AI will upend lives. Society needs guard rails while waiting for Congress to act.

America must begin to regulate artificial intelligence carefully, before it shapes everyone’s lives in perhaps unimaginable and irrevocable ways.

Yet in Congress, little has emerged in the way of strong legislation, even though more than 1,400 tech leaders wrote an open letter in March seeking oversight of AI technologies because of the potential “profound risk” to society.

While Congress dawdles, the Illinois Legislature should see what it can do to guard against AI abuses, just as state lawmakers stepped in to protect residents’ biometric information privacy when Congress failed to do so.

The Legislature took a step in that direction when it passed the 2020 Artificial Intelligence Video Interview Act, which requires employers who use AI in the interview process to notify applicants beforehand, explain to applicants how the AI evaluates them and obtain consent. The law also prohibits unnecessary sharing of videos and requires their deletion by all parties within 30 days if asked to do so by applicants.

‘We should all be thinking about it’

So-called artificial intelligence already is providing many benefits, and it promises to yield many more in the future. On Wednesday, it was announced that the Air Force Research Laboratory for the first time used AI to fly an XQ-58A Valkyrie jet without a crew. AI can reduce human error in many tasks; take over monotonous chores; do jobs that are hazardous to people; be the presence behind chatbots; and analyze big data sets more quickly to reach conclusions.

But critics worry about its powerful ability to upend society. AI can make it easier to impersonate people in writing, communications and images. It can manipulate people by spreading propaganda and perhaps tilting elections more effectively than was possible in the past. It can increase the power of the surveillance state, and when its algorithms make serious mistakes, people will suffer. It also could, and likely will, eliminate jobs.

”I think we should all be thinking about it,” said state Rep. Kelly Cassidy, D-Chicago.

But regulating fast-changing AI systems won’t be easy. How do you allow for maximum benefits while building in protections for civil liberties and privacy? These are issues lawmakers grappled with even before AI.

For example, Matt Erickson, a longtime executive in the online privacy space, points to a past case in which a man’s insurance rates quietly went up because he subscribed to a motorcycle magazine, even though he was a quadriplegic and could not operate a motorcycle.

AI can exponentially multiply such hidden effects on people. It has the potential to make decisions that affect individuals in various ways, without people realizing how it alters their lives.

“Washington is still trying to come to grips with it,” Erickson told us.

Controlling the revolution

Some efforts to get out in front of the oncoming AI revolution are underway.

In May, the National Science Foundation announced it would spend $140 million to launch seven new national AI research institutes.

In June, Illinois Attorney General Kwame Raoul and 23 other attorneys general asked for federal policies requiring consumers be informed if companies are using AI and that federal regulations take into account how AI affects ordinary Americans. The attorneys general also want companies that use or develop high-risk AI to submit to external audits of their systems. In addition, the letter pointed out the need for federal AI legislation on such issues as privacy.

In July, President Joe Biden said seven U.S. companies voluntarily committed to ensure their AI products are safe before they release them.

But with something as fast-growing, varied and quickly evolving as AI, a single new set of regulations, guidelines or commitments won’t be enough. Rules will have to be updated as fast as AI changes, and regulators will have to be agile and vigilant.

Cary Coglianese, a professor at the University of Pennsylvania law school and an expert on AI issues, said Congress should enact a law clearly authorizing and requiring regulatory agencies to address AI within their domains — something some already are doing to a certain extent — and to set a deadline for doing so.

Congress should also leverage industry expertise to identify risks and establish a clearinghouse to monitor what agencies are doing, Coglianese told us.

Getting all the benefits artificial intelligence offers while avoiding its dangers will be a challenge. The time to put wise and effective oversight on the fast track is now.

The Sun-Times welcomes letters to the editor and op-eds. See our guidelines.

Send letters to letters@suntimes.com.