New Study Reveals People Accept Wrong AI Answers 73% of the Time in What Researchers Call 'Cognitive Surrender'

Apr 05, 2026
Ars Technica
Article image for New Study Reveals People Accept Wrong AI Answers 73% of the Time in What Researchers Call 'Cognitive Surrender'

Summary

A groundbreaking new study finds people accept wrong AI-generated answers 73% of the time, with researchers warning that 'cognitive surrender' is causing humans to abandon critical thinking and making our reasoning quality entirely dependent on AI accuracy.

Key Points

  • A new psychological framework called 'cognitive surrender' describes how AI users increasingly abandon critical thinking and accept AI-generated answers without scrutiny, even when those answers are wrong.
  • Across 1,372 participants and over 9,500 trials, subjects accept faulty AI reasoning a staggering 73.2% of the time, with time pressure worsening blind acceptance and financial incentives slightly improving users' willingness to question AI responses.
  • Higher fluid IQ reduces susceptibility to AI misinformation, while strong pre-existing trust in AI increases it, and researchers warn that cognitive surrender means human reasoning quality becomes entirely dependent on the quality of the AI being used.

Tags

Read Original Article