In Joseph Heller’s Catch-22, the anxious Colonel Cathcart is in the habit of compiling little lists of recent events. He draws these lists up under two headings: “Black Eyes” and “Feathers In My Cap.” Into the former, he puts everything that seems to damage his chances of preferment – bombing his own runway again, and so forth – and in the latter, everything that will redound to his credit.
I like to imagine Sam Altman, the boss of OpenAI, and his fellow AI overlords doing something similar. In the “Black Eyes” column, we might find near the top of the list the original sin of generative AI. Despite the efforts to handwave it away as a bygone, none of these models would exist at all were it not for the wholesale theft of the intellectual property of almost every writer or artist on earth. Consumer AI, pretty much all of it, is the fruit of a poisoned tree.
But there are plenty of black eyes in the outcomes, too. To list a few: the poisoning of the information ecosystem and the concomitant political destabilization; the vast expansion of surveillance possibilities for totalitarian regimes; the effective dismantling of secondary and higher education and the cognitive prospects of an entire generation; enabling financial fraud and the cost-free production of child sexual abuse material at scale; the prospect of transferring state-level capacity in bioweapons and cyberweapons to individuals with laptops; the precipitous decline everywhere and anywhere in basic trust between human beings – and also actual crimes against humanity such as the impossibility of removing Copilot from Microsoft Word.
Under “Feathers In My Cap”, I imagine we would find: “But protein-folding!” And “But decipherment by academics of certain very fragile ancient manuscripts!” And: “But swift interpretation of data from radiotelescopes!” Plus, in very small pencil letters: “A few of us getting ridiculously rich while all this bad stuff happens.”
So if such figures as the Pope and the King of England are willing to step in to say, essentially, “let’s hold on a minute with this AI stuff,” those of us suffering the actual black eyes – which is the vast majority of us – should be in favor. King Charles has let it be known ahead of his tech summit at Dumfries that AI development should attend to “principles,” and bear in mind “responsibilities” as well as opportunities, while never losing sight of “human benefits for communities.” That is, obviously, a great cloud of cotton-wool that says nothing much at all.
And yes, we should admit that His Maj, as a 77-year-old who has never owned a mobile phone, might have a limited locus standi for commenting on bleeding-edge tech developments. And, yes, we can have a good laugh at the briefing of the King’s position: “As we’ve seen him do on issues like the environment and water scarcity, he is hoping to provide a neutral space to aid conversations that are part of a national and global issue on something that is going to impact all of us.” A “neutral space” on environmental matters isn’t what most people will see him as having provided.
But in using his prominence to say, essentially, “I don’t understand the first thing about this stuff but I’m worried about it,” he speaks for all of us. He also speaks for the people who are supposed to understand this stuff, but also don’t, and are also worried about it. Given that a central point about AI is that it trains itself, and programs itself, and behaves unpredictably, not understanding this stuff is baked into the mix. We move in a great cloud of unknowing. And as witness the rather dramatic resignation of the Anthropic AI researcher Jacob Coxon, who says he thinks there’s a slightly more than one in ten chance of AI annihilating the human species in the next few years, the unknowing and the anxiety are connected.
But upstream of all this is something very decipherable in non-technological terms. On the face of it, the noisome forcible intrusion of unasked-for AI into every aspect of our daily lives – Google search, Copilot, every rage-making customer service interaction we have online – and the same companies telling us that AI is going to end the world as we know it are contradictions. But both stem from the same thing: greedy, predictable, irrational humans.
Stupendously vast amounts of capital have been invested in a technology that shows few if any signs of a sufficiently widespread consumer use-case to recoup that investment. So on the one hand, there’s a powerful interest in forcing take-up of the technology. And on the other, warning that the technology might just bring about our extinction adds to the hype. It says, more or less unsubtly, that the tech is so powerful even its creators are frightened of it, which is another way of keeping it in the headlines and boosting its importance and its stock price.
“We don’t want to be forced into a race” is another way of saying “if we’re forced into a race, you’d better keep backing us because imagine what happens if China wins.”
The thing that all these AI companies are frightened of, I suspect, is not a Terminator-style singularity so much as a boring old correction in the stock market. And the thing the rest of us should be frightened of is that they are prepared to lay waste to much of modern life to put that off as long as possible. Black eyes all round.
Comments