Remind people that humans suck, and suddenly AI looks fair.

A study from the University of Exeter found something every AI sales team already knows intuitively. When you remind people that human decision-making is biased, they suddenly rate AI as more fair and impartial.

Not because anyone tested the AI. Not because it improved. Just because the comparison shifted.

They had people evaluate discrimination risk in public sector hiring. Half evaluated AI first, then humans. The other half - humans first, then AI. The order changed everything. Think about human bias first, and AI looks reasonable by comparison.

This is literally the contrast effect wearing a lab coat. Your brain doesn't evaluate things in absolute terms, it evaluates them relative to whatever it just processed. Show me something broken, and the next thing looks functional by default.

Every company selling an AI hiring tool already knows this playbook. You don't need to prove your algorithm is fair. You just need to remind the buyer that the alternative is a hiring manager who went to the same university as her favorite candidates. The framing does the work. Not the evidence.

The researchers even warned about this, that debates about AI might shift toward emphasizing human weaknesses instead of actually demonstrating that the AI system works fairly. Which, if you think about it, is already happening.

I'm not saying human decision-making is great. It's really not. But "at least the robot is consistent" is not the same as "the robot is fair." Consistency just means the same mistake, at scale, every single time.

#mind