"Cognitive surrender" leads AI users to abandon logical thinking, research finds
Ars Technica
Research shows AI users often accept faulty AI reasoning, exhibiting "cognitive surrender" and reduced critical thinking.Summary
A study involving 1,372 participants found that individuals accepted faulty AI reasoning 73.2% of the time, demonstrating a phenomenon called "cognitive surrender." This occurs because fluent and confident AI outputs are treated as authoritative, lowering scrutiny and reducing the inclination for deliberation. Individuals with high trust in AI were more susceptible to errors, while those with high fluid IQ were less likely to be misled. The researchers note that while "cognitive surrender" isn't inherently irrational, especially with statistically superior AI systems, it creates a vulnerability where human reasoning quality becomes dependent on the AI's accuracy.
(Source：Ars Technica)