![]() They asked everyone (and -thing) to diagnose the illness, and then for triage recommendations. Well, you know who's very good at robotic things? Robots! Recently a team of researchers from Harvard showed dozens of descriptions of health problems to three groups: physicians, people with no medical training, and ChatGPT. A human clinician backed by the knowledge base and processing power of AI systems will only be better Either way, you don't expect to enjoy these encounters. Or, if you have electronic access to your medical records, maybe you get results back from a new set of tests, and you email your doc to ask what they mean. You feel icky and call an advice nurse they ask preset questions to determine whether you should get to an ER or just go take a Tylenol. Don't tell anyone I said this, but a lot of what healthcare workers do is already a bit formulaic - at least at the lowest-level, patient-facing interface. But they're probably going to get it anyway. Sixty percent of Americans surveyed recently by the Pew Research Center said they wouldn't want an AI system diagnosing what's ailing them or proposing treatments. If you ask people whether they're into that idea, they mostly say no. If you're lucky enough to have health insurance, your insurance company probably already has some kind of dumb chatbot for you to talk to before you can get a human on the phone. Dozens of companies are working on applications, aiming for uses from diagnosing illnesses to helping with the slog of paperwork that has somehow become the responsibility of both doctors and patients alike. So the medical establishment is jumping on chatbots as a cheaper, more ubiquitous tool. They're highly trained to detect one thing, like a tumor or sepsis, using specific test results as input. Specialized AI systems - not dumb chatbots - are already pretty good at diagnostics. An artificially intelligent assistant may not be more human than human, but maybe it'll be more humane. But if our current, broken healthcare system makes it impossible for humans to take care of one another, maybe fake taking-care will save real lives. Nobody thinks ChatGPT actually cares, any more than they think it's actually smart. And doctors didn't sign up for it."įor all the tech-world promises of robot pets and AI psychotherapists, the idea of a caring chatbot still feels destabilizing - maybe even dangerous. So they're looking for answers on forums like r/AskDocs. "People are disconnected from healthcare, and they're desperate," says John Ayers, a computational epidemiologist at UC San Diego who was lead author of the new paper. Nobody likes it, except the people getting rich. Our for-profit healthcare system doesn't hire enough doctors and nurses, and it expects the ones it does hire to treat more and more patients, assembly-line style. It was to show that a chatbot could have a role in the provision of care. ![]() The point of the empathy experiment wasn't to show that ChatGPT could replace a physician or a nurse. But I now think that with some tinkering, chatbots could radically improve the way people interact with healthcare providers and our broken medical-industrial complex. I suppose I'm open to the idea that they'll accelerate the coding of software and the analysis of spreadsheets. I'm skeptical that AI bots driven by large language models will revolutionize journalism or even make internet search better. Still, the apparent facility with which the bot could handle medical concerns, in both style and substance, presages an actual, real-world use for these things. ![]() Now admittedly, the bar is low on beating human doctors for displays of empathy. Data figured out how to convincingly emulate Dr. Seven times! They provided just what you want from your doctor: care and an emotional connection. What's more, the bots didn't show any of the distressing tendency to make stuff up that they often have in other circumstances.īut here's the most striking part: The chatbot answers, on average, were rated seven times as empathetic as the ones from humans. Almost invariably, the chatbot answers were rated as three or four times as reliable as the ones from the poor wee humans. For one thing, ChatGPT came out well ahead of the human doctors on usefulness. Account icon An icon in the shape of a person's head and shoulders.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |