Can artificial intelligence diagnose and treat patients? Will it replace doctors, or will it help them provide better care? And how is the doctor-patient relationship changing at a time when more and more people are consulting ChatGPT before visiting a doctor’s office? These questions were at the heart of the debate on artificial intelligence in medicine, organized as part of the HealthTech Forum 2026 and moderated by Associate Professor Sorin Păun, M.D., of the “Carol Davila” University of Medicine and Pharmacy, Department of Surgery, Bucharest Emergency Clinical Hospital, and Cristian Taslițchi, Product Owner at Soft 31.
The discussion brought together surgeons, family physicians, imaging specialists, patient advocates, and experts in technology and communication. Despite their differing perspectives, the participants agreed on one point: artificial intelligence can become an important tool in medical practice, but its use must be accompanied by professional competence, oversight, and responsibility.
AI can support diagnosis and treatment, but the doctor remains responsible
In surgery, artificial intelligence offers applications at all stages of care: preoperative, intraoperative, and postoperative. Participants presented examples of CT and MRI image analysis, the identification of hard-to-detect lesions, surgical planning, and the use of video footage to train surgeons.
During laparoscopic or robotic procedures, visual analysis solutions can help identify important anatomical structures. These applications have been discussed as technological opportunities, not as practices that are already widespread in Romanian hospitals.
Andreea Curcă, a representative of LightXVision, presented the use of AI for analyzing X-rays and identifying pulmonary nodules, highlighting its potential to aid in early diagnosis. However, she drew attention to false-positive results and described the case of a patient whom an AI system had erroneously flagged as having a cerebral aneurysm, causing him to panic. The message of her presentation was that automatically generated results must always be interpreted and validated by a physician, even when they come from certified medical applications.
Less bureaucracy, more time for patients
One of the most concrete opportunities identified was reducing the time spent on medical documentation. Doctors described the ever-increasing volume of forms, consent forms, and information that must be entered into computer systems. In family medicine, where an appointment is often scheduled for 15 minutes, this task reduces the time available for dialogue with the patient.
AI solutions for transcribing and structuring consultations could automatically generate draft observation sheets, medical letters, and clinical summaries, with the physician then reviewing and validating the information. Applications for scheduling appointments, sending notifications about routine checkups, and collecting preliminary information through virtual assistants were also discussed. However, participants emphasized that these systems must be able to distinguish between routine requests and potential medical emergencies. Automation must not delay access to care.
Cecilia Cherciu, a cybersecurity expert, explained that the effectiveness of AI assistants depends on the instructions and rules they are configured to follow. When it comes to medical documents, they must use standardized formats, refrain from adding information that has not been discussed, and flag uncertain elements for verification.
The patient comes to the doctor with a diagnosis suggested by ChatGPT
Another important topic was the changing doctor-patient relationship. Dr. Crenguța Lanbă, a senior family medicine physician, described the case of a patient who, after noticing a lump in his neck, consulted ChatGPT and arrived at the clinic convinced that he might have a serious cancer. This example illustrated the risk of self-diagnosis: AI tools can provide useful explanations, but they do not always have access to the patient’s complete medical history, do not examine the patient, and can lead to misinterpretations.
From the perspective of medical communication, participants recommended that doctors not dismiss patients’ questions based on information obtained online, but rather explain the differences between general recommendations and individualized clinical decisions. Radu Gănescu, President of COPAC and a patient advocate, also highlighted the potential benefits: a better understanding of the disease, the ability to prepare questions for the consultation, and faster access to information about treatment. In this context, AI can become a tool for patient education, provided that the information is critically evaluated and discussed with the doctor.
Without high-quality data and professional training, AI cannot produce reliable results
The discussion also addressed the conditions necessary for implementation: interoperability of IT systems, data quality, protection of medical information, and computing infrastructure. Elena Paraschiv, a Research Scientist at the National Institute for Research and Development in Informatics, presented the RO AI Factory project, highlighting the investments required to develop and host AI models, as well as the challenges posed by the rapid evolution of technology. Participants emphasized the need for well-defined objectives, the use of accurate data, and the continuous verification of results.
A separate topic was the training of future doctors. Alexandra Costinescu, who participated in the debate as a representative of young professionals, pointed out that quick access to AI-generated answers can discourage the pursuit of deeper knowledge. “It’s very tempting for us because it takes less time,” she explained, advocating for the preservation of independent study and critical thinking.
Along the same lines, several participants argued that medical education should include the responsible use of AI, without replacing clinical training or a physician’s ability to make independent decisions.
AI should be the doctor's tool, not the other way around
The debate highlighted that the benefits of artificial intelligence should not be evaluated solely based on the performance of algorithms, but rather on the outcomes achieved for patients and healthcare professionals. Reducing administrative time, identifying medical issues more quickly, supporting clinical decisions, and improving communication are promising areas. However, all of these require validated systems, protected data, and trained staff.
The panel's conclusion: The future of medicine does not involve choosing between doctors and artificial intelligence, but rather the responsible integration of technology into medical practice, so that professionals can devote more time to patients and make better-informed decisions.
About the HealthTech Forum 2026
The 9th edition of the HealthTech Forum, held November 5–6, 2026, at the Bucharest Chamber of Commerce and Industry under the theme “Smart and Interoperable Hospitals, ” brought together hospital administrators, doctors, and representatives from government agencies, academia, and technology companies to discuss the modernization of the healthcare system.
The HealthTech Forum 2026 was held with the support of partners who promote dialogue and collaboration among healthcare professionals, government authorities, academia, and the medical technology industry: Giv Solutions, Clinomic Romania, InfoWorld, Tremend Publicis Sapient, Medtronic, Philips, IQVIA, Docviser, Soft31, Innovate IT, POLITEHNICA University of Bucharest, “Gr. T. Popa” University of Medicine and Pharmacy in Iași, Right Partner, the Association of Public Hospitals, the Association of Medical Product Suppliers, the HL7 Association, the Romanian United Foundation, Global Romania, PALMED, the National Institute for Research and Development in Informatics, the AI Human Leadership Organization, the Romanian Society of Anesthesia and Intensive Care, and the Bucharest Chamber of Commerce and Industry. Organizer: Oameni și Companii.






