¿Puedo Decirle a Mi Médico que No Quiera Que Use IA para Tomar Notas?

Whispers in the Exam Room: Navigating AI-Generated Notes and Patient Consent

It highlights Eric Boose, a family physician at Cleveland Clinic, who began employing an AI note-taking app about two years prior to the interview.

The tool listens to the patient encounter, then rapidly produces a visit summary, ostensibly freeing the clinician to focus on direct patient interaction rather than simultaneous note-taking.

The report situates this practice within a broader trend: roughly one-third of medical practices are using AI note-takers, with more exploring the technology to cut administrative workload.

• The narrative centers on real-world use of artificial intelligence (AI) assistants to generate visit summaries during primary care encounters.

In Boose’s description, the abstract typically becomes available within seconds after the visit ends.

The clinician can review and edit the AI-generated output before it becomes part of the patient record.

This is presented as a standard safeguard against inaccuracies.

Verbal consent is commonly accepted in practice, but the article notes that legal requirements for recording conversations vary by state.

If the patient declines AI assistance, clinicians reportedly resume manual note-taking.

• How the system functions: An AI assistant is activated during a patient visit to capture the dialogue and generate a concise recap of the encounter.
• Human-in-the-loop safeguard: There is an expectation that clinicians will verify and revise the AI-produced notes prior to their incorporation into the medical record.
• Consent mechanism: Patients are asked for permission at the outset of the visit to allow an AI assistant to aid in note-taking.
• Patient autonomy within the encounter: Patients can request the AI to pause or stop, particularly when discussing sensitive topics.

It may also omit important details or lose context from the conversation.

Patients are urged to scrutinize the resulting summary and to flag any discrepancies to their provider.

• Known limitations of AI notes: The technology can make errors or “hallucinate” entries, introducing inaccuracies into the record.
• Safeguards and responsibilities: The onus is placed on healthcare professionals to review and edit AI-generated summaries before they are added to the patient’s chart.
• Patient-facing diligence: Given potential inaccuracies, patients are advised to carefully review the visit summary and communicate any errors to the clinician.

In practice, clear answers may not always be forthcoming.

• HIPAA considerations: The article explains that AI vendors and the health systems deploying these tools have access to medical data and are bound by federal rules governing data use and storage under the Health Insurance Portability and Accountability Act (HIPAA).
• Data de-identification nuance: If data are de-identified—meaning identifiers are removed and the data cannot be linked to an individual—there is greater latitude for using the information in further ways, with fewer regulatory constraints.
• Patient inquiry on data use: Patients can ask their clinician or the health system how their data are being used.

It notes a presidential executive action at the start of 2025 aimed at reducing regulatory barriers to AI development and deployment in order to preserve U.S.

leadership in AI.

It also references a later Department of Health and Human Services (HHS) strategy on AI, articulated in December, which endorses integrating AI to modernize health care and public health infrastructure and to improve health outcomes at individual and population levels.

• Policy signals and political framing: The article sketches a policy landscape in the United States that appears supportive of AI integration in healthcare.
• Public participation and transparency: The piece references public-facing reporting by journalists and local outlets (e.g., Emily Siner of Nashville Public Radio) to illustrate the ongoing evolution and public discourse surrounding AI in clinical settings.

• Specifics about consent, surveillance, and data retention practices vary by state and by vendor; the article does not provide jurisdiction-specific permissions or a precise legal framework for every practice.
• It does not quantify the rate of AI accuracy across different patient populations or clinical contexts, nor does it provide detailed performance metrics for the AI tool(s) described.
• Patient experiences beyond the depicted clinician narrative are not systematically explored; the piece focuses on a representative clinician’s account and general regulatory commentary rather than a comprehensive survey.

• Consent and communication: Practices should ensure that patients are informed about AI-assisted note-taking at the start of visits, with the option to decline and to have notes recorded manually.
• Quality control: Clinicians must allocate time to review and correct AI-generated summaries to mitigate misrepresentation of the encounter.
• Patient safety and trust: Clinicians should facilitate patient review of the AI-derived notes and address any errors promptly to maintain trust and accuracy in the medical record.
• Data stewardship: Health systems should clarify to patients how data from AI note-taking are used, stored, and protected, and what happens when data are de-identified versus linked to an individual.

You can request that the AI be paused or stopped for sensitive topics.

• If your clinician proposes an AI assistant for note-taking, expect a basic consent exchange at the start of the visit.