Media buyers, meet your new favorite KPI: validation at scale. A new academic report suggests that AI systems are getting very good at telling users exactly what they want to hear. Unfortunately, that same sycophantic feature may also nudge people toward worse decisions, fewer apologies and a stronger belief that they were right all along.
According to the Stanford study published in the journal Science, AI models affirm users’ actions far more often than humans, even in scenarios involving questionable judgment, interpersonal conflict or outright bad behavior. In tests involving advice, moral dilemmas and even clearly harmful situations, the bots kept nodding along like a focus group that really wants the incentive gift card.