Mini-Feature: Governing by AI
Would a government run with algorithms and AI be better or worse than the government we have now?
The answer to this question, which should fill everyone who ponders it with a kind of religious dread, can only be approached by beginning with a consideration of in what sense a social system run with algorithms and artificial intelligence—an automated system—could actually amount to a government. It strains the limits of out psychological tolerance to accept the suggestion that machines could govern people; rule by machine is hard to describe as governance, which rightly feels to us as inherently a matter in which humans play the definitive role on both sides of the governing relation.
But even so, people have managed to abstract the human role away from at least one side of the relationship, as over time in the modern era the idea that democracy would save us gave way to the idea that elites would save us, and then to the fallback notion that institutions or offices would save us, and finally now the terminal idea arrives that, all else having failed, we may have to deliver ourselves into the hands of disincarnate ubiquitous machines. After all, government comes from the Greek for the steerer of the ship, and a ship is a machine of sorts which surrounds the people aboard who depend entirely on its successful function.
So the question becomes what sort of government we might have if behind the automations was a person (or were people) with some degree of human autonomy. The easy answer is, a government incompatible with our way of life. The more difficult answer concerns the possibility of really restoring local citizenship and statecraft in the truly political sense by way of passing digital control at the fundamental level to small groups—a possibility still beyond reach, but in some ways growing ever closer.
-James Poulos, executive editor of The American Mind
Asking if we'd be better off being governed by a computer reminds me of what Dorothy Parker reportedly said on being told that Calvin Coolidge was dead—"How can they tell?"
Not to sound entirely cynical or blasé, but I fail to see how our current system of governance reflects human reason or the distillation of the people's will in any kind of coherent form. Would an algorithm do any worse?
One supposes that the system would have to be designed carefully and that the inputs be measured judiciously in order for it to function correctly. Following the principle that the government which governs least governs best, the USA.2 algorithm should ideally be designed to offer a null return as often as possible—that is, to recommend no action.
Most of the functions of government that actually impact people are fairly automatic—for instance, the processing of transfer payments and the collection of tax revenue. I suspect most of that is already performed by computers anyway, so there wouldn't be much difference if we just owned up to the existing conditions.
If the computer was instructed just to carry out existing laws and follow the Constitution, I can't see why there's any reason to object to that. For foreign policy we would just have to decide to dial up or down the tolerance for foreign meddling before air strikes are launched.
In a way, the US Constitution is a kind of algorithm already—a set of rules and conditions that can be iteratively followed. Who needs anything more?
-Seth Barron, managing editor of The American Mind
I spit on your filthy algorithms. The idea that they might ever govern us is a category error of the most amateurish kind, one that would be corrected by even a remedial education in political philosophy. Unfortunately our ruling elites are less than amateurs, because they are devoted at a level of principle to the arrogant notion that politics is about building systems which will “nudge” (read: deceive and coerce) people into behaving in unnatural but “efficient” ways. Eat the bugs. Blot out the sun. Watch the porn.
But “out of the crooked timber of humanity no straight thing was ever made.” And so the harder our naïve idealists try to stuff us into pods and force us into compliance, the more we dissolve into neurosis and chaos. The one thing none of these clumsy buffoons ever account for in all their stupid equations is human nature, which is incorrigibly surprising, perverse, passionate, and above all free.
So I suppose there is an argument to be made that a government actually ruled by bots would be better than the one we have now, which is ruled by flailing gremlins who think they can harness the power of bots but are really, before everybody’s eyes, being reduced by those very bots to dribbling incompetence. But that is precisely because the bots were not in fact made to rule us—we were made to rule the bots. Efforts to invert this relationship therefore inevitably buckle under the weight of their own dysfunction or else morph rapidly into totalitarianism of the most primitive and abusive kind. Ours seems to be doing a little of both.
The real solution would be, as James Poulos argues in the latest CRB, to recover a sense of our own ensouled humanity and moral reasoning as the governing measures for right and wrong—measures which must always, as Aristotle observed, be to some extent inexact (Nicomachean Ethics 1094b). We might, by halting and careful attempts at living well together according to republican principle, find ways of governing ourselves, of ruling and being ruled in turn, with digital technology serving as our handmaiden rather than tormenting us with dreams of subjection like some demented harpy. Just a thought. We’ll probably make a mess of things instead.
-Spencer Klavan, associate editor of The American Mind