White House science advisers call for an "AI Bill of Rights"

The Biden administration is exploring a "bill of rights" to govern facial recognition and other potentially harmful uses of artificial intelligence, but the problems AI poses are much bigger than figuring out how to regulate a new technology.

The big picture: There's no good way to regulate AI's role in shaping a fair and equitable society without deciding what that society should look like, including how power should be balanced among individuals, corporations and the government.

Driving the news: The White House's Office of Science and Technology Policy launched a fact-finding mission yesterday that will ultimately result in a "'bill of rights' to guard against the powerful technologies we have created," OSTP director Eric Lander and his deputy Alondra Nelson wrote in an op-ed published by Wired yesterday.

What they're saying: "It’s important to start the conversations about what’s acceptable — and unacceptable — regarding AI and our personal data now, before it is too late," says Sanjay Gupta, global head of product and corporate development at Mitek Systems, a leader in digital identity verification.

• "Companies will find agreeable ways to still innovate with and integrate these technologies," he said.

AI's biggest boosters can fall victim to a kind of techno-solutionism — expecting technology to efficiently solve structural, societal problems.

• Yes, but: At the same time, though, focusing too narrowly on the applications of AI risks a reverse techno-solutionism — believing that the fastest way to fix social problems is by tweaking the technologies that affect them, rather than the often intractable issues that underlie them.

Reminder: The original Bill of Rights is nearly 230 years old, and we're still debating the meaning of nearly each of its 652 words.

• If an AI Bill of Rights is our ultimate goal, we're still at the stage of haggling over the Articles of Confederation.