Biased AI improves human decision-making but reduces trust

TL;DR

Biased AI systems, while reducing trust, can enhance human decision-making and engagement by providing culturally diverse perspectives, challenging the notion that AI should always be ideologically neutral.

Contribution

This study demonstrates that culturally biased AI can improve decision quality and engagement, revealing potential benefits of strategic bias in AI design.

Findings

Partisan AI improves human decision performance

Biased AI increases user engagement and reduces evaluative bias

Exposure to opposing biases narrows perception-performance gap

Abstract

Current AI systems minimize risk by enforcing ideological neutrality, yet this may introduce automation bias by suppressing cognitive engagement in human decision-making. We conducted randomized trials with 2,500 participants to test whether culturally biased AI enhances human decision-making. Participants interacted with politically diverse GPT-4o variants on information evaluation tasks. Partisan AI assistants enhanced human performance, increased engagement, and reduced evaluative bias compared to non-biased counterparts, with amplified benefits when participants encountered opposing views. These gains carried a trust penalty: participants underappreciated biased AI and overcredited neutral systems. Exposing participants to two AIs whose biases flanked human perspectives closed the perception-performance gap. These findings complicate conventional wisdom about AI neutrality, suggesting that strategic integration of diverse cultural biases may foster improved and resilient human decision-making.