Government should err on the side of caution when it comes to regulatory intervention in nascent industries like AI.
Government's approach should foster AI development, not thwart it.
P olicy-makers need to be very cautious before proceeding with anything other than a minimalist regulatory approach to AI. While early intervention in an industry may sound good in theory, the government is likely to make the wrong regulatory calls, prevent AI from reaching its potential, and put the U.S. at a significant disadvantage in the global AI race, a race that has economic as well as geopolitical implications.
Early regulation of the still emerging AI industry is likely to make the wrong calls precisely because the unknowns about the industry are greater than the knowns. How do we ensure that the technology operates safely if we don’t yet know what its main uses will be? How should we guarantee appropriate levels of transparency if the technology is still unpredictable and emerging? And how do we effectively address the market’s competitive environment when neither the main players nor the market has emerged?
Of course, congressional hearings and even litigation can unearth information about the current state of AI. But those efforts are typically inefficient and incomplete because it is far from clear where the technology is headed or even, in some cases, how it works. Even well-intentioned legislative and regulatory efforts in fast-moving industries like AI are usually outdated by the time they are implemented and can cause collateral damage.
Some of that collateral damage is that regulatory heavy-handedness would stunt AI progress. If AI innovators are bogged down in bureaucracy, their determination to thrive downgrades to an effort to survive. Investment also becomes less likely, as investors turn toward fields with less government resistance. French president Emmanuel Macron only recently warned us against regulatory overreach in AI and other fields when he lamented that the EU is “overregulating and underinvesting. In the two to three years to come, if we follow our classical agenda, we will be out of the market.” Learn from Europe’s mistakes; don’t repeat them.
While the AI technologies and some of their implementations are new in some sense, the business practices behind many implementations of it are old hat. Take, for example, algorithmic AI, which helps businesses better address dynamic decision-making, including pricing. Out of fear that the technology has been (allegedly) used to facilitate price-fixing, the Department of Justice has recently taken action against a property-management software company that uses AI to help set lease terms.
The reality, though, is that algorithmic decision-making has been in use for over 20 years. It has helped the government more effectively manage toll roads, or car purchasers find the best deals possible, without a price-fixing concern in sight. Simply because more advanced AI technologies are now involved should not be a basis for regulating what hasn’t been regulated before.
Legislators and regulators may counter that they are being proactive to protect competition as AI evolves. But incumbent firms have the resources to deal with the costs of regulations and the influence to shape them. Who gets hurt? Start-ups. And when start-ups suffer, so does innovation, because plenty of research shows that start-ups are often more innovative than incumbents. So when regulators intervene in a market too early, they may hinder innovation and blunt competition, the opposite of what they wish to achieve, especially given AI’s potential.
The global race for AI leadership is under way and fierce, with China in particular putting pressure on U.S. technologists to keep up. Many factors will determine how well the U.S. does in that global competition. But regulators’ appetite for allowing the still developing industry to, well, develop, will be a significant one.
None of this is to say that there should be no regulation, or that government should follow wherever technologists lead. But it does mean that government should err on the side of caution when it comes to regulatory intervention in nascent industries like AI. Only recently, the State of Utah created the Office of Artificial Intelligence Policy, whose mission is to strengthen trust in AI “through data-driven policy, timely regulatory adjustments, and innovation-enabling regulatory relief.” A similar federal approach, one that seeks to foster rather than thwart AI, is needed.