A new study published in Nature Health reveals that patients provide less detailed symptom information when interacting with artificial intelligence compared to human medical professionals, a phenomenon that could undermine the accuracy of digital diagnoses. The research, led by Professor Wilfried Kunde and Moritz Reis from the University of Würzburg, in collaboration with experts from Charité Berlin, the University of Cambridge, and Berlin hospitals, highlights a critical psychological barrier to the integration of AI in healthcare. As digitalization accelerates in the medical sector, AI chatbots and digital symptom checkers are increasingly becoming the first point of contact for self-triage, where patients assess their own urgency for treatment. While the technical capabilities of these systems continue to grow, the study suggests that their effectiveness relies heavily on the quality of information provided by users. The research involved 500 participants who wrote simulated symptom reports for common conditions like headaches and flu. Participants were led to believe their reports would be reviewed either by an AI or a human doctor. The findings showed a measurable decline in the suitability of descriptions when participants thought they were communicating with a machine. Reports intended for an AI averaged 228.7 characters, whereas those for a human doctor averaged 255.6 characters. Although the difference in character count may seem minor, the research team emphasizes its practical relevance. In medical diagnostics, even high-performance AI models cannot provide accurate advice if essential details are missing. Consequently, the success of digital initial assessments depends less on computational power and more on the patient's willingness to provide comprehensive information. The primary reason for this reluctance appears to be a psychological phenomenon known as uniqueness neglect. Professor Kunde explains that many people assume AI cannot grasp the individual nuances of their situation and will merely match standardized patterns. This belief, combined with skepticism about algorithmic diagnostic capabilities and privacy concerns, leads users to unconsciously withhold specific details. Reis notes that if patients do not trust a machine to understand their unique circumstances, they may not provide the data necessary for precise assistance, effectively lowering the quality of the diagnosis before the system even begins processing the case. The study concludes that technical advancement alone is insufficient to ensure the safe and effective use of medical AI. To address these human factors, the researchers advocate for the intelligent design of user interfaces. Developers should implement features that encourage detailed reporting, such as providing concrete examples of high-quality descriptions and programming the AI to actively request missing information. By fostering a more detailed dialogue, the healthcare system can reduce the risk of misdiagnosis and mitigate the burden on medical professionals. Without addressing the psychological gap between human and machine interaction, the potential of AI in reshaping digital diagnosis remains limited.