The current landscape of AI development is riddled with instability and impending collapse, yet from this chaos, we can extract valuable lessons and potential pathways forward. But here’s where it often gets controversial: many believe the AI boom will inevitably lead to lasting success, while in reality, it's fundamentally a bubble—one destined to burst, leaving behind a trail of wasted resources and shattered expectations.
As a science-fiction author, my craft revolves around constructing futuristic narratives that serve as allegories for our present socio-technological realities. My aim isn't to predict the future—nobody truly can. If we could, the future would be a fixed, unchangeable script, and that would render us powerless to steer it. The unpredictable nature of the future is actually a good thing because it preserves our capacity to shape what lies ahead.
That said, there's a widespread misconception that science-fiction writers are oracles—people who can foresee what’s to come. Some colleagues of mine even believe they possess the ability to “see into the future.” Meanwhile, a fan base that delves into speculative stories often takes a step further, convincing themselves that they’re reading the future itself. Sadly, an increasing number of these enthusiasts have aligned themselves with the AI industry—often called “AI bros”—who continuously hype the idea that their advanced autocomplete tools will someday become sentient and dominate us, turning us into literal paperclips in a scenario known as the “Paperclip Maximizer” (see https://en.wikipedia.org/wiki/Instrumentalconvergence#Paperclipmaximizer). This myth has led many journalists and conference organizers to repeatedly ask me to comment on the future of AI. But here's where it gets controversial: I’ve historically resisted such predictions, because like everyone else, I know that no one can truly forecast the future.
I spent two years patiently explaining to crypto enthusiasts why I believed their industry was flawed—and faced relentless accusations. First, that I simply didn’t understand crypto; then, once I demonstrated comprehension, that I was secretly paid by rival interests. Engaging in disputes with such groups is often like arguing with Scientologists: exhausting and unproductive. Nonetheless, the questions persisted, so I’ll take this opportunity to share my perspective on AI and how to critically evaluate its development. Specifically, I mean: how to criticize AI in ways that genuinely prevent harm, rather than merely cheerlead for its unchecked growth.
The Reverse Centaur Phenomenon
In the realm of automation, a “centaur” is a human operating alongside a machine—think of a driver assisted by autopilot or a writer using AI to generate ideas. Conversely, a “reverse centaur” is a human being reduced to a fragile, fleshy component servicing a machine—an individual whose role is primarily to support an unfeeling algorithm.
Consider an Amazon delivery driver trapped in a van surrounded by AI-based surveillance systems. These systems monitor the driver’s eye movements, penalize them for looking in unauthorized directions, police their speech—disallowing singing—and report any failure to meet quotas. The driver’s entire effort is dictated by the van—incapable of self-driving— which requires human labor to bring parcels from the curb to your doorstep. The human is merely a piston in the machine, which demands superhuman stamina and speed.
Being a “centaur” is generally favorable—humans working with AI to augment their abilities—whereas the “reverse centaur” is a dehumanizing situation where people are essentially appendages to machines they cannot control. Many AI tools are seemingly designed to transform workers into reverse centaurs, which none of us want.
My job as a science-fiction author is to think beyond just what a gadget does. I focus on who it benefits and who it harms, challenging the narrative pushed by tech elites who try to convince us that there's only one way to use technological innovations. Mark Zuckerberg wants us to believe that we can’t communicate privately without his surveillance; Tim Cook insists on controlling the software installed on our devices, extracting a cut from every transaction; Sundar Pichai implies that access to the web depends on them watching you every step. This is a kind of vulgar determinism akin to Margaret Thatcher’s infamous slogan: “There is no alternative.” It’s a cynical demand to shut down dissent and alternative visions driven by the false belief that the current model is the only possible one.
As a writer, I challenge these assertions by contemplating at least a dozen alternate paths before breakfast—alternate ways technology could be used, regulated, or reimagined. So let’s unpack what’s truly happening with this AI hype cycle, who benefits, and who’s being served—distinguishing myth from material reality.
How the AI Bubble Grows
A core issue fueling the AI bubble is the dominance of monopolistic tech giants. These companies don’t merely compete—they swallow entire sectors, either individually or in collusive cartels. Google and Meta dominate digital advertising; Google and Apple control the mobile ecosystem—Google pays Apple over $20 billion annually to prevent it from developing a competing search engine, contributing to Google’s 90% search market share.
One might assume such monopolies are beneficial for these corporations, but in fact, they are a crisis for the broader economy. Growth stocks—companies that are expanding rapidly—are highly prized by investors, who pay exorbitant prices relative to earnings, measured by the “PE ratio.” When a company stops growing—becoming “mature”—its PE declines sharply; for example, Target’s PE might be around 10, whereas Amazon’s could be 36. This disparity means that, in growth phases, companies like Amazon can issue shares instead of cash for acquisitions, leveraging their high valuation. A dominant company can continually fund expansion through stock sales, fueling a cycle of hype and inflated valuation.
But once market saturation hits—say a company controls 90% of its niche—it becomes challenging to grow further. Investors panic, causing massive drops in valuation, as happened to Facebook in early 2022, when a slight slowdown prompted a $240 billion sell-off in a single day. These are the perils of monopolistic dominance: shareholder expectations for perpetual growth create a relentless urge to hypedify or inflate value through new bubbles—whether in video content, crypto, NFTs, or now, AI.
The primary goal of this obsession isn't necessarily to produce viable, useful AI but to maintain the illusion that the company will keep expanding, until the inevitable next bubble takes over. That’s why the current AI hype is so pervasive: it’s built on the material foundation of hundreds of billions of dollars already invested.
The Myth That AI Can Replace Your Job—And Why It Can’t
Many industry narratives suggest AI will revolutionize and automate entire labor markets, promising investors that their machine algorithms will replace human workers, leaving employers with fat profits and a smaller wage bill. The fantasy: CEOs firing workers wholesale, sharing savings with AI firms, and still maintaining productivity.
But the truth is far less rosy. AI cannot do your job—at least not yet. It can assist, yes, but not replace humans in complex, judgment-heavy tasks. Take radiology: AI might occasionally detect tumors missed by radiologists. That’s promising—but only if AI is used to aid doctors, not to replace them wholesale.
Imagine a hospital that implements AI tools with the understanding that radiologists will review every diagnosis, with AI providing a quick second opinion. This enhances accuracy without cost-cutting or layoffs. But the market narrative sells a different story. Seasoned AI sales pitches go like this: “Firing nine out of ten radiologists saves millions, with the remaining overseeing AI diagnoses that are mostly right but occasionally wrong—and when they’re wrong, it’s the human’s fault.” This creates a “reverse centaur”: a human being forced to oversee an imperfect machine, bearing the blame for its errors.
This dynamic is at the heart of the AI hype—an “accountability sink,” as Dan Davies words it—where human workers are placed in the position of taking responsibility for AI failures they cannot realistically control or fix.
The core deception: AI doesn’t replace the job—it replaces the high-wage professionals most capable of catching subtle errors. And if it’s profitable to do so, the message is clear: replace those workers first.
What About AI-Generated Art?
AI art often serves as a marketing ploy, designed to generate buzz while exposing how little it actually contributes. Many illustrators are impoverished and insecure; consultants claim AI will obliterate their industry, but the economic impact is negligible. The real purpose behind AI art: to persuade the public that AI is “creative,” thus distracting from its true function—cost-cutting.
Can AI create genuine art? Deeply experienced artists understand that true art emerges from the artist’s profound, ineffable feelings—personal insight infused into a medium like paint, words, or sound. AI, however, only regurgitates patterns based on prompts; it lacks consciousness, emotional depth, or original intent. Its images and texts are, at best, eerie echoes of human expression—implying presence or intention where none exists—a phenomenon Mark Fisher called “eerie.”
Rethinking Copyright and Its Limitations
Many advocates argue that extending copyright protections could help artists—yet this is a misguided strategy. Currently, AI training involves scraping publicly available data, something legal under existing law, and analyzing that data—processes that don’t require licenses. Publishing factual information about works is also legal, since copyright generally does not cover facts.
Proposing to expand copyright protection to activities like data scraping or analysis risks creating a legal monster—favoring corporate interests over creators’ rights. It would merely confer new rights to large media conglomerates, allowing them to control AI training by patenting datasets or algorithms, while leaving individual artists worse off.
The key safeguard exists in the current legal stance of the US Copyright Office, which has maintained that AI-generated works cannot be copyrighted because true authorship requires human creativity. This position has been upheld in court repeatedly. This ensures that AI-created content remains in the public domain—meaning big corporations cannot monopolize it by applying for copyrights. The consequence? Companies must pay living human creators to produce original work, reinforcing a “centaur” model—humans and AI working together.
How Workers and Society Can Fight Back
All hope is not lost. Artists and their allies can organize and push for stronger sectoral bargaining rights—similar to what writers achieved during their historic strikes—allowing entire sectors to negotiate collectively. This collective power is crucial because most workers have been prohibited from such bargaining since the Taft-Hartley Act of the late 1940s.
The battle against the AI bubble isn’t just technological—it’s fundamentally political. It hinges on challenging the myth that AI can replace human labor entirely and exposing the parasitic economic practices that sustain these bubbles. By resisting the hype and insisting on policies that protect human creativity and labor, we stand the best chance of deflating this toxic, destructive bubble.
The Inevitable Burst of the AI Bubble
Bubbles are inherently destructive—transferring ordinary people’s savings into the coffers of the wealthiest few, only to explode and wipe out those investments. Unlike frauds like Worldcom, where the underlying infrastructure persisted—fiber optic networks still exist—most of the current AI investments are built on ephemeral, unprofitable foundation models that are unlikely to survive the bust.
When the AI bubble inevitably bursts, most companies will fold, datacenters will shut down, and only a handful of skilled coders working on open-source, low-cost models will remain. These models will power useful applications—from transcription to image recognition and automation—accessible on personal devices and fostering innovation outside the monopolistic hype.
If the bubble hadn’t inflated, we might instead have seen steady, unglamorous progress—like the development of useful plugins—rather than a costly frenzy for unprofitable giants. But the bubble’s collapse will be messy. Its aftermath will leave behind the real, unglamorous work of applied statistics, affordable hardware, and open-source tools—elements that, in the long run, can empower society far more than the illusion of endless growth.
The key to deflating this bubble is to focus on the myth that AI can fully replace human labor—especially high-wage roles—and to recognize that growth-driven tech companies rely on continual, irrational bubbles to survive. The fight is both economic and ideological—and it’s worth engaging in seriously, because the stakes are nothing less than the future of work, creativity, and social equity.
Are you ready to question the narratives fed to us by tech giants? Do you see through the hype, or are you convinced AI will reshape everything? Join the discussion below.
