Why Washington Is About To Repeat Its Worst Regulatory Mistake

It’s harvest season across the Midwest, which means combines comb neat rows in the fields while politicians race into the midterm election to collect votes.

Two decades ago, in 2005 and more ambitiously in 2007, Congress decided it knew how to fix America’s energy problems. With the country at war and gasoline prices climbing, families were feeling the squeeze at the pump. New laws required billions of gallons of ethanol to be blended into our fuels. Corn that once fed people and livestock went into gas tanks.

Food prices soared. One World Bank economist estimated that biofuels in the United States and Europe drove roughly three-quarters of the global increase. A law sold as making fuel affordable and assuaging consumers’ fear made food unaffordable. The bill hit the world’s poorest households hardest.

The World Bank estimated that the price spike pushed 100 million people into poverty, and the U.N. Food and Agriculture Organization counted 75 million more people going hungry in 2007 alone. Emissions eventually fell because of an innovation no mandate anticipated: fracking. The mandate is still on the books.

Policymakers in Washington are about to repeat the renewable fuel standard’s mistakes twice over for AI by letting fear drive policy, and in an election season, nothing sells as well as fear.

Why does fear so routinely drive policymaking? Psychologist Roy Baumeister and journalist John Tierney call it the power of bad. We are wired to feel bad news more strongly than good news.

We need about four good experiences to offset each bad experience. Every alarming story about a chatbot or data center outweighs the thousands of quiet, unseen gains in medical diagnosis, personal tutoring, and productivity. Legislators react to that imbalance too, and they respond with laws to meet the alarmism.

The first mistake follows directly. Just as with ethanol, Washington is rushing ahead before anyone has clearly defined the problem. The legislative process, with its fact-finding, hearings, and debate, is designed to name a problem and weigh responses. Congress has not worked that way in a long time, and it has certainly not examined how artificial intelligence fits the laws we already have on liability, copyright, and fraud.

Before new laws pass, supporters should show that the harm is real, that it is widespread or persistent, and that competition, the courts, and current law cannot remedy it. When legislation is still needed, it should set the standard for safety or security and let companies race against each other to find the best path. Medicine, autonomous cars, art, and national defense raise different questions. One sweeping AI law answers none of them well.

The second mistake is permitting 50 legislatures to enact conflicting laws for technologies used and provisioned nationally. States passed more than 100 AI laws in the first half of this year alone. A start-up in Florida must now answer to rules hastily drafted in Sacramento, Albany, and Hartford before it ships a product. That is a tax on start-ups and entrepreneurs and a gift to the few big firms that can keep compliance lawyers on retainer.

It is also a gift to the Chinese Communist Party. By comparison, Chinese developers answer to a single regulator. Congress should do two things: preempt most state AI laws and examine how these technologies fit with existing statutes before writing new ones. A Justice Department task force or an agency policy statement is no substitute.

None of this means the fear and worries are imaginary. Harms from AI will emerge, and someone must answer for them. Much of that can happen under laws already on the books. A chatbot that defrauds a customer is still committing fraud. A hiring algorithm that discriminates still violates civil rights law. A defective self-driving car still exposes its maker to liability. Courts have adapted time-tested doctrines to new technologies from the railroad to the radio and the Internet. They can do it again.

The human costs of a bad law are more difficult to see than the political costs of a bad headline. Families facing food insecurity in 2008 never made the news in Iowa, where corn growers cashed in, or in Washington, where lobbyists had moved on. No one writes stories about the treatment that was never developed, the stronger building material that was never invented, or the start-up that never got off the ground. Lawmakers ignore what they cannot see. That is the power of bad at work.

Critics of preemption assume that more rules from more legislatures mean more safety and accountability. Usually, they mean less. As Washington prepares to repeat its costly ethanol mistakes on the frontier of artificial intelligence, it risks trading real, unearned fear for unseen, generational loss. To avoid the tragic, invisible human costs of the past, Congress must let the courts apply time-tested doctrines and have the patience to regulate with clear objectives, or once again, we risk starving the world of tomorrow to ease today’s political anxieties.

***

Kent Lassman is the president and CEO of the Competitive Enterprise Institute.