How good are ‘AI doctors’ — and will they take over medicine?

Nature spoke with medical specialists in early June 2026 about the burgeoning capabilities of artificial intelligence (AI) in diagnosing medical conditions and its potential to reshape the future of healthcare. The consensus among experts is that while AI systems are rapidly advancing and demonstrating impressive diagnostic accuracy in specific areas, they are not yet poised to entirely replace human physicians. Instead, AI is increasingly viewed as a powerful tool to augment clinical decision-making, improve efficiency, and potentially expand access to care, particularly in underserved regions.

These AI systems, often trained on vast datasets of medical images, patient records, and research literature, are showing proficiency in tasks such as identifying cancerous tumors in scans, detecting diabetic retinopathy, and analyzing electrocardiograms. For instance, some AI algorithms have achieved diagnostic performance comparable to or even exceeding that of experienced radiologists and ophthalmologists in controlled studies. The appeal of AI lies in its potential to process information at speeds and scales beyond human capacity, reducing diagnostic errors, and freeing up clinicians' time for more complex patient interactions and personalized care.

However, significant challenges remain before AI can be widely integrated into routine clinical practice. Ethical considerations, including data privacy, algorithmic bias, and accountability for diagnostic errors, are paramount. Furthermore, the "black box" nature of some AI models, where the reasoning behind a diagnosis is not fully transparent, poses a hurdle for clinician trust and regulatory approval. The integration of AI also necessitates substantial investment in infrastructure, training for healthcare professionals, and robust validation processes to ensure patient safety and efficacy across diverse populations and clinical settings.

Specialists anticipate a future where AI acts as a collaborative partner for doctors, offering diagnostic suggestions, flagging potential risks, and streamlining administrative tasks. This symbiotic relationship could lead to earlier disease detection, more precise treatment plans, and a more equitable distribution of medical expertise. The ongoing development and rigorous testing of these AI tools are crucial to realizing their full potential while mitigating risks, ensuring that technology serves to enhance, rather than supplant, the human element of medicine.