The White House Office of Science and Technology policy has released draft regulatory principles for the governance of AI. The memorandum proposes policy considerations to guide regulatory and non-regulatory oversight. Notably we also have a legislative definition of AI adopted from the John S. McCain National Defense Authorization Act:
“(1) Any artificial system that performs tasks under varying and unpredictable circumstances without
significant human oversight, or that can learn from experience and improve performance when exposed to
(2) An artificial system developed in computer software, physical hardware, or another context that solves
tasks requiring human-like perception, cognition, planning, learning, communication, or physical action.
(3) An artificial system designed to think or act like a human, including cognitive architectures and neural
(4) A set of techniques, including machine learning, that is designed to approximate a cognitive task.
(5) An artificial system designed to act rationally, including an intelligent software agent or embodied robot
that achieves goals using perception, planning, reasoning, learning, communicating, decision-making, and
Although there have been a couple of other legislative type definitions, it is fair to say that this one looks a lot more like a legislative definition than the others. I would not be surprised that greater adoption of this definition will follow and that it will start to guide other definitions. I’ll revisit the definition in a further post.
The policy proposes that the principles guiding AI regulation should engender 10 items: public trust in AI, public participation, scientific integrity and information quality, risk assessment and management, benefits and costs, flexibility, fairness and non-discrimination, disclosure and transparency, safety and security and interagency coordination.
A schedule to the memo proposes technical guidance on the rule making process.
The memo remains open for comment for 60 days, after which it will become final.
I was appreciative of this statement in the memo: “Given that many AI applications do not necessarily raise novel issues, these considerations also reflect longstanding Federal regulatory principles and practices that are
relevant to promoting the innovative use of AI.” As I have said before, not everything about regulating AI should be seen as requiring completely new facets of law.
Much like recent Canadian standards well laid out roadmaps like this memo should have significant effect. As AI may have to be dealt with legislatively by federal, provincial and state governments, a document that presents guidance that all can follow could ease unnecessary incongruity between different legislative efforts. If this leads to legislative efforts in the short term, we are likely to see the US setting the standards that others follow.
Photo by Aaron Kittredge on Pexels.com