Exploring AI in Healthcare: Legal, Regulatory, and Safety Challenges

Stanford Legal

Artificial Intelligence holds the potential to transform much of our lives and healthcare professions are embracing it for everything from cost savings to diagnostics. But who is to blame when AI assisted healthcare goes wrong? How is the law developing to balance the benefits and risks? In this episode, Pam and Rich are joined by health policy expert Michelle Mello and Neel Guha, a Stanford JD/PhD candidate in computer science, for a discussion on the transformative role of AI in healthcare. They examine AI’s potential to enhance diagnostics and streamline workflows while addressing the ethical, legal, and safety challenges this new technology can bring. The conversation highlights the urgency of adapting regulatory frameworks, the complexities of liability among hospitals, developers, and practitioners, and the need for rigorous testing to ensure patient safety as AI integration in healthcare advances.

Connect:

• Episode Transcripts >>> Stanford Legal Podcast Website
• Stanford Legal Podcast >>> LinkedIn Page
• Rich Ford >>> Twitter/X
• Pam Karlan >>> Stanford Law School Page
• Stanford Law School >>> Twitter/X
• Stanford Lawyer Magazine >>> Twitter/X

Links:

• Michelle Mello >>> Stanford Law page

(00:00:00) Chapter 1: Understanding AI in Medicine
The episode begins with a broad introduction to AI's applications in medicine. Neel Guha explains generative AI systems and their rapid advancement, including practical applications like chatbots, imaging, and decision-making tools. Michelle Mello highlights AI's widespread integration, from diagnostic tools like radiological imaging and predictive algorithms to administrative uses that aim to reduce physician burnout.

(00:07:04) Chapter 2: The Benefits and Risks of AI in Healthcare
The group explores the advantages of AI in medicine, such as enhanced diagnostic precision, reduced administrative burdens, and improved patient outcomes. Michelle Mello identifies potential risks, like automation bias, where reliance on AI might lead to unchecked errors, highlighting the tension between time-saving tools and maintaining human oversight.

(00:08:22) Chapter 3: Legal Challenges and Liability in AI-Driven Medicine
The conversation turns to the legal implications of AI in healthcare. Neel Guha outlines scenarios where AI contributes to patient harm, discussing negligence claims, product liability, and the complexity of determining accountability. Michelle Mello and the hosts analyze how liability standards might evolve, comparing AI's systematic errors to human fallibility and addressing the interplay of human-AI collaboration in preventing mistakes.

(00:14:47) Chapter 4: The Challenges of AI and Transparency in Decision-Making
The group explores parallels between medical and anti-discrimination fields in understanding machine learning's opaque decision-making. Neel Guha delves into the evolution of AI systems from rule-based programming to complex machine learning, emphasizing challenges in identifying points of failure across stakeholders like hospitals, physicians, and developers.

(00:17:35) Chapter 5: Regulation and Liability of AI in Healthcare
Michelle Mello discusses the regulatory framework for AI as a medical device, comparing outdated 1976-era regulations to modern challenges. The conversation shifts to gaps in tort liability and the risks of developers limiting their liability through contracts. Proposals for redistributing liability to incentivize better private governance are examined alongside the need for robust AI quality assurance akin to crash tests or clinical trials.

(00:23:13) Chapter 6: The Road Ahead: Balancing Innovation and Safety
The speakers analyze the distinct challenges of regulating AI across diverse healthcare environments. Neel Guha and Michelle Mello discuss adapting evaluation practices to align with AI's real-world complexities. Optimism prevails as Michelle highlights AI’s potential to address critical issues like diagnostic delays, advocating for guardrails to ensure safety without stifling innovation. The episode concludes with reflections on Stanford's interdisciplinary approach to these pressing issues.