Hospitalists

Hospitalists do many tasks, and AI is mostly helping with the routine ones. For example, writing discharge summaries (the notes doctors make when patients leave the hospital) has been tested. One study found ChatGPT could draft these notes about as well as a junior doctor ^[1].

However, other research shows much of the needed information isn’t in the electronic chart, so doctors still have to add details from memory ^[1]. In practice, AI might give a first draft that the doctor edits. Similarly, hospitals are using AI models to predict which emergency patients will need to be admitted ^[1].

These tools help plan bed assignments, but the final admission decision stays with the doctor.

Hospitals have mixed reasons to adopt AI. On one side, there’s a strong incentive: doctors spend many hours on paperwork ^[1], so automating routine notes and orders could save time and money. Indeed, some health systems are funding AI projects to automate such tasks ^[1].

On the other side, costs and trust are barriers. Buying or building AI systems and training staff is expensive. Most importantly, people must trust the technology.

Surveys show many doctors appreciate AI’s potential but worry about its “black box” nature and who would be responsible if it makes a mistake ^[1]. Reviews note that clear legal rules and oversight are needed so clinicians and patients feel confident using AI ^[1]. In short, AI can bring efficiency and help with staff shortages, but wide use will only come as these tools prove safe, reliable, and well-regulated in hospital practice.