- AI-powered medical systems matched or exceeded doctors in specific diagnostic tasks in new studies
- Mira AI diagnosed emergency cases with 87.1% accuracy versus 78.1% by physicians
- Google's Amie AI created treatment plans more aligned with guidelines than 21 UK doctors
Artificial intelligence may be closer than ever to becoming a trusted clinical assistant. A new study published in Nature has found that advanced AI-powered medical systems performed as well as, and sometimes better than, doctors in specific diagnostic and treatment-planning tasks. The findings add to growing evidence that AI is rapidly improving its ability to analyse medical information, recognise patterns, and recommend evidence-based care.
One system, called Mira, outperformed physicians when diagnosing patients using electronic health records from emergency departments. Another tool, Google's conversational medical AI known as Amie, generated treatment plans that aligned more closely with clinical guidelines than those produced by a group of practising doctors. While the results have generated excitement among researchers, experts caution that these tools were tested under controlled conditions and are not yet ready to independently manage patients in real-world healthcare settings. The study nevertheless marks an important milestone in the evolution of medical AI and raise questions about how doctors and machines may work together in the future.
What Did The New Studies Find?
According to research published in Nature, researchers tested two specialised medical AI systems on clinical tasks that are typically performed by physicians.
The first system, Mira, analysed more than 500 emergency department cases using electronic health records. The AI achieved a diagnostic accuracy of 87.1 per cent, compared with 78.1 per cent for a panel of physicians reviewing the same cases. The tool was designed to function as an autonomous medical reasoning agent capable of synthesising information and generating diagnostic suggestions.
The second system, Amie, built on Google's Gemini AI platform, was evaluated on its ability to interact with patients and formulate treatment plans. Researchers found that the AI's recommendations aligned more closely with established clinical guidelines than those generated by 21 practising physicians in the United Kingdom. The system also performed strongly when handling complex medical scenarios.
Together, the studies suggest that AI is progressing beyond simple symptom checkers and entering a phase where it can assist with sophisticated clinical reasoning.
Why Is AI Improving So Rapidly?
Modern AI systems are built on large language models (LLMs), which are trained on vast amounts of medical literature, clinical guidelines, textbooks and anonymised patient data.
Over the past two years, researchers have reported dramatic improvements in AI performance on medical licensing examinations and clinical reasoning benchmarks. In some studies, advanced AI models have demonstrated diagnostic capabilities comparable to experienced clinicians when presented with structured patient information.
Unlike human clinicians, AI systems can rapidly analyse large volumes of information simultaneously and instantly cross-reference medical guidelines, research findings and treatment recommendations. This ability makes them particularly useful for pattern recognition and decision support.
Could AI Replace Doctors?
Despite the impressive results, researchers emphasise that AI is not ready to replace physicians. One of the biggest limitations is that these studies were conducted in controlled environments. Real-world medicine involves far more than reviewing records or generating diagnoses. Doctors must perform physical examinations, assess non-verbal cues, understand social and cultural factors, communicate uncertainty, and make decisions when information is incomplete.
Experts interviewed by Nature noted that an AI outperforming physicians on a narrowly defined task does not necessarily mean it can replicate the complexity of real clinical practice.
Furthermore, AI systems can still make mistakes, generate inaccurate information, and occasionally produce recommendations that appear confident despite being incorrect. Researchers behind the Mira study acknowledged that the tool sometimes deviated from best-practice clinical approaches.
What Are The Risks?
While AI may improve efficiency and reduce diagnostic errors, concerns remain about reliability, bias, accountability and patient safety.
A large study published in Nature Medicine earlier this year found that although AI models performed well when tested independently, members of the public using those same systems did not consistently make better medical decisions. Researchers concluded that human interaction with AI remains a major challenge and that strong safeguards are necessary before widespread deployment.
Other experts warn that AI systems may inherit biases present in their training data and may not perform equally well across different populations or healthcare settings. Transparency, regulatory oversight and clinical validation will therefore be critical before these tools are integrated into routine care.
How Could AI Change Healthcare?
Most experts believe the future is likely to involve collaboration rather than competition between doctors and AI. AI could help clinicians identify overlooked diagnoses, summarise patient histories, suggest evidence-based treatments, interpret medical records, and reduce administrative workloads. This may allow physicians to spend more time on direct patient care and complex decision-making.
Researchers increasingly view AI as a powerful "second opinion" tool that could enhance diagnostic accuracy and support healthcare systems facing workforce shortages and rising patient demand.
The latest Nature studies provide some of the strongest evidence yet that specialised medical AI systems can match or even outperform physicians in certain diagnostic and treatment-planning tasks. However, the findings should not be interpreted as a sign that doctors are becoming obsolete.
Medicine involves far more than identifying diseases and recommending treatments. Empathy, communication, ethical judgement, and the ability to navigate uncertainty remain uniquely human strengths. For now, the most likely future is one in which AI serves as a highly capable clinical partner, helping doctors deliver faster, safer, and more personalised care rather than replacing them altogether.
Disclaimer: This content including advice provides generic information only. It is in no way a substitute for a qualified medical opinion. Always consult a specialist or your own doctor for more information. NDTV does not claim responsibility for this information.