The AI Boom Looks a Lot Like a Bubble

-

 The current wave of excitement around generative AI is hard to miss, but a closer look suggests a widening gap between promise and reality. The dominant story says AI is on the verge of awakening, that it will suddenly become reliable, intelligent, and transformative if we just feed it more data and more money. That story does not hold up.

The more convincing analysis is that today’s AI systems consistently underperform relative to the hype. They are often best understood as sophisticated add-ons to existing tools rather than world-changing breakthroughs. Video editing, text generation, and image creation have improved, but not in ways that justify the breathless claims being made on their behalf.

Capabilities, Errors, and Incentives

One of the defining traits of current AI systems is their tendency to hallucinate. They produce confident, incorrect outputs in ways that are difficult to predict or prevent. From a user’s perspective, this is a serious flaw. From a company’s perspective, it is sometimes a feature. Errors can reduce development costs, shift responsibility onto users, and still generate revenue.

This mismatch between user needs and corporate incentives mirrors the dynamics of enshittification. When companies are insulated from consequences, product quality stops being the priority.

The Economics Do Not Add Up

The financial side of the AI industry looks even shakier. Estimates suggest that between 600 billion and 1 trillion dollars have been poured into AI infrastructure and development. Annual global revenue, even with generous accounting, appears to be around 60 billion dollars.

Worse, AI does not benefit from the usual advantages of scale. Each new user and each increase in system sophistication tends to raise costs rather than lower them. This is the opposite of the classic tech growth model, where scale brings efficiency. The result is an industry where growth makes the underlying economics worse, not better.

Spending continues anyway, largely as a signal to investors. Companies are not optimizing for sustainability or usefulness. They are trying to prove they belong in the race, even if that means burning money.

Valuations and Fragility

Large tech firms rely on continued growth to justify high valuations and generous stock-based compensation. As long as growth appears endless, price-to-earnings ratios remain inflated. Once growth slows, especially in mature or monopolized markets, valuations can collapse quickly.

This creates a desperate environment where hype becomes essential. The AI boom increasingly resembles earlier cycles around crypto, blockchain, and the metaverse. Big promises, vague timelines, and a media ecosystem willing to repeat them.

A Political Shift Worth Noticing

Despite the bleak economics, there is a reason for cautious hope. Across the US, Canada, Europe, and parts of Asia, there has been a noticeable increase in antitrust and anti-monopoly activity. This suggests a shift in political winds, even if powerful interests remain deeply entrenched.

This matters because AI’s problems are not just technical. They are structural. Concentrated corporate power limits accountability, blocks democratization of computation, and accelerates enshittification. Any serious improvement depends on restoring discipline through competition, regulation, and user control.

Choosing Hope Over Hype

The most persuasive conclusion here is neither optimism nor despair, but hope. Not the passive hope that technology will save us on its own, but the active hope that comes from political engagement and collective action.

AI is not going to wake up and fix itself. The bubble may well burst, and when it does, the damage could be significant. But the growing resistance to monopoly power suggests an opening. If that opening is used to democratize technology and rein in corporate excess, something better could emerge from the wreckage.

That outcome is not guaranteed. It is possible. And that, right now, is reason enough to stay engaged.

Popular posts from this blog

The Dual Faces of Technology: Enhancing and Replacing Jobs

-
Image
Automation has been on my mind a lot lately, more immediately due to a book I read , but also because of the looming reality of artificial intelligence. To understand what's involved, making the distinction between replacing technologies and enabling technologies is crucial. As we traverse the timeline of the 20th century, we observe how each type of technology has distinctly influenced employment. Automated technologies have transformed industries by replacing manual labor with machines. Consider the automated switchboards introduced in the early 20th century. Previously, telephone operators manually connected calls using physical connectors on a switchboard. With the advent of automated switchboards in the 1920s and 1930s, many of these jobs were eliminated as calls were routed through electromechanical systems. Similarly, in agriculture, the introduction of mechanical harvesters like the combine harvester in the 1930s drastically reduced the need for manual labor. These machine...
Read more

The Technology Trap: Capital, Labor and Power in the Age of Automation (Book Review)

-
Image
In his compelling book, " The Technology Trap: Capital, Labor, and Power in the Age of Automation ," Carl Benedikt Frey offers a sweeping examination of the historical and ongoing impact of mechanization and automation on society. Starting with the onset of the first Industrial Revolution, Frey deftly navigates through the evolution of technology up to the modern age of artificial intelligence, presenting a nuanced view of how technological innovations have reshaped labor markets and power dynamics. Frey categorizes technological innovations into two types: enabling technologies, like the typewriter, which enhance worker productivity without displacing the labor force, and replacing technologies, such as robotic machinery in factories, which eliminate certain job roles altogether. This distinction forms the backbone of Frey’s analysis, as he explores the varying implications of each type of technology on different segments of the workforce. The book traces a historical patter...
Read more

Deputy Product Owners

-
Image
"The Product Owner is one person, not a committee. The Product Owner may represent the desires of a committee in the Product Backlog, but those wanting to change a Product Backlog item’s priority must address the Product Owner." - The Scrum Guide There are a lot of ways in which my approach to Scrum differs from what's in the official guide, and since that document refers to Scrum as 'immutable,' let's just call what I do 'Notscrum™.' In Scrum, there is only one product owner per team. That person is the sole person responsible for the product backlog, and the person who advocates for the work to be done. In Notscrum there is one product owner, but there can be multiple deputy product owners. Here's how it works. When my program at Viacom was first forming I met with the then-director of the Library & Archive group and learned that while he set the priorities on work done, the archivists were organized in three distinct areas with a team lead ...
Read more