Three Inverse Laws of AI

Susam Pal proposes the Three Inverse Laws of Robotics as a set of guidelines for human interaction with AI systems, arguing that current practices risk uncritical trust and anthropomorphism. The first law cautions against anthropomorphising AI, warning that attributing emotions or intentions can distort judgement and lead to emotional dependence. The second law urges users not to blindly trust AI outputs, emphasizing the need for independent verification and the use of neutral language such as "queried" instead of "asked". The third law stresses that humans must remain fully responsible and accountable for decisions involving AI, noting that AI systems do not bear the costs of failure and that responsibility must fall on the people who design and deploy them. Pal illustrates how vendors can mitigate these pitfalls by adopting a more robotic tone and providing clear warnings about the limitations of AI-generated content. He also recommends adding automated verification layers, such as proof checkers or unit tests, for high‑stakes applications. The article concludes that while no finite set of laws can be foolproof, adopting these inverse laws can help maintain clear thinking about AI’s capabilities and limitations.

By framing the guidelines as laws that apply to humans rather than robots, Pal seeks to shift the focus from controlling AI behaviour to ensuring responsible human use. He argues that the current trend of treating AI as an authority is dangerous and that developers and users should treat AI as a tool, not a moral agent. The piece calls for industry‑wide adoption of these principles to prevent the inadvertent erosion of critical thinking and accountability in AI‑driven workflows.