The AI Hater's Manifesto
A tool that helps debug a crash log can be useful. A trillion-dollar industry built around the promise that it will eventually think, work, and decide for us deserves a far harder audit. The question isn’t whether language models can do anything. It’s whether their demonstrated value matches the certainty, spending, and social pressure surrounding them.
Useful tools and the boundary of judgment
Outsourcing work versus thought
The financial story under the hype
Inference costs and productivity proof
Growth, power, and human investment
Future tense and accountability
The maintenance burden…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.