It turns out that recently published studies show that it is impossible to completely trust the techniques of medical imaging presenting with the help of artificial intelligence. An international research team from the University of Cambridge, led by Anders Hansen, announced that deep learning tools that transmit high-quality images through quick scans could create numerous distortions of artifacts that, unfortunately, will distort diagnosis. Artificial intelligence systems are very advanced and can describe x-ray images, computer tomography, or magnetic resonance imaging. Due to these possibilities, it was decided to use AI for image reconstruction. However, a problem arose; SI systems have serious issues. Unfortunately, they may miss small structural changes such as cancerous tumors. The situation can occur even with a slight disturbance, yet the patient’s movement, and be reproduced as severe artifacts in the output image. Hansen said researchers need to test system stability. Most systems are unstable. The problem is dangerous because it is not possible to explain mathematically how such systems work. Without specific testing, catastrophically wrong results will be presented. Fortunately, such systems are not yet used in practice.