Source: NUS Law Research Blog

In this article, Tan Cheng-Han and Daniel Seng Kiat-Boon explore the intersection between agency law and artificial intelligence (AI). The rise of automated systems and artificial agents, relying on machine learning and neural networks, has led some commentators to suggest that AI could be treated as legal agents. These artificial agents replace human intervention by making decisions based on built-in models, but without personal desires or cognition.

While AI may play a critical role in processes that lead to legal liabilities, the authors argue that artificial agents are not true legal intermediaries in the same way human agents are. A legal agent must be a person, and until legal personality is granted to AI—similar to corporate legal personhood—AI cannot be considered a legal agent. Moreover, legal agents must have cognitive capacity, which AI lacks.

Instead of viewing AI as independent legal entities, the authors propose treating AI as instruments or tools used by humans or corporations. This "instrumentality principle" would allow liability issues to be addressed by recognizing AI as a tool rather than a separate legal agent. This approach would avoid the complexities of assigning legal agency to AI and still hold operators accountable for AI-related outcomes.

The article emphasizes that many legal issues involving AI can be resolved by recognizing AI as an instrument rather than conferring it legal agency status. As AI continues to evolve, the authors suggest that the legal system must adapt, but not by equating artificial agents with human agents.