Upgrade your account to a Community Supporter account and remove most of the site ads.
Enchanted Trinkets Complete--a hardcover book containing over 500 magic items for your D&D games!

Reply to thread

Message: <blockquote data-quote="Gorgon Zee" data-source="post: 9832247" data-attributes="member: 75787">Maybe if we had more positive news stories about .. how it is used to help in medicine we'd have people with a more balanced view of it.I'm not sure that it makes sense to characterize medicine as a "highly specialized field". It's often quoted as being 15% or more of the GDP of the United States, and, furthermore, it absolutely is being targeted by the tech companies. In an earlier post you talked about the difference between using GenAI for creative tasks and non-creative tasks. I agree with you that the creative uses look like they are self-limiting. We are not seeing much in the way of improvement recently; the input data they use is close to exhausted and new data is heavily polluted.  But where I disagree is that the GenAI industry cannot survive without being used for creative purposes. Certainly GenAI is being hyped and 2025 as the "year when agents will take over menial human tasks" never materialized, and looks unlikely to in 2026. Or, possibly, ever.But like any hyped technology, what I expect to happen is that the boring, mundane solutions that are value-for-money will be refined and grow, and the hyped areas will wither away.For everyone's consideration, here are a number of scenarios in healthcare where GenAI has as a strong ROI (and would have at 10x the current cost) and, I would argue, are also uses that make the world a better place:Imaging Incidental FindingsIf you get a mammogram X-ray, or to see if you broke a rib falling off a ladder, the person viewing that set of images is not looking for other features. They might not spot an issue with your lungs because they aren't looking for that -- it's probably not their job or specialty. A GenAI solution can alert an image technician to things they might not have seen. Obviously good for the patient, it also saves money for the healthcare provider. For context, the US does over a quarter of a billion x-rays per year (not including a billion+ of dental ones) and 100m+ CT scans.New Patient SummarizationsYou are a doctor in a local clinic well away from a hospital, and a patient comes in with a serious condition that needs specialist attention , so you send them immediately to a big hospital to be seen asap. That hospital needs to review the patient's medical records rapidly. If we are lucky, the records are electronic. If we are unlucky, the doctor faxes images of the patient records to the hospital. Typically a nurse will have to review this and summarize -- potentially overnight for a surgery the next day. This means a nurse will have to read something that about a third of the time is longer than Moby Dick in the middle of the night and make sure they don't miss anything important. GenAI summarization is really good at this task, improving accuracy and assisting the nurses to get the job done faster.Automatic TranscriptionYou may have experienced this yourself -- your doctor may have asked your permission to use AI to record a conversation with them, which is then summarized and can be used as the basis for the doctor's notes. This is an area I've personally done validation studies on (<a href="https://ai.nejm.org/doi/abs/10.1056/AIoa2500945" target="_blank">https://ai.nejm.org/doi/abs/10.1056/AIoa2500945</a>) and it not only saves doctor's time (especially work outside work hours), but the resulting summaries are really quite good -- we used a couple of ways of measuring quality and compared them to non-AI workflows. -------------------I have been pretty unimpressed with GenAI creativity. Studies also show that although it can be shown to be more creative than a single random professional, because it is more creative in the same way for everyone, it's less creative than a group of people. But the boring "non-creative" uses I think are quite strong. Not enough to justify the current hype levels of company valuation, but I do think they are sufficiently sustainable to ensure that GenAI will continue to be a seriously large industry.</blockquote>

Verification