In the near future, doctors could work alongside assistants powered by large language models such as GigaChat to help diagnose conditions and suggest treatment plans. This possibility was discussed in a recent interview with socialbites.ca, featuring Andrey Belevtsev, Senior Vice President and Head of Sberbank Technologies. He outlined a clear path for how such AI companions might integrate into medical practice and affect patient care.
Belevtsev noted that Sber is already advancing toward this vision. He highlighted that GigaChat’s productive artificial intelligence recently passed a professional medical exam as part of a collaborative effort with medical experts from the National Medical Research Center that shares the same name. The neural networks were trained on a substantial dataset provided by VA Almazova, which comprised multiple gigabytes of medical information to build a robust knowledge base for clinical reasoning.
During the evaluation, the AI was presented with the doctor’s exam, a basic measure of clinical competence. The model achieved a score of 4 on the oral portion, while on a more extensive 100-question written assessment it earned an 82 percent, above the traditional 70 percent passing threshold. These results were presented as evidence of the system’s ability to engage with medical material at a level approaching human proficiency in standardized testing scenarios.
Belevtsev emphasized that, given the current pace of AI advancement, intelligent assistants for doctors are likely to become a practical reality within the next few years. He suggested that mobile formats will be the most probable delivery method, enabling clinicians to access AI-driven insights on the go, alongside patient records and other essential data streams.
The executive pointed out a key advantage of neural networks: they operate continuously, without sleep or fatigue. They can process a full patient history and comprehensive anamnesis, analyzing information in ways that may help clinicians avoid oversights and consider possibilities they might not immediately recall. In this sense, such AI tools could function as valuable physician assistants, supporting decision making while clinicians retain ultimate accountability and supervision.
Further exploration into the development of Russian AI technologies and the prospects for systems like GigaChat continues, with ongoing collaboration among technology teams and medical institutions. The progression of these tools reflects a broader trend toward integrating artificial intelligence into health care, aiming to enhance efficiency, accuracy, and access to medical expertise in complex clinical settings.