Originally published at: How the Writers Guild sunk AI’s ship – Pluralistic: Daily links from Cory Doctorow
After a grinding, 148-day strike, the Writers Guild of America ran the table, conceding virtually nothing and winning virtually everything.
The most consequential outcome will be data on streaming viewership. For the studios, these numbers are state secrets, revealed on a need-to-know, burn-before-reading basis, even within the studios themselves.
Hiding streaming data has two obvious benefits to the studios. First, it lets the studios keep creative workers in the dark about the success of the shows they work on, so that their agents and unions can’t bargain for higher pay on the basis of having knocked the last one out of the park.
Second, it lets execs hide how many dismal failures they greenlit from investors and stock analysts, whose rage and incredulity would send the share prices tumbling and lead to calls for the board to replace top management.
So the fact that these numbers will finally be disclosed is a Big Hairy Deal, but that’s not where everyone’s attention went on this strike.
For both writers and their supporters, the most visible deal-point wasn’t streaming data, it was AI.
For the AI sector, a 148-day news-cycle about how AI threatened writers’ jobs was a gift from the heavens.
It’s not just that there’s no such thing as bad publicity — it’s that this bad publicity was very good for AI’s narrative of being the most powerful, amazing and transformative technology since fire, or possibly language.
Here’s how that works: “If the writers are convinced that AI can write scripts so well that they will all be out of a job in five years, then you should absolutely invest in AI technology, because it is obviously incredibly powerful.”
And here’s how that works: “If the writers are convinced that AI can write scripts so well that they will all be out of a job in five years, then you should absolutely buy an enterprise site-license so that you, too, can fire all of your workers and replace them with a chatbot.”
The term of art for this kind of bad publicity is criti-hype, wherein a critic is suckered into repeating the booster’s own claims, but as criticism.
Think of the Prodigal Techbros’ claim that they created a mind-control ray in the process of refining big data-driven ad-tech, learning how to “hack our dopamine loops” in a way that bypassed our free will and turned us into puppets who’d buy whatever they were selling.
Critics repeated this outlandish claim — “these sorcerers made a mind-control device to sell your nephew a fidget spinner, then let evil billionaires use it to turn your uncle into a QAnon — and in so doing, made the ad-tech huckters’ pitch for them.
“See? You should absolutely pay a 40 percent premium to run Facebook ads, because — as even our critics admit — we are sorcerers who will hack your customers’ dopamine loops to make them buy your products!”
The AI sector is utterly dependent on criti-hype. They are burning tens of billions of dollars on engineering salaries, custom chip fabs, human data annotation, data-center rents, racks and racks of GPUs and ASICs, whole gridsworth of electricity and entire aquifers’ worth of fresh water for cooling.
To keep that reservoir full, the AI industry needs to convince fresh rounds of “investors” to give them hundreds of billions of dollars on the promise of a multi-trillion-dollar payoff.
That’s where the “AI Safety” story comes in. You know, the tech bros who run around with flashlights under their chins, intoning “ayyyyyy eyeeeee,” and warning us that their plausible sentence generators are only days away from becoming conscious and converting us all into paperclips.
It’s pure criti-hype: “Our technology is so powerful that it endangers the human race, which is why you should both invest in it and use it to replace all of your workers.”
This form of criticism is entirely distinct from the legitimate realm of “AI ethics,” whose emphasis is on how bad AI is at the things that will supposedly generate those promised trillions. Things like bias, low-quality training data, training data attacks, data ordering attacks, adversarial examples, the endless stream of confident lies, and the high degree of supervision they necessitate.
Add to that the exploitative labor pipeline, the environmental damage, and the public safety risks and a very different critique emerges —one that’s grounded in AI’s shortcomings, not the supposed risks arising from its incredible power.
The studios fought like hell for the right to fire their writers and replace them with chatbots, but that doesn’t mean that the chatbots could do the writers’ jobs.
Think of the bosses who fired their human switchboard operators and replaced them with automated systems that didn’t solve callers’ problems, but rather, merely satisficed them: rather than satisfying callers, they merely suffice.
Studio bosses didn’t think that AI scriptwriters would produce the next Citizen Kane. Instead, they were betting that once an AI could produce a screenplay that wasn’t completely unwatchable, the financial markets would put pressure on every studio to switch to a slurry of satisficing crap, and that we, the obedient “consumers,” would shrug and accept it.
Despite their mustache-twirling and patrician chiding, the real reason the studios are excited about AI is the same as every stock analyst and CEO who’s considering buying an AI enterprise license: they want to fire workers and reallocate their salaries to their shareholders.
The writers new deal doesn’t ban using AI in the writers’ room — instead, it bans paying writers less whether or not AI is in the loop.
Which is fine — for writers. There’s doubtless ways that AI tools can help writers at the margins. Writers have been using oracles, tarot cards, dice, and decks of cards to stimulate their creative juices since the year dot.
Having a florid liar produce plausible sentences on demand is just another way that writers can take a directed randomwalk through the infinite problem-space of making up stories about imaginary people that our brains are tricked into caring about.
But if the studios can’t use AI to fire writers and replace them with chatbots, they have no reason to shell out millions for enterprise licenses.
And those enterprise licenses are absolutely vital for the AI companies. Mister Market isn’t going to keep pouring billions into Big AIs’ cash reservoirs forever. Eventually, they’re gonna have to rug us, get us to stop talking to each other and start buying things.
When the AI companies inevitably shut off the expensive supply of free amusing chatbot queries, individual writers might pony up a few bucks a month for a throttled trickle of stochastic parrot queries.
But the personal amusement price-tier can’t sustain the incredible burn-rate of AI companies, and it certainly can’t pay off for the suckers who handed billions to the eyeball-snatching would-be evil sorcerers who promised to make them all rich well before they were dissolved into grey goo.
Now, there are plenty of other workers in the film-production pipeline that bosses are hoping to make unemployed so that their salaries can be rendered unto the shareholders.
Hollywood is one of the most heavily unionized sectors in America, but there still gaps where workers are atomized and bosses can act without fear of a strike.
For example, there’s the visual effects (VFX) workers, famously unorganized, whose wage-bills are a big chunk of any film budget. These workers already have a heavily automated pipeline driven by sophisticated computer technology, so perhaps they can all be fired and replaced with software?
Well, maybe (especially if you don’t care whether the VFX the software produces is good, and only care about satificing).
Which is why Disney/Marvel’s VFX workers just voted to unionize. What’s more, if Disney doesn’t give them a contract ASAP, the hard-charging general counsel of the National Labor Relations Board is going to simply force the company to honor the terms that other unionzed workers get through their contract.
Unions are back, baby, and the NLRB has instituted a new, zero-tolerance policy for union-busting. Everyone is unionizing, from strippers to Starbucks baristas, and the writers just proved that no matter how horny your boss is for replacing you with an AI, a union will cool his ardor.
This is big takeaway from the writers’ strike, the part that goes beyond the entertainment sector. It’s the pin that pops the AI sector’s hype-balloon.
Because even if you believe their hype that AI is capable of putting 13 trillion dollars’ worth of workers on the breadline, the writers’ strike makes it clear that workers won’t let it.
Remember, this is the true message of AI: not that it can do amazing things, but that it can shift value from workers to owners.
That story gets wrapped up in the mythology of “technological progress,” as though there is only one way that a technology can be used. Margaret Thatcher’s dictum that “there is no alternative” was always a demand dressed up as an eternal truth.
What Thatcher meant was stop trying to think of an alternative.
There is nothing foreordained about how — or even whether — a technology is deployed in our society. That what a technology does is secondary to who it does it for and who it does it to.
This is always and forever up for grabs, and the writers just grabbed it.
And in so doing, they established the playbook for every other group of workers that AI shills hoped to use as props in their big con.