Many people will be familiar with two modes of thinking – fast and slow. Fast thinking is intuitive, whereas slow thinking is deliberate reasoning  

Researchers are proposing a third mode, one that they’ve called “cognitive surrender” – the tendency to accept AI-generated answers without critically evaluating them, which is effectively farming-out decision making to AI.  

They gave participants a variety of different problem-solving tasks and allowed some of them to access AI, while others could only rely on their own fast and slow modes of thinking.  

They found that:  

• Participants often followed AI advice even when it was incorrect. When AI was right, performance improved substantially; when AI was wrong, performance dropped below the level of those working without AI assistance.  
• One of the most striking findings was that participants became more confident after consulting AI, regardless of whether the advice was correct or incorrect.  
• People who had higher levels of trust in AI were more likely to follow incorrect AI recommendations and less likely to question them.  
• People with stronger analytical thinking skills were more likely to identify faulty AI advice and override incorrect recommendations.  
• Incentives for accuracy reduced cognitive surrender by encouraging people to question and verify AI-generated responses.  
• The findings suggest the greatest risk is not AI making mistakes, but people becoming less likely to recognise and challenge those mistakes.  
• For organisations, there is potential for a gradual erosion of independent thinking if employees routinely defer judgement to AI, particularly in roles that rely on professional expertise, critical thinking and accountability.  

It’s interesting to read this alongside other research which has shown that more experienced workers use AI better than newer workers, as those with more experience had a better understanding of their jobs and could spot when AI was making mistakes or offering unhelpful suggestions. 

LISTEN ABOVE