A New Social Contract? Written by Whom?

Last week, when OpenAI CEO Sam Altman repeated his belief that AI will usher in such changes that states would require a new social contract, observers were rightfully alarmed. Altman supports a universal basic income, likely because he expects that AI will lead to mass layoffs as machines replace labor. He is not alone in predicting an enormous shift; Dario Amodei, the CEO of AI company Anthropic, has also warned of massive changes in the relationship between workers and technology. Without seeming to realize it, these predictions are hitting dystopic sci-fi notes. Amodei’s candid babble about where AI is headed included the bald admission that AI’s creators — these are the people with their feet on the gas pedal, remember — expect that AI will soon be “better than humans at almost everything.”

Although the tech oligarchy may, in their breathless excitement, be mistaken about this or that detail, the arc of technological development and their capture of state apparatuses around the world makes it tough to bet against them. Upheavals seem inevitable.

A shift toward an economy and system of production dominated by AI will indeed require a comprehensive reassessment of the social contract — Altman is correct in that regard. If new production technologies are swiftly and thoroughly adopted, they could displace workers across all sectors — from blue to white collar. It has become widely accepted that the AI “revolution” is essentially a new Industrial Revolution, one that promises to be as disruptive as the major industrial transformations that have recurred since the eighteenth century.

Of course, any revision of the social contract must include democratic oversight and control. Yet saying that it should is not the same as ensuring that it will. Controlling what happens next will require struggle and determination. And, no matter what the tech oligarchy tells us, it will involve class antagonisms at the state level and in the workplace. This will be at once a new and old battle.

Marx Was Right About Robots

Marx recognized in the nineteenth century that mechanization — a sort of proto-AI — would replace workers and serve as a tool for the capital class to exploit workers. Machinery intensified exploitation and discipline and deskilled labor, which contributed to alienation. Yet he also saw the potential for mechanization to free us from workaday drudgery. Under different social relations, such technologies and their associated productivity gains might free workers from endless toil, allowing us the use of newfound time for more fulfilling pursuits. But this outcome would depend, of course, on who controlled these technologies and to what end.

In the twentieth century, critical theorists — particularly some members of the Frankfurt School — identified both the promise and danger of contemporaneous machinery. Herbert Marcuse, echoing Marx’s insights, noticed this dual nature of technology. In Eros and Civilization, he argues: “The very progress of civilization under the performance principle has attained a level of productivity at which the social demands upon instinctual energy to be spent in alienated labor could be considerably reduced.” In other words, civilization was nearing the point where automation might liberate us. However, he remained skeptical. Who, after all, would robots work for — and against?

As I’ve argued before, robots won’t set us free unless we control the robots. We might own the smartest Roomba known to history and humankind, but we don’t own the factories that produce them, nor do we control — as is increasingly more important — the algorithms or programming that make them possible. We don’t control the chain of production that makes them possible, nor is anyone about to seize the means of production. But there are certain tangible things we can do right now.

Practical Steps for an AI Paradigm Shift

If we want a paradigm shift in production, we can arm policy with research and advocacy, driven by organizations advocating for public dialogue and oversight, like the Center for Human Technology and the Ethics and Governances of Artificial Intelligence Initiative. The work of developing and advocating for better state policy to restrain and direct the development and use of AI is critical. State regulation, and regulations across countries, remain utterly essential for producing better outcomes from a technology that isn’t going to disappear.

As frustrating as electoral politics is — rage-inducing, even — it is also the primary avenue for producing better policy. We can back candidates, referenda, petitions, and ballot initiatives that support reforms and legislative action that put AI to work for more people rather than fewer. These approaches are even better if complemented by broader theoretical strategies and projects to bring Big Tech to heel and make it work for us, like participatory AI and algorithmic accountability.

In Paris, ahead of a major AI summit, one researcher is preaching the power of transparency to guide decision making on AI. The European Union already has its proposed AI Act, which aims to assert some state control over the use of the technology, for instance, and Canada has a directive on automated decision-making.

We can also hold regulatory agencies accountable while struggling to ensure they have the teeth they need to bite when they need to bite. These agencies ought to hold regular public consultations on technological developments and hold strong oversight powers that are antagonistic toward industry, rather than letting tech oligarchs write their own regulations or, worse, operate without any at all.

Revisiting the Social Contract on Our Terms

Revisiting the social contract in the face of AI requires addressing two key dimensions: redefining the relationship between the individual and the state — what each owes the other — and rethinking the bargain between workers and industry, particularly regarding ownership and control. We ought to be inherently distrustful of any universal basic income schemes as dreamed up by the technocratic class — schemes that are all too likely to include a bare-bones approach that will dismantle what’s left of the social safety net. Instead, we need to determine what state support workers will require in the face of another industrial restructuring — and that decision must be driven by workers themselves.

Capital, by its very nature, will never use technology for the primary benefit of workers; only workers will. We must also fight for the democratization of the ownership of AI technologies and control over its deployment in industry. To the extent capital will ever care about using technology responsibly, it will be only to protect production and profit. When oil companies pretend to care, for instance, about the environment or mitigating the effects of climate change, you can bet it’s not for the sake of the poor suckers who stand to lose their uninsurable home to flood or fire. Rather, it’s to make sure there’s a market to sell to — and a human remainder, somewhere, with minimal viable standards of living to produce widgets or Roombas.

That disruptive potential of AI for both industries and the state is undeniable, and the need to renegotiate our social and political agreements will naturally follow. But how we do this isn’t preordained. It will entail a struggle over control that demands democratic direction in government and in the workplace. In that sense, the AI revolution presents a critical juncture and a powerful opportunity to build a better world — one in which the robots free us, because we control them for our collective good.