AI penetration in the health care industry is about to reach critical mass, as we hear from several AI subject matter experts in our
Part B News 2026 Predictions issue [
subscription required]. Here with additional commentary is
Alaap Shah, technology and data regulatory attorney and co-chair of the AI Practice Group at Epstein Becker Green in Washington, D.C.
What areas of health care will see use of AI increase in 2026?
Clinical support functionality will likely expand first into areas such as patient triage and screening recommendations, with AI becoming the “front door” to care by handling symptom intake, steering patients to the appropriate setting, and surfacing risk signals before a clinician engages.
Continuous patient monitoring solutions will likely mature from mere alerting to recommendation to intervention -- escalating, initiating workflows, and automating documentation in response to deterioration signals.
Other likely areas to see AI augmentation will be in clinical decision support and productivity related to ordering suggestions, care-plan drafting, and real-time summarization during patient visits.
What specialties are most likely to see an impact?
Computer vision-driven specialties such as radiology, pathology, dermatology, and cardiology will likely see rapid scaling of AI powered clinical workflows as a means to compress human workload and time pressures, which will likely boost efficiency and revenue capture to create material ROI.
What should AI adopters be careful about?
Overconfident clinical automation, where AI crosses from suggesting to acting without a human-in-the-loop, will continue to pose the highest risk. For example, hallucinated orders or prescriptions (without human review) could create patient safety risk and liability exposure that are likely to draw attention from the FDA, consumer protection authorities, patient advocacy groups, and other stakeholders.
Also: Bias and discrimination at scale resulting from AI algorithms will also remain a significant concern in the contexts of triage, resource allocation, and screening because flawed models can misdirect care for certain populations that have been historically underrepresented in the medical literature or ignored in establishing clinical best practices.
And “Shadow AI” -- unapproved tools, personal GPT apps, and undocumented automations -- will also continue to plague the industry. Use of shadow AI by clinicians will create compliance headaches and rogue workflows along with patient safety, regulatory compliance and reputational risks.
What’s the bottom line?
AI is decisively expanding beyond administrative tasks into the clinical flow of care, and success will hinge on reliable high-quality patient outcomes, continuous workforce training and upskilling, and strong AI governance mechanisms to detect and remediate issues early and often.