Gleamer

An AI scribe typically works in four steps:

• Capture: a phone, computer or room microphone records the consultation after the patient is informed or consents.
• Transcription: speech recognition converts audio to text and separates speakers.
• Summarization: a language model extracts relevant content and drafts a note in the chosen format, such as a SOAP note, sometimes with codes or orders suggested.
• Review: the clinician edits and signs the draft, which is then filed in the EHR, often through an integration.

Each step can introduce errors, so final responsibility for the note stays with the clinician. Ask vendors whether audio is stored, whether data is used to train models, and how fabricated or missing content is measured.

General diagnostic imaging software stores, displays and manipulates images, for example PACS viewers, 3D reconstruction or measurement tools operated by the user. Radiology AI adds automated analysis that produces its own output, such as a detection, triage flag or quantitative result.

The distinction matters for evaluation:

• Viewing and processing tools are judged mainly on image quality, usability and integration.
• AI analysis tools also need evidence on accuracy, populations studied, failure modes and monitoring.

Both can be regulated medical devices depending on intended use; in the US many image management and processing functions fall under device classifications. Many vendors now combine the two, so ask which functions are AI-based and which are covered by a regulatory authorization.

Hallucination is when a generative AI tool produces content that sounds plausible but is false or unsupported by its inputs, such as a symptom, medication or exam finding that was never mentioned in a consultation.

It matters most in documentation, summarization and question-answering tools, because errors can be copied into the medical record. Common safeguards are:

• Clinician review and sign-off of every generated note or summary.
• Links from generated text back to the source audio or record.
• Vendor testing that measures omission and fabrication rates in realistic conditions.
• Clear policies on what the tool must not be used for.

WHO's guidance on large multi-modal models highlights inaccurate or false outputs as a key risk.