A representative of the press service of Sber

[tanjimajuha20]

The service will allow customers of the Russian online pharmacy Sber Eapteki to receive support on pharmaceutical issues - drug selection, administration schedule, contraindications and side effects. "The digital assistant will automatically generate a response to the request, and the contact center operator will evaluate the quality of the generated response and only after checking will send it to the client. Connecting the service will reduce the waiting time for a pharmacist's response, and will also reduce the workload of contact center employees," said a statement from Sberbank.
Eapteka told a ComNews romania whatsapp resourcecorrespondent that the model's answers are fully checked by pharmacists with special training and comply with the law: "At the same time, the final responsibility for consultation on drugs, rules of administration and side effects will remain on the side of the pharmacist. The concept of pharmaceutical consulting and what it includes is defined in the order of the Ministry of Labor and Social Protection of the Russian Federation dated May 31, 2021 No. 349n "On approval of the professional standard of the "Pharmacist".

Ilya Petukhov, Head of AI Solutions Development Projects at Directum LLC (Directum is a developer of electronic document management systems), mentioned that responsibility for the consultation remains with the operator, who must check the recommendations of artificial intelligence (AI) for errors and contraindications, and also request additional information from the person making the request. "In terms of requirements for operators, they will remain the same, since it is necessary to check what the machine has generated in any case. At the same time, AI will simplify the task for a person, the work will speed up, and the same team of specialists will be able to process more requests," Ilya Petukhov noted.

Alexander Nesterov, a research fellow at the Strong AI in Medicine laboratory at the AI ​​Research Institute (AIRI), noted that the medical knowledge embedded in GigaChat, combined with such LLM system design techniques as Retrieval Augmented Generation (RAG), allows for highly accurate answers to reference questions, which include drug selection and other similar information: "It follows from the developers' statement that all GigaChat responses are validated by pharmacists, which reduces the risks of potential inaccuracies. With the correct organization of the work of LLM systems of an advisory nature, possible risks are preventable; the main thing for the user is not to ignore the recommendation and seek additional clarification from the doctor if it appears."