• •TNW reports a study finding participants using AI advice were three times less accurate than manual counterparts.
• •The same study observed that AI-assisted users reported significantly higher confidence levels, effectively doubling their self-assurance.
• •It remains unclear whether this 'automation bias' stems from an inherent trust in model outputs or a lack of mental effort during verification.

A study cited by The Next Web reveals that reliance on AI advice leads to a threefold decrease in user accuracy, even as it doubles their self-reported confidence. This phenomenon mirrors 'automation bias' documented in earlier research regarding GPS and autopilot systems, where human oversight diminishes when users trust a machine's output. However, the study leaves a critical question unanswered: whether the confidence boost causes the errors or if the AI output simply blinds users to their own critical thinking gaps. Whether this trend persists across different industries, such as medicine or law, remains the next major milestone for human-AI interaction research.