When one looks at the progress made in “large language models” such as ChatGPT, one cannot help but be amazed. Questions from all areas of life are answered reliably and linguistically sophisticated within seconds. Even though the sources of information are opaque and the AI sometimes hallucinates, the results are nonetheless impressive and will change our everyday lives, our information society, and our way of learning sustainably. AI is already in use for medical questions.And ChatGPT & Co are also strong with medical questions. Nowadays, it is no longer a problem to enter medical symptoms and receive a diagnosis in no time. If one consistently thinks further about what is already possible from every smartphone today, then professional systems that have learned from correspondingly large data sets will in the future provide much more reliable and probably also more economical medical diagnoses and treatment recommendations than any (human) doctor.
Will our profession still be needed then? What is medical art in the future if one can diagnose cancer at home on the PC or put together the best treatment for a chronic disease?
AI as a supplement for medical questions
I do not share these substantial concerns about the future of our profession. Of course, medicine will change through “artificial intelligence” and is already doing so. Image and pattern recognition improve radiology or dermatology, automatic interaction checks improve drug safety, robot-assisted surgery or neuroprosthetics would not be possible without the use of AI. Even these examples illustrate that AI can help make the medicine of the future better and more reliable – which is initially a great gain.
Especially in neurology, where we sometimes deal with very complex and very rare disease patterns, I would find it absolutely beneficial to be helped by AI systems in diagnosing and selecting the most sensible therapy. This would save time and make our work more targeted. In this respect, it is even rather the case that we are lagging a bit behind the possibilities in Germany.
As far as patient autonomy is concerned, which will be further increased by the “large language models”, I also have no worries. I am rather relaxed about the scenario that makes many medical colleagues nervous, namely that the patient already comes to the practice with his diagnosis and brings clear ideas of which therapy should be prescribed – especially because we already know this situation. Since the availability of search engines on the internet (which are also getting better and better), we have been confronted with Dr. Google and its effects.
Information from AI requires classification by experts
But ultimately, the development so far has rather taught us how important the real doctor-patient contact is in times of unbridled information acquisition. No matter how good and precise the AI diagnosis is, there will always need to be a doctor who, with his understanding of the human body, his knowledge of pathophysiological processes, and his experience in clinical medicine, puts this information in the right context and accompanies the patient through his illness. Even if computer-supported systems can be astonishingly empathetic, I believe that, in the end, human contact is needed in medicine. Our profession will change, it may get different priorities, but it will certainly not become dispensable. And accordingly, I personally look forward to the upcoming technical developments with curiosity and anticipation rather than with worry.
This post was translated from German to English with the help of AI.
