The Reverse Centaur's Guide to Life After AI summary

Have you read the new Isabel Allende novel, Tidewater Dreams, in which the Chilean author brings her trademark magical realism to the climate crisis? Or what about Percival Everett’s The Rainmakers, a futuristic neo-Western?

Your answer is no to both.

In 1950s America, a market researcher named James Vicary, claimed that during screenings of the film Picnic at a cinema in New Jersey, he'd flashed the words "Drink Coca-Cola" onto the screen for about a three-thousandth of a second at a time. The result, he announced, was a 18.1 percent jump in Coca-Cola sales. Moral panic ensued.

Vicary’s claims didn’t stand up to scrutiny. In fact, it’s doubtful Vicary’s marketing experiment even took place. But that didn’t quell people’s fears about the newly-named subliminal advertising. And the advertising industry was just fine with that: rumours that advertising was so powerfully persuasive it could bypass the consumer’s conscious mind entirely were good for business.

Ask the AI-doomers what their biggest anxieties are and they’ll feed you a dystopian scenario where artificial intelligence far outstrips human intelligence and, to put it bluntly, the robots have taken over.

But while the doomers are focusing on what overhyped catastrophes AI might cause in the future, current AI is causing real problems. Look at where the money flows. Real AI investment isn’t betting on robots that replace people outright. It's betting on "decision support," where an AI uses past data to guide human decisions in the present. Decision support currently helps doctors spot tumours on a scan, corrections officers set bail and parole, teachers grade essays, and HR workers sift résumés.

So we know AI executives fear-monger on purpose. And the louder we shout about how terrifyingly powerful AI is, the more we do free marketing on their behalf. The same logic applies to the outrage provoked around AI’s use in art.

To be clear: there is something outrageous about AI-generated art. Art is a communicative act, as well as a stylistic one. AI art can match aesthetic choices, but it can never communicate in the way human art does. What’s more, all these aesthetic choices that AI is making are the product of large-scale theft: the models that generate art were trained by scraping the work of living artists who didn’t give their consent and didn’t receive any payment.

The AI bubble will burst. But every bubble leaves a residue behind. Some of that residue is grimy and uninteresting. Take what remains of the crypto bubble: some teachable moments about security and a large quantity of worthless digital tokens. But when the dotcom bubble burst, back in 2000, there was plenty left to salvage in the residue. True, hundreds of overhyped internet companies went under. But the crash also left behind cheap office space, idle servers, and a pool of out-of-work engineers. These were the building blocks of the next generation of tech companies.

So, what will the legacy of today’s bloated AI enterprises be?