The One-Click Delusion

Let's talk about the one-click delusion.

Google wants prompts to become Chrome skills. Anthropic wants routines. The sales pitch is obvious: reduce the distance between intention and execution until it disappears into a single button. No friction, no ambiguity, no messy thinking, just instant output on demand.

It sounds efficient. It also quietly strips away the most important part of using AI well. The prompt was never just an input box. It was the place where you actually had to think. To write a real prompt, you have to define the problem, set constraints, and imagine what a good result even looks like.

Once that gets collapsed into a pre-baked button, the cognitive work moves somewhere else, usually upstream to the people designing the product. The user gets convenience, but loses authorship. We stop learning how to steer the machine and start selecting from a menu of approved outcomes.

This is the McDonaldization of cognition. Summarize Meeting. Fix Bug. Write Email. The interface becomes cleaner while the human gets flatter. We trade synthesis for convenience and call it progress because the latency went down.

At the same time, the market is still pretending that ever-larger valuations prove ever-deeper intelligence. But a giant price tag is not evidence of insight. It is often just evidence that enough people are still willing to believe the next model will justify the last narrative.

The real gain is not in one-click execution. It is in the gap before the click. The struggle to phrase the request, fail, refine it, and discover what you actually mean. That friction is not a bug in the workflow. It is where the thinking happens. Remove all of it, and the tool may get faster while the user gets dumber.