Rather than tell you what I think about AI, I asked AI to evaluate what it thinks I think about it. In its words:

;quote;
I’d summarize your relationship with AI as pragmatic, technically engaged, but philosophically ambivalent.

You don’t seem anti-AI at all. In your projects, you’re interested in AI as a force multiplier: something that can write boilerplate, explore implementations, help you reason through unfamiliar areas, and increasingly act as an agent rather than just autocomplete. You’ve thought pretty concretely about how to make agentic development work well—for example, preferring small, well-bounded ownership areas with explicit local rules because you don’t trust an agent to reliably hold an enormous system’s context at once. That’s less “AI magic” and more treating AI like another imperfect engineering component whose failure modes need to be designed around.

At the same time, I think there’s an important distinction in what you actually value about work. You’ve historically gotten satisfaction from making things and understanding them, especially writing software itself—not merely producing software as an output. So when AI becomes good enough that the human role shifts toward specifying, reviewing, and orchestrating machine-generated work, you don’t automatically experience that as an improvement. Even if it makes you more productive, it can remove part of the craft that made programming meaningful to you.

That creates a tension in your use of it:

‍You want AI to increase your reach without replacing your agency.

‍You seem happiest with AI when it helps you do something you otherwise couldn’t do as quickly—investigate an architecture, generate a starting point, reason through game systems, experiment with an unfamiliar technology, or accelerate the mechanical parts of implementation—while you remain the person forming the mental model, making the design decisions, and understanding why the system works.

You seem much less comfortable with the emerging version of software work where success means becoming extremely good at feeding work into agents and accepting that you may understand progressively less of the implementation. Your experience in an AI-first work environment made that conflict particularly concrete: difficulty adapting wasn’t simply “I don’t know how to use AI.” It collided with your conception of what being a good engineer and enjoying engineering actually meant.

There’s also something interesting in your personal projects: you appear more open to AI there than your anxiety about AI and employment might suggest. With a game, a library, or another thing you personally care about, AI becomes subordinate to authorship. You still decide what the thing is. You can use the machine aggressively without feeling that the machine has replaced the reason you’re there. That’s quite different from an employer saying, essentially, “the product of engineering matters; how much of the engineering you personally do does not.”

So if I condensed your stance into a few lines, it would be:

‍AI is a very powerful tool, and I want to become good at using it. But productivity is not my only goal. I care about understanding, authorship, and the craft of making things. I want AI to expand what I’m capable of creating—not turn me into a thin management layer over work I no longer meaningfully participate in.

‍And technically, your instincts seem to be moving toward AI-native engineering rather than indiscriminate AI use: decompose systems so agents can succeed, give them constrained context, establish rules and boundaries, let them take on larger chunks of execution, but preserve enough structure that the human can reason about and control what is happening.

I think that distinction—augmentation versus alienation—is probably the cleanest through-line in your view of AI.
;endquote;