I don't want to automate the jobs of medical professionals, but AI is going to replace them nonetheless.

I'm a research engineer in the field of medical imaging AI. Most of the past decade was spent training deep learning models to better diagnose disease from MRI and CT scans. We automate the jobs of radiologists because we can. AI can do it faster, more accurately, and cheaper than any human radiologist. And while it may be interesting to talk about the ethics of replacing some of the most highly paid professions in medicine, AI is going to replace them regardless.<br><br>When I tell clinicians or medical students about AI replacing medical professionals, they always respond with some form of "well, we have 8 years of medical school and 6 years of clinical residency. It's not that easy." And they're right, it's not that easy. However, AI learns through a process called deep learning, which is much different than the way humans learn. So while a clinician might spend 14 years studying for their profession, we can train an AI to do the same thing in just a few months. And we don't even need to train a different AI model for each disease - a single model can be trained to recognize thousands of diseases in a single training session, making it arguably better than any human clinician.<br><br>The first AI model that could accurately diagnose diabetic retinopathy was published in a Nature Medicine paper back in 2016. Less than a decade later, and that same AI model is being used in primary care centers all over the United States, and has even received FDA approval to be used as a diagnostic tool. The FDA has since approved AI models for diagnosing everything from age-related macular degeneration to brain hemorrhage. AI is better at detecting brain tumors than radiologists.<br><br>But it's not just radiology that's going to be replaced - AI is being used in dermatology and cardiology as well. AI algorithms can be used to classify skin lesions as either benign or malignant, or automate stress tests and detect abnormal heart rhythms. For example, SinoaTics is an AI-powered dermatoscope capable of classifying skin lesions into 4 different categories: Actinic Keratoses, Basal Cell Carcinoma, Seborrheic Keratoses, and Melanoma. It's also handheld, so a patient can simply walk into their local pharmacy and ask for a screening.<br><br>AI is going to replace a lot of jobs in the future, many of which we consider to be high value and require a lot of skill. This might be bad for the clinicians that spent thousands of hours training, but it will undoubtedly be better for patients and the overall quality of healthcare. Unfortunately, this is just a consequence of technological progress. We automate jobs because we can, and because it's better for everyone.