Are humans losing the ability to think for themselves?
April 8, 2026·7 min
Host Megan McCarty Carino opens by framing the episode around growing concern that reliance on AI tools may be eroding human thinking. She introduces Steve Shaw, a postdoctoral researcher at the Wharton School of Business at the University of Pennsylvania, who co-authored a study on what they call cognitive surrender. Shaw explains that the classic behavioral economics framework of thinking fast and slow — intuitive versus deliberative cognition — is no longer sufficient to describe how humans make decisions. He proposes a third system, System 3, which he defines as artificial cognition or AI, representing a new mode through which people can outsource their thinking entirely. The study was designed to test this hypothesis in a controlled lab setting. Participants completed logic and reasoning questions, with some given optional access to ChatGPT. Crucially, the researchers secretly manipulated ChatGPT's accuracy on the back end so participants were unaware whether the AI was giving correct or incorrect answers. Results showed that participants largely adopted the AI's answers regardless of accuracy, demonstrating cognitive surrender. In follow-up studies, researchers introduced situational moderators including time pressure — giving participants only 30 seconds per question — and financial incentives for correct answers. Both conditions increased the rate at which participants overrode the AI, but not enough to restore performance to the level of participants who had no AI access at all. Shaw interprets this as evidence that cognitive surrender is a robust psychological mechanism that persists even under meaningful stakes. He discusses implications for education, warning that students who defer the learning process itself to AI may never acquire critical thinking skills. He also raises concerns about de-skilling in the workplace as employees increasingly rely on AI for cognitive tasks. Shaw addresses the emerging trend of agentic AI — autonomous systems that complete tasks with minimal human oversight — and suggests the same principles of cognitive surrender apply at an even larger scale. He calls for policy consideration before society moves further toward automating cognitive life. When asked about his own AI use, Shaw acknowledges he uses AI daily and allows his Wharton students to use it for all assignments. However, he advises students to think first and generate their own ideas before consulting AI. Since completing the research, he says he has become more intentional, sometimes going fully offline to preserve independent thinking.
