If the AI Bubble Pops, It Won’t Be the End
Like past waves of automation, AI isn’t going away. Boom or bust, the fight is over whose interests it will serve.
Whether the AI hype deflates or investment keeps pouring in, capital is committed to forcing AI into the workplace, reshaping jobs and lowering labor standards. (Alex Wong / Getty Images)
A report from MIT says that the overwhelming majority of generative artificial intelligence implementations are failing — 95 percent of them, in fact. That’s a lot of failure. For a series of technologies touted as the answer to every question ever asked, literally and figuratively, the incapacity of companies to effectively integrate AI into their work says a lot about the next stage of industrial mechanization, our economic system, and us. But what it doesn’t say, whatever one may hope, is that AI is doomed.
The trick here is that you’ve got to read beyond the headline. The MIT report didn’t conclude that AI doesn’t work, as flawed as it is, but rather that many integrations aren’t working, at least as of now. As Jowi Morales puts it for Tom’s Hardware, integrations are failing to produce returns “because generic AI tools, like ChatGPT, do not adapt to the workflows that have already been established in the corporate environment.”
All-In on AI
The corporate response to these findings, or to similar ones, won’t be to abandon AI, but rather to force it even further upon workers and workplaces — reshaping established workflows, routines, and jobs themselves to fit the AI, rather than the other way around. Setbacks shouldn’t fool anyone about the direction of travel.
Over at Meta, for instance, Mark Zuckerberg is freezing hires in the AI division. Scale AI, in which the Facebook parent company has invested over $14 billion, is cutting 14 percent of its staff. The headlines are fueling talk of an AI slowdown or crash, of the bubble bursting. The pace and scale of AI investment in recent years has been accompanied by breathless talk of AI revolutions, of the imminent and irreversible shift to a new technological and industrial order. One is hard-pressed to click more than twice now while browsing the internet or opening an app or service without encountering some AI integration, from word processors to retail shop chat bots to social media and search engines and beyond. It all has the air of a Ponzi scheme on the brink of being exposed.
That scheme is already leading to tens of thousands of job losses, with more expected. The risks for workers are real. As companies go all-in on AI, even should they scale back a scooch, they remain committed to transforming work through automation — that is, to improving “efficiencies” and eroding both labor power and cost. AI is poised to kill millions of jobs around the world sooner than one might think. Yet its vaunted automation, efficiency, and autonomy rest on vast amounts of human labor — moderation, annotation, data labeling, troubleshooting, repair, and more. AI may render countless jobs redundant, but it also creates and transforms others. The catch is that many of these new jobs are miserable, and the overall shift drags down labor standards and pay.
Nonetheless, it won’t matter how much anyone insists that the AI emperor has no clothes; the hundreds of billions invested in AI have established, and will maintain, an internal hope, logic, and expectation that the series of technologies are and must be the future of knowledge, manufacturing, and service. Faced with AI’s limits and the attendant enshittification, as Cory Doctorow might put it, of so many aspects of life, one might expect technocrats and oligopolists — corporate and political alike — to pause, take stock, consult with the population at large, and adjust. But that’s unlikely. In short, the corporate world is “pot committed,” to borrow a poker term, to the AI revolution, such as it is.
Anti-AI optimists might see a bubble bursting or a revolution failing. But for those struggling over the future of labor and the democratization of AI, the smarter reading is that this is, at best, a strategic retreat or consolidation of AI integrations. This means preparing, strategizing, and responding accordingly. Meta itself framed its hiring freeze as “basic organizational planning: creating a solid structure for our new superintelligence efforts after bringing people on board and undertaking yearly budgeting and planning exercises.” As much as one might hate to admit it, the smart money right now — that is to say, the power — is on Meta and its competitors shaping the industrial space as they wish. This is especially true amid democratic backsliding in the United States and its deepening entanglement with the tech oligarchy.
The Wrong Question
To ask whether certain AI technologies will ever “work” is to ask the wrong question, or at least one of secondary, even tertiary, importance. There’s a global race to develop and deploy AI, a contest that at once encompasses a kind of geopolitical strategic jockeying that puts states at odds with one another, and a broader capitalist urge to transform technology and work in the interests of their class across state borders. The capital and political heft behind the AI movement is accordingly robust, with its devotees viewing efforts as strategic necessities rather than technological frivolities.
Setting aside that there’s a difference between types of AI, their deployments, and their integrations — some of which are already working — the core questions for labor, mass politics, and consumers are: To what ends will AI be deployed, to whose benefit, and under whose control? As things stand, there are plenty of reasons to believe the answer to each question is oligarchs and oligopolists under, perhaps, some nominal control of a technocratic ruling class that will act as, at best, a regulatory rubber stamp — and institutional investor or underwriter. And with the state itself backing the broader AI rollout, that is, as they say, the ball game.
The risk of buying into the idea that the AI bubble is bursting is that it risks lulling us into a false sense of comfort, into complacency. For those who want a technological – and political and economic – future marked by deep democratization and collective control, the starting point must be that AI, in one form or another, is here to stay. The task now is not to wait for collapse, but to organize for control.