Clinicians Must Be AI-Ready
What happened
- A JMIR survey of Veterans Health Administration clinicians found that clinicians are the key interface for AI tools in care. - The study highlighted gaps in knowledge, attitudes, and training needs for primary-care clinicians nationwide. - It argues clinician training and governance are essential to prevent AI from adding opaque steps to patient care. (formative.jmir.org)
Why it matters
Primary-care clinicians are already using artificial intelligence, but a new Veterans Health Administration survey found many still lack the training to judge its risks and limits. (formative.jmir.org) The study, published April 22 in *JMIR Formative Research*, surveyed 170 Veterans Health Administration primary-care providers in October 2025 from a pool of 989 clinicians, a 17.2% response rate. (formative.jmir.org) Two-thirds of respondents, 66.5%, said they were already using AI, most often generative AI tools and decision-support systems. About 70.6% said they felt mostly enthusiastic, or more enthusiastic than apprehensive, about AI. (formative.jmir.org) Artificial intelligence in clinics often works like a prediction or drafting engine: it can summarize notes, suggest next steps, or generate text for messages and records. In primary care, those tools sit inside routine visits, inboxes, and charting, where clinicians—not software—still make the final call. (formative.jmir.org) That makes clinician judgment the safety check. In the survey, only 36.5% said they were confident they understood sources of AI bias, and 47.6% said they were confident about ethical issues tied to AI use. (formative.jmir.org) The paper also found that very few respondents could give an accurate technical definition of AI when asked in an open-ended question. The authors said that gap can complicate adoption when health systems roll out tools meant for frontline care. (formative.jmir.org) The Veterans Affairs department has been building that rollout infrastructure. Its AI office says the agency is adopting “trustworthy” AI, maintains an AI inventory, and offers workforce resources and guidance for generative AI use across the department. (department.va.gov) Outside the Veterans Health Administration, the training problem is not unique to one system. A 2023 *JMIR Formative Research* review on doctors’ AI roles and competencies found that physicians need skills in evaluation, implementation, and communication as AI moves into everyday care. (formative.jmir.org) That lines up with a broader push in medicine to treat AI as a clinical tool that needs oversight, not just software that can be dropped into workflows. A 2025 *JAMA* summit report said AI in health care should be developed, evaluated, regulated, disseminated, and monitored. (jamanetwork.com) The survey’s basic finding is narrow but practical: if AI writes, predicts, or flags something in a patient’s chart, the clinician using it has to know what the tool is doing before the patient ever sees the result. (formative.jmir.org)
Key numbers
- (formative.jmir.org) The study, published April 22 in *JMIR Formative Research*, surveyed 170 Veterans Health Administration primary-care providers in October 2025 from a pool of 989 clinicians, a 17.2% response rate.
- (formative.jmir.org) Two-thirds of respondents, 66.5%, said they were already using AI, most often generative AI tools and decision-support systems.
- About 70.6% said they felt mostly enthusiastic, or more enthusiastic than apprehensive, about AI.
- In the survey, only 36.5% said they were confident they understood sources of AI bias, and 47.6% said they were confident about ethical issues tied to AI use.
What happens next
- (formative.jmir.org) Artificial intelligence in clinics often works like a prediction or drafting engine: it can summarize notes, suggest next steps, or generate text for messages and records.
- (formative.jmir.org) The paper also found that very few respondents could give an accurate technical definition of AI when asked in an open-ended question.
- The authors said that gap can complicate adoption when health systems roll out tools meant for frontline care.
Quick answers
What happened in Clinicians Must Be AI-Ready?
A JMIR survey of Veterans Health Administration clinicians found that clinicians are the key interface for AI tools in care. The study highlighted gaps in knowledge, attitudes, and training needs for primary-care clinicians nationwide. It argues clinician training and governance are essential to prevent AI from adding opaque steps to patient care. (formative.jmir.org)
Why does Clinicians Must Be AI-Ready matter?
Primary-care clinicians are already using artificial intelligence, but a new Veterans Health Administration survey found many still lack the training to judge its risks and limits. (formative.jmir.org) The study, published April 22 in *JMIR Formative Research*, surveyed 170 Veterans Health Administration primary-care providers in October 2025 from a pool of 989 clinicians, a 17.2% response rate. (formative.jmir.org) Two-thirds of respondents, 66.5%, said they were already using AI, most often generative AI tools and decision-support systems. About 70.6% said they felt mostly enthusiastic, or more enthusiastic than apprehensive, about AI. (formative.jmir.org) Artificial intelligence in clinics often works like a prediction or drafting engine: it can summarize notes, suggest next steps, or generate text for messages and records. In primary care, those tools sit inside routine visits, inboxes, and charting, where clinicians—not software—still make the final call. (formative.jmir.org) That makes clinician judgment the safety check. In the survey, only 36.5% said they were confident they understood sources of AI bias, and 47.6% said they were confident about ethical issues tied to AI use. (formative.jmir.org) The paper also found that very few respondents could give an accurate technical definition of AI when asked in an open-ended question. The authors said that gap can complicate adoption when health systems roll out tools meant for frontline care. (formative.jmir.org) The Veterans Affairs department has been building that rollout infrastructure. Its AI office says the agency is adopting “trustworthy” AI, maintains an AI inventory, and offers workforce resources and guidance for generative AI use across the department. (department.va.gov) Outside the Veterans Health Administration, the training problem is not unique to one system. A 2023 *JMIR Formative Research* review on doctors’ AI roles and competencies found that physicians need skills in evaluation, implementation, and communication as AI moves into everyday care. (formative.jmir.org) That lines up with a broader push in medicine to treat AI as a clinical tool that needs oversight, not just software that can be dropped into workflows. A 2025 *JAMA* summit report said AI in health care should be developed, evaluated, regulated, disseminated, and monitored. (jamanetwork.com) The survey’s basic finding is narrow but practical: if AI writes, predicts, or flags something in a patient’s chart, the clinician using it has to know what the tool is doing before the patient ever sees the result. (formative.jmir.org)
