AMA launches Center for Digital Health and AI to keep physicians in the loop
The American Medical Association has created a new Center for Digital Health and AI after two years of research into how physicians use AI and what they need next. The goal is clear: keep doctors involved in design, rollout, policy, and oversight so AI supports care instead of adding more work.
AMA leaders say clinicians must be at the center of technology decisions. Without that, tools get built in ways that miss clinical realities and create new administrative burdens.
Why this matters for clinicians
AMA data shows a quarter of physicians feel more concerned than optimistic about AI in practice. Top issues: data privacy, reliability, safety, and the fear of being blamed for model errors.
As AMA's CEO Dr. John Whyte noted, digital tools are everywhere, but if they don't fit how care is delivered, they won't stick. Clinical context and workflow understanding are non-negotiable.
What the new Center will focus on
The Center will work with regulators, policymakers, health systems, and tech leaders to turn AI from promise to practical value. Priorities include:
- Policy and regulatory leadership: Create clear benchmarks for safe, effective AI use in medicine and digital health.
- Workflow integration: Co-develop tools with physicians so they improve patient care and reduce friction for clinicians.
- Education and training: Equip physicians and teams to evaluate, implement, and monitor AI tools confidently.
- Cross-sector partnerships: Connect tech, research, government, and healthcare to build patient-centered clinical AI.
The larger trend
Physician use of AI has nearly doubled from 2023 to 2024, with about two-thirds reporting they've adopted at least one AI-driven tool. Still, 47% say stronger oversight would increase their trust.
Doctors want explicit data privacy protections (87%) and do not want to be held liable for model errors. In response, the AMA introduced an eight-step governance framework covering accountability, oversight, and staff training to guide responsible deployment. See AMA's resources on AI in healthcare for context and updates: AMA: AI in health care.
What this means for your organization
- Identify 2-3 high-value use cases with clear outcome metrics (e.g., turnaround time, documentation time, readmissions).
- Bring clinical leaders and frontline staff into vendor selection, design reviews, and pilots early.
- Map workflows before and after implementation; remove extra clicks and duplicate entry.
- Set data safeguards up front (access controls, logging, de-identification where possible).
- Clarify liability and escalation paths; ensure human review for high-risk outputs.
- Start narrow pilots, measure impact, and expand only if safety and value hold.
- Train for real-world use: indications, limits, bias risks, and failure modes.
- Use a governance checklist aligned with AMA guidance for approval, monitoring, and sunsetting.
Policy and oversight touchpoints
Expect more coordination across federal and state bodies on AI transparency, accountability, and post-market monitoring. For reference on medical AI oversight, see the FDA's page on AI/ML-based software as a medical device: FDA: AI/ML SaMD.
Voices from the AMA
AMA leaders emphasize that physicians need a seat at every stage-from design and governance to day-to-day use-so tools stay clinically valid, ethical, and consistent with standards of care. Dr. Margaret Lozovatsky has reinforced that point: doctors should be full partners across the AI lifecycle to protect the patient-physician relationship.
Next steps
- Set up or refresh your clinical AI committee to align with AMA's governance approach.
- Audit current AI tools for privacy, safety, and workflow fit; retire what's not delivering value.
- Engage with vendors and the AMA's initiatives to share real clinical needs and outcome data.
- Plan ongoing training so staff can use, question, and escalate AI outputs appropriately.
Bottom line
AI will help medicine only if it works for the people delivering care. The AMA's new Center is built to make that happen-by centering clinical reality, raising the bar on safety, and cutting the noise so useful tools reach the bedside.
If you're building a clinician upskilling plan, these curated options can help you survey practical AI training paths: AI courses by job.
Your membership also unlocks:
