AI recommendations change how humans decide

2 min read · January 30, 2025
New Power Labs

Research by Kuhl and Bush (2025) shows that biased AI recommendations don’t just skew decisions, they change how people make decisions themselves. 

In a controlled experiment simulating hiring decisions, participants followed biased algorithmic recommendations about 70% of the time, even when candidate qualifications were equal. More importantly, when participants went back to deciding without the AI, those exposed to biased recommendations continued to echo the same bias unless the system had first provided counterfactual explanations. This means that, rather than merely automating bias, AI systems can train human decision-makers to internalize it.

The tools used in hiring, lending, admissions, and screening don’t just influence outcomes, they reshape how people judge merit over time. As biased systems condition decision-makers’ internal criteria, inequity becomes harder to detect and correct. 

In an AI-driven economy, decision design quietly shapes who and what gets ahead.

Narinder

New Power Labs

Like what you’re reading? Subscribe to get weekly Equity Shots in your inbox.