Alarming Study Finds That Most People Just Do What ChatGPT Tells Them, Even If It’s Totally Wrong

A study by UPenn researchers examined whether people accept ChatGPT answers even when they are wrong.

In experiments where ChatGPT use was optional, over half of participants chose to use the chatbot.

In one experiment (n=359) participants followed correct AI advice 92.7% of the time and faulty advice 79.8% of the time.

Researchers described this willingness to accept AI output as 'cognitive surrender' and linked it to reduced critical checking.