Harvard DSR
Social Sciences
How humans judge AI's advice—and why we're terrible at knowing when to trust it
When an AI suggests something, we often look at it the way a jury looks at an expert witness: we trust the confident tone more than the actual accuracy.
This means as we integrate AI into workplaces and hospitals, we need to design interfaces that help us spot bad advice, because our instincts are unreliable judges.
Bug reported: No