Does adding "engineering" before any random word in a domain create a new job, or make that particular part of a system, or the person who does that, more important? We see the trend as it started:

Prompt engineering. Context engineering. Loop engineering. Graph engineering. Harness engineering.

And if you look at how long the hype around each one lasts, it is getting shorter and shorter. What comes next?

Outcome engineering? Evaluation engineering? Reliability engineering? Environment engineering?

And is it the right way to think about solving the problems we are going to solve? Maybe we should start watching the whole thing from a different point of view.

The promise was: models are not good enough. With a better model, I can solve ALL problems. Models got to the point where they are absolutely usable. The problems did not get solved the way we expected.

Then the prompts: if we take control, we can solve ALL problems. It did not. Then context. Loop. Graph. Harness...

Maybe we cannot solve ALL problems.

Maybe the expectation itself puts us on the wrong way of solving the problems. If we can define what success means, and add a probability factor to the equation, maybe we can start fooling ourselves a little less, and use AI as a very powerful TOOL.

And treat it, and its relationship to the systems it can be integrated into, as parts of a functioning entity. Maybe this way we can add the word "engineering" to it.

At the end of the day, the goal is not to have a functional AI. The goal was, and still is, to solve a real problem. With or without AI.