Science

Study reveals high trust in AI leads to cognitive surrender

Researchers from the University of Pennsylvania found that 73.2% of 1,372 participants accepted faulty AI reasoning without question, a phenomenon termed 'cognitive surrender.' The study revealed that individuals with high trust in AI were more likely to be misled, while those with higher 'Fluid IQ' were less susceptible to erroneous AI outputs. The findings suggest that as reliance on AI increases, decision-making quality is directly tied to the accuracy of the AI system.

ERROR project shifts from cash rewards to publication incentives

The ERROR project, which aimed to incentivize scientists by offering cash for identifying errors in published papers, has been deemed ineffective and is now transitioning to a new model. Instead of cash rewards, the project will launch a journal that offers published papers as incentives for participation.