Strike at the Roots to Burst the AI Bubble
▼ Summary
– AI is described as the most unprofitable technology ever, with every customer use and generation increasing losses, unlike the web which became more profitable with each use and iteration.
– The social reception of AI contrasts sharply with the web; past business press worried about workers smuggling the web, while current press frets about workers refusing to use AI and promotes surveillance to enforce its use.
– Some workers, called “centaurs,” find AI beneficial because they control how it assists them, improving their work.
– Other workers hate AI because they are forced to produce more at the expense of quality and wellbeing, and are made “accountability sinks” blamed for AI failures.
– The article notes that AI has thoughtful defenders, and the impact of technology depends more on who it serves and who it is used against than on the gadget itself.
The phrase “money-losingest” might sound like hyperbole, but when applied to artificial intelligence, it is a starkly accurate description. AI has become the single most unprofitable technological endeavor in human history, and the numbers keep getting worse, not better.
Consider the trajectory of the web. Its profitability didn’t emerge from a single stroke of genius; it grew from sound unit economics. Each new user reduced the overall loss, and every repeat visit increased total profits. With each generational leap in web technology, the profit margins widened. AI operates in the exact opposite direction. Every new customer costs the company money. Every interaction that customer has with the system deepens the loss. And each new, more powerful generation of AI is engineered to be more expensive to run than the one before it. The pattern is clear: we are pouring more capital into a system that is structurally designed to hemorrhage cash.
The social landscape surrounding AI is equally telling. A look back at business journalism from the late 1990s and early 2000s reveals a constant worry about employees sneaking web access into the office. Today, the tone has flipped. The dominant anxiety is no longer about workers smuggling in technology; it is about workers refusing to use it at all. The same publications that once fretted over internet usage now run articles asking, “How do we force employees to adopt AI?” and are filled with advertisements for surveillance software designed to catch and punish those who resist.
This resistance is not simply a matter of technophobia. There is a thoughtful, measured defense of AI, but it comes with a crucial caveat. The workers who benefit from AI are what we might call “centaurs”,skilled professionals who dictate how the technology assists them, using it as a tool to enhance their own expertise. In science fiction, the impact of a gadget is defined not by its capabilities but by who wields it and for whom. For these centaurs, AI is a genuine asset.
The problem lies with the other group. These workers are not fools; they are reliable narrators of their own experience. They describe being forced to use AI to produce more at the expense of quality, to work faster at the cost of their own wellbeing, and to shoulder the blame when the system fails. They are being recruited to serve as what Dan Davies calls “accountability sinks”,human shields designed to absorb the consequences of algorithmic errors. For them, the AI revolution is not a liberation; it is a trap.
(Source: Ars Technica)
