The White House Blueprint For An AI Bill Of Rights – Time For More

The White House has released their Blueprint for an AI Bill of Rights. What it does is provide a great overview of the issues involved in the responsible use of artificial intelligence (AI). What it doesn’t do is provide anything prescriptive for the administrative and legislative branches of our government. I would have preferred the White House produce a legislative blueprint rather than an introduction similar to those that have been published elsewhere.

The blueprint is very well written and clear. The first few pages are an overview of the five core principles identified by the authors as critical to societal protections against misuse of AI. Those five are:

· Safe and effective systems

· Algorithmic discrimination protection

· Data privacy

· Notice and explanation

· Human alternatives, consideration, and fallback

The introductory sections have very clear, high-level descriptions of each of the factors. The rest of the document provided more detailed explanations and examples. The longer sections are misnamed the Technical Companion. They’re not technical and those who wish to know more about this issue shouldn’t be at all scared to read the expanded information.

The problem is in that generality. For instance, in the section about safe and effective systems, there’s the sentence “You should be protected from inappropriate or irrelevant data use in the design, development, and deployment of automated systems, and from the compounded harm of its reuse.” Why, yes, we should. Independent organizations such as the Organization for Economic Cooperation and Development (OECD) and trade organizations are appropriate for stating such vague things, but I expect the government to provide stronger guidance and suggest legislation.

There are a few references to executive orders, the Privacy Act of 1974 (1974?), and similar. What do we need to do to update and broaden the laws for the ubiquitous nature of today’s computing and underlying data use? For instance, the paper mentions Executive Order 13960, which “requires that certain federal agencies adhere to nine principles when designing, developing, acquiring, or using AI for purposes other than national security or defense.” Why only “certain federal agencies”?

I’ve seen enough white papers and articles suggesting concepts for the responsible use of AI. It’s up to our governments (federal, state and local) to do something. The published blueprint is a nice overview of the issue, but that’s not what should be happening at the federal level. This is an issue that should be able to cross the aisles of our increasingly split political arena. The White House should be doing more to work with Congress to create clear legislation to implement the blueprint.