Pilots' experience offers lessons on avoiding AI deskilling

Pilots' reliance on autopilot has long been a cautionary tale for workers increasingly dependent on AI, according to Steven Shaw, an assistant professor at King's College London. Shaw, coauthor of a study on 'cognitive surrender,' drew parallels between pilots' use of automation and the outsourcing of cognitive tasks to AI. Federal aviation authorities have long warned about the risks of continuous autopilot use, with a 2013 FAA alert highlighting the potential for skill degradation. This concern was reiterated in a 2017 alert, urging pilots to practice manual flying periodically. Shaw warns that overreliance on AI could lead to cognitive atrophy, a risk highlighted by OpenAI's Sam Altman and Mistral's Arthur Mensch. Research shows people often accept AI answers without critical thinking, with experiments indicating 80% acceptance of incorrect AI recommendations. Shaw advises developing personal judgment before seeking AI assistance, a sentiment echoed by experts like Geetha Rajan and Ryan Deiss, who advocate for a '10-80-10' approach to AI integration. Rep. Alexandria Ocasio-Cortez also warned at a town hall that AI could diminish human thought, aligning with concerns about deskilling. Vivienne Ming's experiment found that 90% to 95% of participants relied on AI for answers or validation, rather than deepening their thinking. Shaw's research underscores the need for critical engagement with AI, emphasizing the importance of maintaining independent cognitive skills to avoid 'cognitive surrender.'