Categories
cybernetics

Artificial Intelligence: Changing Everything Except Ourselves

Artificial intelligence promises to change the world. The trouble is that we’re using it to preserve the world we’ve already got.

There is a peculiar absurdity in watching the world’s largest technology companies spend hundreds of billions developing artificial intelligence, only to explain that nobody quite knows how to manage the consequences. Apparently, we can build machines capable of extraordinary feats of reasoning, prediction and synthesis, but asking them to help prevent the economic, ecological and social destruction accompanying their deployment is a bridge too far. The problem isn’t insufficient intelligence, it’s what that intelligence has been employed to optimise, whose interests define success, and who is obligated to carry compound consequence, cost and catastrophe when the machinery goes sideways.

The difficulty is that AI is being developed within a corporate and economic structure whose participants largely imagine the future as an extension of the present. They can conceive of extraordinary technological innovation, provided the ownership, wealth, authority and institutional arrangements remain comfortably familiar. Established structures have a remarkable capacity to embrace transformation while resisting changes to themselves. AI is consequently employed to optimise the machinery that already exists, including the inequalities, dependencies and vulnerabilities it reproduces. The limits of technological imagination become the limits of institutional self-preservation.

I’m sympathetic to those who want to tell the entire AI industry to fuck off. But here’s the uncomfortable proposition: use AI to dismantle the conditions that make AI dangerous. Use it to expose dependencies, model systemic risks, interrogate institutional incentives, explore alternative economic arrangements and reveal possibilities that existing concentrations of wealth and authority have little incentive to contemplate. Not another automated authority deciding what’s best for humanity, but an instrument for transforming the relations through which authority, dependency and control reproduce themselves. The intelligence required to challenge the machinery may already be inside the machinery. The trick is to stop treating the world we’ve inherited as the only world worth optimising. Fuck AI. Use AI. Put humans back in the loop.

An inability to use artificial intelligence for the common good is inseparable from an unrepentant determination to preserve the world, its status hierarchies, and its revenue bandwidth, as they are. The technology offers the possibility of radically restructuring our societies, yet the corporations, governments and institutions directing its development resist the axiomatic, foundational changes that would make this possible. Many of the problems we attribute to AI are consequences of that resistance. Once again, we are demanding that the future accommodate the past.