No Plan

 Our ruling class has been sneering for months at the local data center backlash that has been springing up across the country. Tech companies ask impatiently who they have to "bribe" to get their centers built in small towns across America. 

As Jasmine Sun says of tech executives in a recent interview: "when I talk to them about the build-out, I’m not hearing any sort of personal moral reckoning with slowing A.I. down. I’m hearing: How do I make the bribes bigger? How big do they need to be?"

And as for our political overlords, they too appear to have decided that new AI infrastructure has to be constructed at all costs. 

There was a time when I was still struck by what appeared to be a discrepancy between the importance of AI to our economy and (potentially) our collective future—and the fact that it had not yet really been politicized as an issue (at least not in our usual binary partisan fashion). 

Now, that steadily appears to be changing. Trump has come out hard in favor of AI "accelerationism," dismissing fears about rapid development as a "hoax." Reportedly, he fears than any slow-down in AI growth could jeopardize one of the few things keeping the stock market afloat. 

Trump's administration has also promulgated new rules that would make it harder for local communities to block or weigh in on new data center construction. 

Yet, there is a tremendous irony in the fact that the same elites who have been sneering at the data center backlash and dismissing it as mere reactionary Ludditism for months are now throwing up their hands and begging government regulators to save them from their own handiwork. 

The same tech companies who have been trying for months to circumvent local democratic review and consultation procedures in order to build more AI infrastructure are now the ones admitting publicly, "we're not on track to keep these machines under control; please save us from ourselves!"

AI executives have been saying for months, if not years: "our technology might displace most if not all human employment, driving people into mass unemployment and poverty if the government doesn't step in to do something about it. Also, it might exterminate all humankind after attaining superintelligence." 

Ordinary Americans understandably respond to this by saying: "that sounds terrible! Don't build that thing near me! We don't want it!"

And the same tech companies who have just been giving them many good reasons to hate and fear their new technology roll their eyes and complain bitterly about the unthinking backward-looking technological conservatism of the masses who resist change. 

Then, as they proceed to develop their technology, they discover that they have lost control of it. They brought about the very thing they warned against. And so now, they turn to the state and say: "please fix this! We're not even sure how!"

I am reminded of a passage in Robert Tressell's The Ragged Trousered Philanthropists. He observes that the Edwardian working class was always told not to trouble their heads about politics or business, because they couldn't hope to understand such complex public affairs. 

Leave those matters to your "betters," they were told. Leave them to the "practical, sensible, men of affairs" who understand such things.

But as soon as the working class turned to these alleged "betters" and "practical, sensible men"—it turned out that the latter "simply did not know what to do, or how to do it," as Tressell puts it. 

"These are the 'practical' men; the monopolists of intelligence, the wise individuals who control the affairs of the world: it is in accordance with the ideas of such men as these that the conditions of human life are regulated," Tressell writes: 

And yet, "[w]hen the Practical, Level-headed, Sensible Business-men are asked why they do not remedy this state of things, they reply that they do not know what to do! or, that it is impossible to remedy it!"

That seems to be where we're at at present in the AI debate. The government is trying to prevent local communities from having a say in building the physical infrastructure of the new technology. They are telling Americans: "don't try to interfere with something you don't understand. Trust your betters." 

But then, those betters turn around and tell us: "we have no idea what we're doing either, help! help! Please do something to slow or prevent this technology from being built!"

As Ezra Klein quotes the person running "alignment" efforts at one of the current leading AI labs: "we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to."

These are the "practical" people, the "monopolists of intelligence," the "betters." And they are telling us quite openly: we "simply do not know what to do, or how to do it." 

Apparently, no one is flying the plane. 

For a long time, this situation made me simply fatalistic about AI development. I thought: if even the people building it have no plan for what happens if this goes wrong, I guess we're screwed. I kept quoting Martin Amis's great line about nuclear weapons: "I'm just going on about them, I don't know what to do about them."

But ordinary Americans apparently decided: if no one else has a plan for how to control this technology and keep it safe, I guess we're just going to have to step in and keep it from being built! 

They said: if even the people building this are crying out that someone needs to "pace" them—then, by all means, let us pace them! Let us stop them from building the physical plant they need to speed up the creation of this technology!

And Americans said and did all this in the face of extraordinary condescension and belittlement from thought leaders, the executive branch, tech oligarchs, and all their alleged "betters." 

Perhaps, when the "betters" are telling us they have no plan and are all out of ideas, that means it's time for the non-betters to have a crack at it. 

When the "betters" are saying: please stop us and save us from ourselves; perhaps we should praise the non-betters for doing so. 

Tressell writes in another passage of one man from Mugsborough: "Being a householder the man had a vote, but he had never hitherto taken much interest in what he called 'politics'. In his opinion, those matters were not for the likes of him. He believed in leaving such difficult subjects to be dealt with by his betters."

Tressell goes on to observe: "In his present unhappy condition he was a walking testimonial to the wisdom and virtue and benevolence of those same 'betters' who have hitherto managed the affairs of the world with results so very satisfactory for themselves."

So too, in our time, I think it's becoming very clear that our tech overlords—despite their public protestations—did not really have a better plan than the rest of us for how to develop AI while ensuring that it would "serve human interests," as all their mission statements said.  

It turns out that they operated, foreseeably, as profit-maximizing ventures typically do, with "results very satisfactory" to their own short-term enrichment—but without any discernible progress toward figuring out how a machine capable of automating all human cognitive labor and hacking every digital system will not in fact prove socially disruptive. 

I talked in a previous post about how public skepticism toward new technology is often seen as a weakness of democratic and liberal institutions. Some have suggested that in order to "win the race" against our authoritarian adversaries we have to adopt some of their methods—making it easier to "just do things" and build whatever needs to be built, regardless of community pushback. 

I quoted in the earlier post Robert Moses's use of this argument to defend the displacement of an Indigenous community from their land in order to build a hydroelectric project, for example—as Edmund Wilson chronicles in his great book, Apologies to the Iroquois.

As Stephen Holmes shows in his book The Anatomy of Antiliberalism, this is a longstanding critique of liberal institutions dating back to Carl Schmitt and other unsavory characters: liberalism is accused of "debility" and of obstructing decisive action. In the place of "just doing things," parliamentary institutions, it was said, produce only "endless chatter." 

(As Holmes points out, this argument for "decisionism" and critique of liberal "debility" is somewhat hard to reconcile with the fact that liberal parliamentary governments beat totalitarian ones in both the Cold War and World War II—including the totalitarian regime that Schmitt himself served.)

AI accelerationists often say the same thing of the data center backlash and local environmental review and community consultation standards today. How are we supposed to beat China in this new arms race, they demand to know, if we can't even build things? 

But now, the same people who wanted so earnestly to "just do things" are also throwing up their hands and admitting "we have no idea what to do." 

Meanwhile, the American public did not submit for an instant to the fatalism and submission that was preached to them. They said: if you don't know what to do about it, let us tell you. If you are saying this technology is dangerous, then we will by all means help you stop it from being built. 

They showed much more ability for collective action in this regard than I was capable of, or than any of our elites were apparently capable of. They stepped in to stop AI construction at the ground level. 

And wasn't this their own democratic way of "just doing things"? Didn't this show much more capacity for decisive action than our elites were apparently capable of, with all their dithering? 

The very institutions—local community review and consultation procedures, etc.—that were accused of making only for "endless chatter" and obstruction—turned out to be the only vehicle in this entire AI debate for anything that even approached a concrete step toward practical solutions. 

So much for our democratic processes producing only "endless chatter"! All the chattering and dithering, in the end, seems to have come from our elites—the "betters." Whereas the people who instantly sprang into action to block the next data center from coming to their community at least actually did something. 

To quote Stephen Holmes's suggestive retort to Carl Schmitt's theory: "Why not conceive the irrepressible right of rebellion as an expression of liberal decisionism?"