Improving Regulatory Oversight of Medical AI
Improving Regulatory Oversight of Medical AI
https://www.theregreview.org/2026/01/15/smith-improving-regulatory-oversight-of-medical-ai/
Publish Date: 2026-01-15 00:22:00
Source Domain: www.theregreview.org
- W. Nicholson Price II, a scholar from the University of Michigan Law School, is urging policymakers to reduce the role of clinicians in overseeing the use of artificial intelligence (AI) in health care.
- According to an American Medical Association study, two-thirds of health care professionals have started incorporating AI into their work but express a need for more oversight.
- Price contends that current oversight places too much reliance on “human in the loop,” requiring individual clinicians to review and incorporate AI recommendations, which is problematic due to clinicians’ lack of expertise and time.
- He argues that medical AI needs regulatory oversight because of design flaws, like algorithms that adapt to specific institutions’ environments rather than universally effective health solutions, and due to potential biases in data used to train AI systems.
- The current regulatory frameworks involve dual layers at the central federal level and local hospital level; however, clinicians are still left as the last line of defense against algorithmic mistakes.
- Price highlights that clinicians’ gaps in AI knowledge, automation bias, and heavy workloads hinder their ability to oversee AI effectively.
- In the short term, Price suggests defining clear roles for clinicians, providing onboarding and training, and offering institutional support to help them oversee AI without compromising their other responsibilities.
- In the long term, Price advocates for AI functions capable of operating independently without constant clinician oversight, reducing dependence on clinicians, and promoting access and democratization of medical expertise.
- He urges regulators to adopt evaluation methods that minimize clinician oversight during AI approvals and suggests that once released, AI systems should undergo random audits for performance validation.
- Price concludes that to improve health outcomes, regulatory frameworks must anticipate clinicians’ limitations, rather than assuming perfect AI performance.
