Post: The potential and pitfalls of AI in health care

The potential and pitfalls of AI in health care

By Robert Coe, AcuityMD

AI is still deep in the hype cycle, but signs of skepticism are emerging. Growing doubt is driven in part by AI proponents who are overpromising and under-delivering. Last summer, Google published a video that they claimed was created by their Gemini AI and showed broad, impressive capabilities – but the video was faked. Google had to take it down and apologize.

In late December 2024, a video of a gymnast doing a tumbling routine from Open AI’s Sora went viral because at times the gymnast had three or four legs and her head rotated around her torso. These public “failures” are obscuring the very real benefits that AI is delivering, especially in health care.

A great example of AI technology that is delivering tremendous benefits in health care is Abridge, which transforms conversations between a patient and provider into structured clinical notes. When thinking of an AI-driven technology that is hugely helpful to healthcare providers, a voice-to-text app probably wouldn’t be the first choice. But if providers were asked what they’d like to have week-to-week, the most popular answer would be “more time.” Abridge’s high adoption rate clearly shows the value in a tool that saves providers hours of transcribing notes each week. AI solutions that enable a user to essentially extend their abilities by automating manual, time-consuming tasks will be a growth segment.

Another example of extending abilities comes from companies like LeanTaaS, which uses AI-based predictions to help schedule infusion clinics and operating rooms. Handling administrative scheduling tasks is not the most futuristic instance, but has real-world practicality of making a shorthanded and overburdened hospital staff much more efficient.

AI is also powerful at synthesizing data across siloed sources. Medical device (MedTech) companies could use this power as a tool that pulls all recent information on a healthcare provider and puts it into a pre-call plan for a sales rep. Now a rep can get all the information they need to prepare instantly, instead of searching through dozens of websites and documents to piece research together. […]

The potential and pitfalls of AI in health care