The deadline for the Hinternet Foundation fall online course on the Voynich Manuscript is almost here. Click below to find out more about it, and be sure to apply by September 15!
You will by now have read of the 27-year-old former Anthropic employee Jacob Coxon, who coupled his resignation with a series of somber posts on X. “The people building AI earnestly believe that it could kill us all by the end of the decade,” he warned, adding that his own estimate of the likelihood of such an outcome is greater than 10%. It’s hard to say when exactly the tipping-point came, in the long slow progress of AI doomerism out of marginal LessWrong forums and into the mainstream. But over just the past few months, and decidedly by the time of Coxon’s public interventions, it really does seem that everyone is talking about it.
You might also recall that in the previous decade similar predictions were made by prominent figures in the fight against anthropogenic climate change. One 2018 retweet from Greta Thunberg, in particular, seemed to online right-wing mobs to communicate the prediction that we had at most until 2023 to solve the problem. When 2023 came around, Charlie Kirk and others seized the moment to point out that we were all still here.
Now I recognize that there are tremendous disanalogies between the two issues, yet I can’t entirely convince myself that there is nothing at all to learn from looking at doomer scenarios in a comparative light, or that doomerism is not perhaps in the end a generic condition of the current moment in history, prone to attach itself to different objects in turn, while consistently failing to take itself as an object. That is, doomers do not habitually reflect on the social and political conditions that make doomerism as such attractive, but instead interpret their attraction to it as a direct rational response to a given concrete threat.
I happen to think that if there is any threat for which doomerism is a rational response, it is neither climate change nor AI, both of which are almost certainly going to continue to cause us major problems, and worse ones than we’ve known so far, in the coming years. It is, rather, war.
States go to war with one another periodically, and it’s been over eighty years since the last big one. That war arrived at its climax when the Americans introduced the world to a very new and different kind of weapon, in August, 1945. Soon after, a handful of other countries acquired the same destructive technology, and the post-war global order took on the character of a stand-off, fundamentally the same, if at a massively larger scale, as what you find in a hostage crisis. If you are 81 years of age or younger, you have lived your entire life under the reign of an unrelenting terror that we still find so exceptional and unacceptable when it takes the form of a single deranged man strapped with a vest of dynamite, but that somehow just becomes a background feature of everyday reality when that man’s role is assumed by the state.
Over the early post-war period there were several near-misses. Now the post-war period is over, and even the pretense of rational deliberation that that era’s institutions upheld —notwithstanding the absolute barbarian madness of the world those institutions pretended to be leading— has disappeared. There is at present more than one global empire explicitly and unapologetically committed to the position that if anyone tries to bring down their current governing regime, they will bring down the rest of the world on their way out. You might say that’s just theoretical, but these same global empires are, as a matter of fact, at the same time, actively working to bring down the governing regimes of the other ones. They are trying their best to do this without letting things get out of control, of course, without stumbling on any tripwires or retaliating in a way that invites escalatory retaliation without end. But this is at best a delicate balance, and there simply is no good reason to suppose that it can be maintained by the relevant state actors indefinitely.
In this light, if a race of rational beings were to visit earth, and we were to query them as to what we must do to keep ourselves together, plainly they would reply: “Ideally all, and no fewer than all but one of the relevant state actors must unilaterally surrender their weapons.” And if we were to say, “OK, but what if the other parties refuse?” the aliens would reply: “It doesn’t matter!” And we would say, “But they don’t have a multi-party system!” or, “Look at their human-rights abuses!” At which point the aliens would righteously interject: “Enough, children! Nothing else matters but disarmament. Now quit wasting time and flip a coin.”
Of course, all doomer scenarios are connected. Climate change exacerbates political instability. Autonomous weapons systems are, obviously, one of the things, in addition to power grids and banks and hospitals, that we don’t want rogue AI taking over for its own ends. But the relative rarity of my own “preferred” doomer scenario at the top of the lists that others make of near-future challenges strikes me as explainable only in terms of some kind of mass-psychological mechanism of repression.
This was not always the case. Much of the freak-out of the 1960s, I have come to believe, may be seen as the cultural processing of the trauma of the Cuban missile crisis and other such close calls. I myself can remember The Day After (1983), a work of televisual schlock that, for all that, at least faced up to the real threat looming over us, and in so doing managed to trigger real reckoning on both sides of the Cold War divide, which may even have played some indirect role in landing Reagan and Gorbachev at the Reykjavik summit a few years later.
I gather there was some other such television series over the past few years — House of Dominoes, I think it was called? Starring Idris Elba? Who knows! Like almost everything else in our current information environment, the cultural reckoning this time around was essentially undetectable. Oppenheimer (2023) came and went, and was instantly elided in the popular imagination with Barbie as its twinned summer blockbuster. The worlds depicted in both cases were processed, one had the impression, as roughly equal in the respective forms of their unreality, either because the characters lived and died in the 20th century, and thus the problems they faced fit entirely within the frame of our own rearview mirror; or because they were made out of plastic, had no genitalia, and were 11 inches tall.
I admit that it was a slow-emerging epiphany, but I think it was in the summer of 2023 that I finally became fully convinced of something that remains no less clear to me today: It’s the world that’s fucking crazy, not me.
When the prospect of a future history-ending war comes up at all, the tone of engagement is almost psychopathically inappropriate. Consider this article in the New York Times, also from 2023, the tone of which is something like “a fun new way to think about global thermonuclear war, for all you fans of Liu Cixin!” You can almost picture the efforts in the newsroom to fit this bizarre piece in alongside such comforting pap as the wifey-softcore “Modern Love” column, or Kwame Anthony Appiah’s frivolous Miss Manners larp in “The Ethicist”, in a way that would permit the casual reader to move back and forth among these offerings without any real cognitive or existential distress. In this way the Times, even when it covers the unspeakable, upholds the illusion of comfort, the implicit assurance that even if this is a bumpy ride we’re on you can continue to trust the people in charge. Theodor Adorno would detect this same function in the horoscope section of your typical Cold War American newspaper.
This illusion is a powerful thing, and almost all Americans live out their lives entirely under its sway. I know a good many politically progressive Americans who really, truly just never think about the fact that the United States is a global power concerned to maintain that status at all costs. When they process news about US military entanglements around the world, they tend to ask questions like, “Should we keep playing the role of global policeman?”, thus taking the state’s official line on what it’s up to out there as basically true. This official line has it that our own country is not like the other ones, is not like any that has ever existed in the history of the world, as its actions are best studied in light of its ideals, rather than in light of the same material and at least somewhat law-like forces to the consideration of which we naturally revert when the subject is the Romans, the Mongols, the Aztecs, or indeed the Soviets. And nowhere is this exceptionalism more evident than in the nearly universal presumption among Americans, or at least those of them living within America’s borders, that their lives will never be seriously impacted, or ended, by war.
I have found that things are different in Europe, and I imagine they are different in Asia too. Yet in their colonial arrogance my progressive American friends never cease to share with me their mostly fanciful plans to flee Trump’s US for peaceable Europe,1 as if there were any plausible scenario on which life in the US would become truly intolerable for domestic political reasons, while life on this continent would go on as before; as if the post-war makeover of Europe as a just and peace-loving place had nothing to do with American projection of its power here, for better or worse; and as if America’s recent chaotic retreat from that role could be expected to have no serious consequences for the ongoing viability of post-war European eirenism!
I can’t entirely shake the suspicion that there is at least some correlation between the sense of comfort and security that is maintained as the ideological frame of life at the heart of the American empire, on the one hand, and on the other the displacement of the apocalyptic imagination onto what are ultimately less important things.
Again, the actual challenges we face are likely going to involve the mixture of more than one existential threat, so that it can easily appear almost indecent, or at least irrelevant, to seek to rank them. At the beginning of the Scientific Revolution, Francis Bacon implored his contemporaries to stop being squeamish and to force Nature to give up her secrets, in the expectation that once they became known to us we would be better able to control her (the gendered language here is Bacon’s). The result of this “hacking through nature’s thorns”, to adapt a line from James Merrill, is indeed a great deal more power, and more responsibility for the management of destructive forces, than the premoderns could have anticipated (unless you are willing to accept the reality of prophecy). In this light it is somewhat myopic to see nuclear weapons as exceptional, since in fact we have pushed nature up to its very limit several times, and in several different domains — notably, the science of chemistry has given us microplastics, and it has given us synthetic fertilizers that disrupt the planet’s nitrogen cycle, to cite just two modern novelties from that field that come quickly to mind; and biological research has brought us to the point where we must also worry about the constant threat of a new global pandemic beginning from a lab leak.
We just keep doing it, I mean: bringing ourselves right up to the limit. In this light what is so intriguing about AI doomerism is that it now shows us, in effect, after a lag of almost a century, that it is not only the early modern period’s effort to mathematize nature that can “kiss awake new powers” (also Merrill) and ultimately destroy us; the mathematization of thought, too, beginning with the first invention of formal languages capable in principle of automated processing, leads us ultimately to the same place.
That we should be most impressed right now with this most recent result of our efforts is understandable — it’s newer, and, on the face of it, stranger. But nature has not been superseded. She’s still there (to speak with Bacon), and she hasn’t forgotten what we’ve done to her. And even if the newest doom scenario arises from the realization that at a certain point automated thought becomes automated will, for which there is no guarantee that it will remain aligned with our own, still there is at least for now plenty of human will left in the world. Considerable power to exercise will is concentrated in sovereign states, which psychopathically take their own continued existence as equal in importance to the continued existence of the world as such. I tell you this remains the greatest danger to humanity.
AI doomerism might be justified, or it might be an ingeniously orchestrated hype campaign by savvy insiders, or something between these two. Time will tell. But either way it is the sovereign power of human beings to end human life by choice, and not the accidental transfer of our power of choice to machines we initially created merely to assist us, that should remain at the front of sober minds right now. If you don’t see that, it’s probably because you’re more in thrall to the American-defined narrative of history than you know — which has always let its subjects indulge at least a bit of their taste for disaster sci-fi, as long as they never forget who, at the end of the day, is in charge.
The ones in academia usually come to understand how fanciful this talk is the minute they learn what my French university salary is. Some of them have noted that it would be more viable for them to move to Paris without expecting to land an academic position here, but instead to continue on a quarter-time basis at their American university, commuting across the Atlantic for six weeks or so a year. It is at that moment also that the alternative scenarios, on which the US descends into pure dictatorship and closes its universities, or on which it’s pure civil war and chaos there, tend to give way to a shrug and a stoic declaration to the effect that “I guess I can stick it out.”