Clinics use AI to draft exam summaries, doctors still must review

In short: More clinics are using AI to draft visit notes and exam summaries, but doctors are still expected to check and approve them.

What's going on

AI systems are increasingly acting like a medical scribe. With a patient’s consent, some tools can record a clinic visit and turn the conversation into a draft medical note, plus a plain-language summary for the patient. A clinician then reviews, edits, and signs the final version.

These tools are also being used to shorten long and messy medical records. They can pull together clinic notes, lab results, and imaging reports into a shorter overview, so a doctor can quickly see what changed since the last visit. Some systems even link each key point back to the original documents, like footnotes in a school report, so details can be checked.

Research suggests these summaries can be as good as, and sometimes better than, human-written ones in controlled tests. One Stanford-led study found clinician reviewers rated AI summaries at least as good as human summaries in 45 percent of cases, and better in 36 percent of cases. But the same body of research stresses a key limit, these tools are meant to assist, not diagnose on their own.

What to watch

The main risk is trust. AI can leave out important details, get priorities wrong, or sometimes invent information, even if that happens less often with newer models in some tests. Health systems and researchers are responding with clearer evaluation checklists and closer monitoring, but for now the safest use is when doctors treat AI like a draft writer and stay responsible for the final medical judgment.

Source: NYTimes