ChatGPT Applications in Healthcare

Episode Takeaways

• With ChatGPT thrusting AI onto the world stage and into the broad public awareness for the first time, there are concerns about whether this technology will ultimately overcome humanity. We’re at the center of something that we didn’t see coming. We thought that COVID was the inflection point, but who saw these large language models coming as fast as they’ve arrived? We’re in the throes of wonder and fear.
• AI has already been given a dystopian mantle of ‘bad’ by Hollywood, but that’s not reality. The system tries to spend more time keeping AI out of the system as opposed to thinking about how to integrate it to improve outcomes in a practical way. For example, it can help us summarize complex ideas and help us avoid our own biases.
• Think about it: technology is already augmenting our tasks. Your emails are half written by prompts and your spelling is corrected automatically. ChatGPT can be so much more than that. It’s an iterative conversation. You can ask for a recipe, and it will give you one. Then you can ask it to refine, adjust, and change the recipe to suit your tastes. If we can think of it as an assistant and a streamlining mechanism, we’ll be able to take fuller advantage.
• It’s exciting to think of the potential for ‘zebra-type’ patients, where no one is piecing together the elements of the patient’s symptoms to identify rare and unusual conditions. ChatGPT can be a souped up search engine with the ability to analyze data and make those connections. Sure, it can be wrong, but so can doctors. We don’t hold clinicians to a standard of perfection, and shouldn’t expect that from GPT at this point, though perhaps someday in the future.