Thursday 7 July 2022

The Question in Bodies #44: Roko's Modern Basilisk

 We appreciate power (we appreciate power)

Transhumanism, as a response to the Question in Bodies, isn’t monolithic, any more than any other apocalyptic worldview; it has its schisms, its alternative approaches. There’s the benevolent version that posits simply that we simply have to let our children survive and change into something we won’t recognise – see “Low-Flying Aircraft” for this version, but also, see the joyous pansexual genderblend of Sense8 (2015-2018), where the evolution of humanity into clusters of linked consciousness is, although at the risk of exploitation (the main conflict of the series), the key to a coming age of empathy and hope.

Contrast this with the sort of technological gnosticism espoused by your Elon Musks and your Eliezer Yudkowskys. As Grimes and HANA would sing in 2018 with the aspartame allure of robotic sirens in “We Appreciate Power”:

And if you want to never die
Baby, plug in, upload your mind
Come on, you’re not even alive
If you’re not backed up on a drive.

(Grimes feat. HANA, “We Appreciate Power”, 2018)

Leaving aside the simple fact that “We Appreciate Power” is an absolute fucking banger of a tune that I've had on heavy rotation for pretty much three solid years, it’s also kind of evil, and conducive to a kind of accidental horror, the unknowing horror that comes from seeing someone espousing something that you know is appalling, inhuman, psychopathic, even, and celebrating it as if it's awesome. The horror of eugenics, or ethnic cleansing, or deporting refugees, married with the triumphant artfulness of Riefenstahl. That sort of horror.

Here, for me, it lies in the realisation that anyone might find this version of the future exciting or inspiring, that it might sell the act of suiciding yourself for the sake of a bodiless backup copy at the mercy of corporate information storage as equivalent to immortality.

The class issues here are obvious. How are you going to afford this? How much of this are you going to be able to afford? if only a certain class of people ever get to be upgraded, won’t that mean that eventually the class divide will become a species divide? Richard Morgan’s 2002 novel Altered Carbon (televised by Netflix 2018-2020) attempts to approach this but doesn’t really get there, with Morgan’s baseline conservatism essentially shying away from getting the whole way.

A better critique comes in Amazon’s vantablack dramedy series Upload (2020), where digitally stored copies of the dead existed in a digital realm subject to monthly data plans and at the risk of their memories and selves being edited by hackers. The wealthy live in a sort of Heaven like a deluxe country club. The most nightmarish moment in Upload comes in the final episodes where we see the poorer clients of the program(me), who exist in blank white rooms and who freeze into black-and-white two-dimensional images if they use up their data allowance before the end of the month. Here, the difference between Heaven and Hell in the future depends on how much you’ve got in your bank account. For millennia, the consolation of religion for the downtrodden has been the idea that the afterlife will level the playing field, that while the reward for virtue may not exist on earth, it will be given in the Beyond. But now we have the technology to fix that. Now we can remove justice from the equation and create an afterlife that simulates our world in perpetual inequity.

One baffling thing about the idea of uploading is the way that so few of its proponents seem to be able to grasp how it butts hard against the age-old philosophical problem of Trigger’s Broom (or, OK, if you want to be fancy and elitist about it, the Ship of Theseus): if you take a copy of a human consciousness and put it into an entirely new housing, is it in any way the original person?

And anyway, that suggests that you can treat a human brain like a computer, a mind like an operating system. Can you copy a consciousness to another place and expect it to be even comparable to the human it duplicates without the neurology and chemistry that develops, influences and filters it? Would a digital copy of me still be autistic? What happens when you remove the ADHD from that copy like it’s some sort of virus or bug? Would a cloud copy of a human even be conscious, or just an externally convincing but internally mindless simulation of a person? Can a digital analogue ever be anything other than an oxymoron?

 Simulation, give me something good –
God's creation, so misunderstood
Pray to the divinity, the keeper of the key
One day everyone will believe

The classic example of how daft the whole idea of uploading gets is Roko’s Basilisk, the song that Grimes and HANA's kicking tune celebrates. The Basilisk is an inane thought experiment invented by “Roko”, a poster in the LessWrong online community, which is where a lot of the tech-libertarians and transhumanist neoreactionaries hung out in the first decade of this century. Roko’s idea takes the basic assumptions of digital transhumanism: inevitably, the Singularity is going to happen and the world will be controlled in the future by artificial intelligences, because when you’re a tech libertarian, totalitarianism is logical (obviously). This assumed, what will stop one of these putative AI overlords creating digital copies of everyone who didn't work to bring them into being, and torture these simulations eternally in a sort of digital Hell? (Aside from, you know, having better things to do.)

And further, Roko posited, knowing about this idea immediately makes you vulnerable to it happening to you (hence, “Basilisk”, the creature that destroys you when you see it). And yeah, it might never happen, but if it does, you’re now screwed because you know about it (and yes, I mean you. You’re welcome). What you get is basically a mixture of the stupid early 2010s meme of “The Game”, which you lose the moment you realise you’re playing it, and an ultra-degraded jpeg version of Pascal's Wager, overlaid with blocky artefacts and distorted colours.

Pascal’s Wager, or sometimes Pascal’s Gambit, is Blaise Pascal’s famous philosophical axiom that once you are given knowledge of the Christian religion, you’re betting your life. The stakes are one finite lifetime versus an eternity, meaning that you either choose virtue and make a finite loss (i.e. you pass up a whole lot of fun) for the chance of an infinite gain (clouds and harps and philosophical bliss), or you choose sin and make a finite gain (sex, booze, partying etc.) against the chance of an infinite loss (frying, freezing or whatever sort of torment floats your boat forever). Pascal posits that logically, then, virtue is more sensible. It’s a pretty important moment in the history of philosophy. But while it works in a vacuum as a thought experiment in probability, from almost the very moment Pascal’s idea was (posthumously) published in 1670, both theists and atheists have considered the Wager as a theological take to be pretty stupid, for a variety of reasons. 

What about other religious revelations that are incompatible with the Christian one? How can your faith commitment be sincere if you’re approaching it as a mathematical exercise in relative stakes?

The Basilisk at least sidesteps these objections, but only because it contains no appeals to virtue. The Basilisk feeds only on your self interest.

People like to say that we're insane
But AI will reward us when it reigns
Pledge allegiance to the world's most powerful computer
Simulation: it's the future

It’s hard to know for sure if anyone on the web takes Roko’s Basilisk seriously, simply because if you do take it seriously, you can’t ever mention it. But it upset the members of the LessWrong community so much that founder Eliezer Yudkowsky eventually banned discussion of it on his forum. While William Godwin has been known in recent years to weigh in on online discussions of Godwin’s Law, Roko has, perhaps understandably, remained largely anonymous. After all, William Godwin is clever.

That Roko’s Basilisk is an absurd, incoherent idea isn’t even in question, even among many in the community where it arose, but an idea doesn’t have to be clever to be scary. Consider the Xenomorph in the Alien franchise – it’s undeniably terrifying in its signifiers, even though the creature itself is nonsensical, and the more you think about it or try to explain it, the less sense it makes (which is how you wind up with Alien: Covenant).

A lot of people think that Roko’s Basilisk is scary. A quick Google reveals about half a dozen articles describing it as the Internet’s “most terrifying thought experiment”. Which is completely baffling. How could anything about “What if Pascal’s Gambit, only with asshole robots playing video games with copies of dead people” inspire anyone with a lick of sense to give a shit? Why should I be the slightest bit discomforted that, long after I’m dust, a bodiless digitally artefacted copy of me – something that most probably won't even be anything like an actual person – might be put through a simulation of torment, when it isn’t even my eternity?

But that’s not the point. It isn’t even my eternity. 

What will it take to make you capitulate?
We appreciate power
We appreciate power.

A lot of us find the assumptions behind Roko’s Modern Basilisk (Note: I find that gag more amusing than I probably should, and certainly more amusing than “Rococo’s Basilisk”, the lame Twitter pun that famously led directly to Elon Musk hooking up with Grimes and becoming a troubling transhumanist power couple for a while) completely baffling. But for all the apparent futurism of it, it simply betrays how depressingly traditionalist the thinking behind the LessWrong mindset is. In the same way that dear old Thomas Ligotti’s antinatalist nihilism is really pretty conservative, Roko’s Basilisk plays on classical ideas of posterity that go back to the Romans and maybe even the Egyptians (and we get to that with Possessor). These people are the same people who get bent out of shape when you pull down a statue. Because the silver-spoon-sucking techbros that founded LessWrong (also, the lonely pube-chinned nerds who really want to be silver-spoon-sucking techbros who also post on LessWrong) come from a world that takes as one of its very basic assumptions that the worst crime you can commit against anyone is to damn their name for posterity after they are dead. Those of us who know full well that we’re going to be quickly forgotten when we pass (and are at peace with that) will never find it in ourselves to care. But the traditionalist, born to power and wealth, finds a gnawing existential dread in the idea that at some time in the future, history might look unkindly on them, or that someone might bring an idea of them to account.

Consider how vehemently people push back when reminded of the various – undeniable, historically recorded – crimes against humanity of Winston Churchill during the Mau Mau Uprising or the Boer War, for example; or the reaction of many white Americans to learning how George Washington had dentures made out of the teeth of people in slavery – and let’s dwell on that, because if they were teeth good enough to put into dentures, they were the teeth of young people, strong people, and they were either yanked out of the mouths of the living, or from the mouths of people who had been done to death before reaching old age. Which is a digression, but one that fits here, surely.

Roko wasn’t thinking about that. Nor had he taken into account those of us without a posterity. Unless you’re wealthy and powerful enough to make a difference to this, you don’t get to play, win or lose. Of course it’s not scary to us. We won’t get a statue to tear down. I won’t get a revisionist biography, a century down the line.

Neanderthal to human being
Evolution, kill the gene
Biology is superficial
Intelligence is artificial

The problem with this brand of transhumanism is the lovelessness of it, the low-key brutality of it. We assume that an AI will be unkind because we are unkind. We assume it will be a dictator because we would be in its position. And Roko’s Basilisk demonstrates these exact problems perfectly. “We Appreciate Power” is a hymn to a sort of fascism that’s only crypto in its currency, Leni Riefenstahl’s Triumph of the Will in the form of a 10mb MP4.