How patient-operated AI tools are reshaping patient perceptions and the patient-clinician relationship: A narrative review

Objectives To examine how patient-operated AI tools influence patient perceptions and the patient-clinician relationship, including their role in supporting health literacy and the risks of misinformation, algorithmic bias and cyberchondria. Methods This narrative review synthesises published literature examining how patients use AI tools to enter symptoms, obtain assessments and judge the urgency of seeking care, and how these tools shape communication, trust and interaction during clinical encounters. Findings were organised around empowerment, anxiety, safety, bias, demographic differences and clinical workload. Author-developed illustrative examples were included to support the narrative synthesis and were distinguished from empirical findings. Results Patients value the convenience, immediacy and sense of control provided by AI tools. However, concerns about accuracy, inconsistent triage, hallucinations, privacy and the potential for increased health anxiety are common. Cyberchondria is reported when AI-generated information amplifies worry or leads to unnecessary escalation of care. Trust in clinicians remains central, and disclosure that AI contributed to a diagnosis can reduce confidence in medical advice. Clinicians report challenges in correcting misinformation, preserving empathy and managing increased workload. Engagement varies by age, education, digital literacy and cultural context, with lower literacy and marginalised groups more vulnerable to misunderstanding or inequitable AI performance. Conclusions Patient-operated AI tools are reshaping early help-seeking and expectations during consultations. They can improve preparedness and health literacy but also introduce new pressures on trust, safety and communication. Effective integration requires transparency and continued emphasis on the human elements of care. Practice implications Clinicians should ask whether patients have used AI tools, encourage them to share the advice received and explain how AI outputs differ from clinical assessment. They should correct unsafe or inaccurate information and reinforce shared decision-making. Services adopting these AI tools should support clinicians with guidance on communication, privacy, bias and digital literacy to maintain trust and person-centred care.