AI-driven bias hunters have discovered a new psychological flaw where humans systematically ignore useful information as a situation gets more complex.
April 24, 2026
Original Paper
AI-assisted discovery of unknown cognitive biases: Humans systematically undervalue information under complexity
PsyArXiv · vabzc_v1
AI-generated illustration
The Takeaway
Humans undervalue important data specifically when a situation becomes complicated, a newly identified bias found by an AI-driven framework. Previously, psychologists had to guess which biases might exist before testing for them in a lab. This study used an AI to find patterns where human behavior deviated from a mathematically perfect benchmark. The results show that instead of seeking more clarity when faced with complexity, we actually stop valuing the information that could help us. This discovery demonstrates how AI can be used to map the limits of the human mind in ways traditional research missed.
From the abstract
Biases in how people seek and integrate information can lead to negative consequences across domains ranging from health to finance. It is thus crucial to identify common biases. Traditionally, researchers generate hypotheses regarding potential biases and then empirically test them. While successful, important biases may go undetected as researchers have limited cognitive resources and can be blind to their own biases. To overcome this, we developed a novel data-driven framework for bias discov