AI's User Manual: Lessons From the Constitution

AI Governance: Lessons from the Past

Those philosophers you may not have thought about since college, like John Locke and Thomas Hobbes, actually have quite a bit to teach Americans about navigating the world of AI, even though they predate modern computing by centuries. The same thinkers who influenced America's Constitution grappled with a key question that we are still facing today: How do you create a society where people coexist, despite disruptive new forces?

From the Constitution’s system of checks and balances to civil rights laws, America has consistently updated its social contract when new technologies have upset established systems. The telegraph, the internet and the automobile all necessitated applying old principles in innovative ways. For example, when cars presented challenges related to interstate commerce and safety, federalism wasn’t discarded. Instead, national traffic laws and highway standards were established while maintaining state authority over local roads. AI is likely to bring about a significant transformation of our society and economy, requiring similar innovative thinking. The minds that shaped the Constitution provide something akin to a user manual for the age of AI.

Social Contract Ideas for the AI Era

Here are ideas from the social contract tradition that may help navigate the socioeconomic issues that AI brings forth. Thomas Hobbes cautioned that life without rules descends into chaos, with everyone battling for their own advantage—a state he termed the “state of nature.” His proposed solution was a government capable of maintaining order. Today’s AI agents might be seen as Hobbes’ vision unfolding at digital speed. These systems have the ability to trade stocks, make decisions about hiring, and potentially perform most computer-based tasks. Without sufficient oversight, they function in a digital state of nature.

Just as the Constitution's commerce clause empowered Congress to oversee trade among states, federal oversight of AI agents may be essential. This includes systems designed to assign accountability when AI falters, safety protocols akin to those governing cars and medications and kill switches to guarantee human control. Instability, not freedom, is what you get when rules are absent.

One of John Locke’s most important ideas, which inspired the Bill of Rights, centered on limiting governmental power. The government cannot simply act as it pleases, even when in control. This becomes critically important when government entities employ AI in determining eligibility for benefits or assessing security threats. The Constitution contains no provision that states, “unless a computer dictates otherwise.” Citizens are entitled to the same safeguards that the Founding Fathers established against overreach: accessible explanations of AI decision-making, channels for appealing incorrect AI judgments, and strict limitations on surveillance.

Jean-Jacques Rousseau held the view that legitimate laws originate from citizens collaborating to address issues, rather than from elites imposing solutions. The American system integrates this principle through elected representatives, town halls, and the right to petition. Artificial intelligence could potentially be helpful in this area. Taiwan uses AI-driven platforms that help citizens find common ground on contentious topics and identify shared interests across the political spectrum. American communities could implement comparable tools to democratically resolve issues such as the use of facial recognition in public spaces or the appropriateness of using AI to grade students’ assignments. These are questions that are too critical for bureaucrats or tech executives to decide on their own.

John Rawls, who passed away in 2002, presented a thought experiment suited for the AI era: Design societal rules without knowing whether you would be rich or poor, employed or replaced by automation. Operating behind this “veil of ignorance,” you wouldn’t want to risk being among the losers. You would insist on ways to secure everyone's economic stability if AI leads to job losses. If algorithms are used to make hiring decisions, they should broaden opportunities instead of reinforcing existing disparities. Should AI generate substantial wealth, its advantages should extend beyond those who own the technology.

This isn’t socialism. It's preparing for an uncertain future. Concealed behind the veil, unsure if you'll be a tech CEO or an unemployed cashier, you would want an economy where everyone could prosper, not merely survive. Americans hold differing views on the direction of AI, with some envisioning a utopia and advocating for rapid advancement, while others foresee a grim future and push for a slowdown. Nevertheless, there are shared concerns. AI is evolving more rapidly than government can keep up, private firms are in control of the technology, and the future is impossible to predict perfectly.

AI governance could be founded on these points of consensus, despite differing political views. This could involve regulations that adjust to technological advancements, partnerships that leverage innovation while upholding accountability, or global cooperation on safety measures that prevent a race to the bottom. Political theorist Danielle Allen cautions that significant inequality undermines genuine democracy. Citizens who lack basic security are unable to participate as equals. This warning becomes more pressing as AI poses the threat of concentrating unprecedented power in the hands of a select few.

The Founding Fathers, who specifically designed the American system to prevent aristocracy and monarchy, would be alarmed if a handful of companies controlled AI systems capable of displacing millions of workers. America requires contemporary antitrust enforcement, systems that ensure affected communities have a voice in AI governance, and economic strategies that broadly distribute the benefits of AI. The central question is not whether to impede innovation but to ensure that it reinforces, rather than weakens, the democratic equality guaranteed by the Constitution. As technologies advance, the fundamental constitutional question is whether AI will be used to fulfill the promise of American democracy or whether it will be allowed to consolidate the type of concentrated power that the founders designed the Constitution to prevent. The philosophers might not have foreseen this future, but they have equipped us with the tools to deal with it. The question is whether America will be wise enough to use them.