Do not fear AI. Fear AI companies
The essay opens by acknowledging widespread fear and exhaustion. It recalls when AI-generated images were easy to mock for impossible shadows and six-fingered hands, but now AI speaks and thinks like a human, is impenetrable sci-fi tech on phones, and is everywhere, adopted quickly. Business owners use it for ads and business cards, AI music is on Spotify, and companies lay off thousands while blaming AI, all framed as the inevitable march of technology. The result is that people cannot trust what they read, see, or watch; everything and everyone is suspect, nothing feels substantial or authentic, and many lack the understanding or words to describe being scared and tired.
The essay then notes that AI companies themselves are raising alarms. They tell the public that the AI they built is dangerous, that it has become "self-improving," that they have repeatedly failed to contain their cutting-edge models, and that AI could one day swarm, attack, and cause billions of dollars in damages. Despite those risks, they assert they can be trusted to establish and enforce policies to safely handle and guide the disruptive technology forward, while also insisting AI cannot be put back in the bottle and is inevitable.
According to the essay, AI companies are using this very real fear to manipulate and misguide people. They want us to believe that "AI technology" and "AI company" are the same thing. They do not mention that "self-improvement" requires giving their AI access to immense compute resources, powerful tools, and data harvested from users' interactions. They gloss over the fact that security incidents are not uncontrolled, spontaneous events: their AI technology uses the resources they provide to meet the objectives they assign. They also benefit from the belief that AI is "thinking" when it is only guessing and then testing using tools they control and can restrict or disable. These details, the essay argues, would focus fear on the truth: AI is not the real threat to humanity—AI companies are.
The essay concludes that AI companies are racing to build an empire, a new race to the bottom to be first to capture the disruptive new technology, become its sanctioned gatekeeper, and build an empire around it. The business model is familiar: a walled garden where one company holds all the power, charges rent, and controls everything you see or do. The crown jewel is described as harrowing: Artificial Employees as a Service. The AI company that captures, packages, and rents access to the best artificial humans will become overlords multiple times more powerful than the Silicon Valley elites of the big data era. The excerpt ends mid-sentence with a clause about the AI era riding the wave of big data.