to confidence and exhibits a significant political skew. Based on these results, I examine the question of neutrality from a data-feminist lens, in terms of how notions of neutrality shift power and further marginalise unheard voices. These findings can hopefully contribute to a more reflexive debate about the normative questions of alignment and what role we want generative models to play in society.