ABSTRACT
Artificial intelligence (AI) is reshaping professional services and complicating liability design. We develop a framework where experts differing in hidden ability adopt AI and choose effort. AI lowers accident risk but raises compliance costs. Even with asymmetric information, strict liability induces efficient AI adoption and care but negligence induces shirking and inefficient adoption, unless punitive damages are allowed. Punitive damages can deter innovation. A mechanism-design approach shows that adaptive standards and scaled sanctions outperform uniform deterrence, which may backfire by suppressing precaution and innovation. Efficient policy must balance innovation, accountability, and risk using information-sensitive liability tools rather than one-size-fits-all enforcement.
Kim, Jeong-Yoo and Kim, Byung-Cheol, Artificial Intelligence and the Paradox of Professional Liability (October 17, 2025).
Leave a Reply