Dvir Aran (Technion – Israel Institute of Technology), Ronen Perry (University of Haifa – Faculty of Law), and Shahar Shelly (Technion – Israel Institute of Technology; Mayo Clinic – Department of Neurology) have posted Clinical Liability Cases Reveal a Coupling Between Legal Defensibility and Medical Procedure Escalation in Large Language Models (Communications Medicine) on SSRN. Here is the abstract:
Doctors increasingly rely on AI in making clinical decisions. This article tested whether AI-generated recommendations meet the legal standard of care (hence helping doctors avoid medical malpractice) and, if so, at what cost. It found striking variation in legal defensibility of AI-generated recommendations among different LLMs, particularly across different generations within each provider family. It also found that higher legal defensibility is generally (and unsurprisingly) associated with an escalation in medical procedure counts and costs but that some models utilize resources better than others, achieving greater (or similar) legal defensibility at lower cost.
Recommended!
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum