Rapid advancements in AI (artificial intelligence) technologies, including machine learning, natural language processing, and computer vision, have developed sophisticated tools capable of performing complex medical tasks. The AI integration in healthcare can revolutionise the industry by improving patient outcomes, optimising resource allocation, and reducing operational costs. However, the AI use in medicine carries certain risks related to ethics and data privacy, shortcomings in the quality of data for training algorithms, and importance of protecting against cyberthreats. There is also a threat of rising medical costs due to the need for a large number of tests and validations of new technologies. This study focuses on the AI application in the diagnostic field, as it is revolutionising the medical industry by offering new opportunities for accurate disease detection, classification, and prediction of treatment outcomes. The diagnostic field specificity is that any changes in it affect both those medical professionals who directly perform diagnostic procedures and those medical specialists who use the results of diagnostic examinations in their work. The research consists of two stages. Stage 1 is a survey of 119 respondents (medical professionals in Ukraine) about their attitude to the integration of AI technologies in diagnostics. Stage 2 is a study of opinions by 10 experts (medical professionals in Ukraine) about their own assessment of AI risk parameters in medical diagnostics. The survey showed the vast majority of Ukrainian doctors (over 84%) had no experience with AI-based diagnostic systems. Simultaneously, 74% of respondents believe AI can be effective in reducing diagnostic errors, and the future of medical diagnostics is associated with AI. They consider its main advantages to be speed, accuracy, objectivity, and ability to detect diseases at early stages. Respondents argue that AI is the most appropriate for diagnosing cancer, genetic research, and chronic conditions with atypical symptoms. Regarding the risks and barriers to AI introduction in medical diagnostics, at the first study stage, respondents named the high cost of implementation, the need for specialised training, and the possible lack of personal interaction between doctor and patient as the main ones. This opinion was clarified at the second study stage. In particular, 10 experts ranked these risks and potential problems in the following order (from the most to least important): unequal access; dependence on technology; ethical issues; legislative and regulatory challenges; lack of personal contact; bias and inequality; data privacy and security; errors in diagnosis and treatment. To mitigate each of these risks, the article develops a set of recommendations.