a suppressed desire is acted out in symbolic form.
I think this confuses "attention-grabbing" with "desire to enact." Most violent video games are much closer to an enacted serial killer fantasy, yet these days we don't say that this reflects an innate desire for people to be mass murderers.
I agree that sadism, suicidality, and aggressive dissatisfaction can generalize in apocalyptic directions.
It seems like you don't explicitly argue for it, but there's a sense from what you wrote that you argue that people in general have apocalyptic desires, particularly ones strong enough to act on, rather than anomalous individuals. If so, I don't think that is right? Or at least, on the margin, we're concerned about anomalies - before you get to the point where a flash of anger could actually end the world, people with a much more sustained desire would've done it first when it was harder.
2I do think that people want to be mass murderers – but this desire is suppressed very well. There are only a few movies in which nobody is killed. There are other activities – like hunting, doomscrolling, or video games – that channel this suppressed desire so well that we have only a few real serial killers.
My view is that the desire to kill, as well as the desire to destroy the world, is built into most people, but it is also effectively suppressed and openly expressed only by some individuals or in special situations. The main problem is that a suppressed desire can change our behavior without our noticing. In the field of x-risks, this means that some decision-makers may make riskier bets than would be rational.
I don't know what violent video games y'all have been playing, but the ones I'm familiar with could not really be described as "serial killer fantasy" but rather "combat fantasy". A serial killer doesn't engage in combat against rival killers on a battlefield; they stalk and murder noncombatants in society. The impulse to do battle and triumph over rivals seems pretty different from an impulse to abduct and murder innocents.
That makes sense. I think if OP argued that many humans have a latent desire to be a warrior, fighting for a just cause, then they'd have a stronger case. And that could still get most of the way to what they're worried about, when it comes to asymmetric catastrophic risks!
In my introspection experience, I have both things and they are different. When I was in kindergarden (and was abused there) I wanted nuclear war to destroy it - and make me free. In school I wrote a novel about how all human disappeared from earth. But I also have a lot of more normal phantasies about gun battles, tanks etc.
Yeah. I'm not even sure the "just cause" is an essential element there. It can just be about who wins, who's better. That's what it is in competitive sports too; which are often described as a sublimation of a warrior impulse.
You know, I'm not sure. Spec Ops: The Line famously had you be the bad guy, deeply morally culpable, and that turned a lot of people who just wanted a combat fantasy off of it. I think a belief that what you're doing is either not real, or is necessary or good, might be a critical component - as I understand it, not believing in what you're fighting for, with no kin bond with your other soldiers to replace it, translates directly to low troop morale.
I do think that people want to be mass murderers – but this desire is suppressed very well.
May i present to you:
https://www.lesswrong.com/posts/baTWMegR42PAsH9qJ/generalizing-from-one-example
It is not my internal experience to want to be a mass murderer.
What I actually observed here is the popularity of books and movies about mass murders - but exactly I don't like them.
Hmm. How do you tell the difference between "most people want to kill but it's suppressed" and "most people don't want to kill, they just find killing very attention-grabbing"? What different predictions do these make? Maybe there's a way to falsify one of these statements.
There's the case of escalatory revenge raiding/killing and kin feuds, which were a historically common cause of killing - but those had to do with ingroup/outgroup dynamics. There's also infanticide, quite common historically - that seems like a premier case of doing something you don't want to do for other reasons. Even for state-driven wars, those were about coordination failures, and elites and societies stealing land and resources from each other, not about murder for murder's sake.
In situations where killing is encouraged, the initial reaction is reluctance - armies have to work really hard to get their soldiers comfortable with killing, and active genocides are carried about by people in really particular circumstances, with drugs, ideology, peer pressure, and authority influencing them - and so on.
So I'm not sure that "most people harbor a desire to be mass murderers" explains this well? You argue that those situations represent a release of the "suppression" - but if the suppression is sticky, and snaps back into place when people leave the contexts where it was suppressed, I think it's just a simpler explanation to say that people generally don't want to kill unless they're actively pulled into it? Otherwise you could argue anything that only occurs in humans in special situations is just a suppressed natural desire - glossolalia, self-immolation, circumcision, etc.
TL;DR: Popularity of the movies about apocalypses tell us that the end of the world is a very attractive idea. There can be several psychological explanations for this unconscious desire. This increases x-risks.
There are many movies about the end of the world. While most end well, they are in some sense similar to stories about serial killers: a suppressed desire is acted out in symbolic form.
Myths about the end of the world exist in all regions of the world. Note that there is a difference between the desire to end the world and cognitive biases in estimating its risk.
While the desire itself is mostly fulfilled via art and literature, it may affect the behavior of some people. Existential terrorists are possible (some sects), but more interesting are the unconscious changes in behavior that can lead a scientist to perform a dangerous experiment, such as gain-of-function research.
There is also a desire to create self-evolving agents, possibly related to the instinct of procreation, which becomes especially dangerous when combined with the desire to end the world. Human interest in gain-of-function research and in recursively self-improving AI are manifestations of it. Recall ChaosGPT, which was one of the first looped agents and was created with the explicit goal of destroying the world.
There are several sources of the desire to end the world:
• Generalized but suppressed aggression (for example, I hated my kindergarten and dreamed that a nuclear war would start and destroy it) — a general dissatisfaction with the whole world, my place in it, and how things work.
• Similarly, suicidal thoughts can generalize into the desire to end the world.
• Sadism and the "pleasure of killing" can also generalize into world-destroying ideation.
• The desire for world transformation — the dancing Shiva, the myth of the deluge, and so on, as well as socialist ideas about revolution.
Desire to save the world and the feeling of self-importance
The desire to save the world (the opposite of the desire to end it, but psychologically close) and the idea that it is "precisely me" who lives in the era when the world ends can stem from a feeling of self-importance and an overblown illusion of personal uniqueness.
Of course, if I really am unique, I am more likely to be in a simulation; but if the illusion of uniqueness is widespread, it negates the update toward the simulation hypothesis.
Conclusion
Understanding the psychological roots of the need for the end of the world is not an academic exercise. In an era when the technological capabilities of a single person approach those of states, and in some cases exceed them, the destructive motivation of a single individual may have civilizational consequences.
And if we are training AI on human values, what about the need for the end of the world? Is the fear of an aggressive superintelligence not someone's secret dream? Could it be that an AI that has learned our values will also assimilate the need for the end of the world? An AI aligned with a preference for global destruction is equivalent to a misaligned AI.