Want to use AI to improve your work? Have it disagree with you

A study of 45 knowledge workers at a U.S. public university found that people who deliberately prompted AI tools to challenge or contradict their ideas produced better work than those who simply accepted the AI's confident answers. Researchers call the passive pattern cognitive offloading, warning it can erode the critical thinking skills that make knowledge workers valuable in the first place.

Examples from the study include a marketing professional who had AI generate skeptical customer personas to surface overlooked preferences, and an attorney who used AI to hunt for legal loopholes companies might exploit, revealing realistic scenarios of hard-to-detect misconduct. The researchers suggest organizations build training programs and safe experimentation spaces to help workers develop these friction-generating habits rather than defaulting to whatever the AI says first.

The full dispatch is available from the source below.