On Being a Little Too Obsessed With AI

Over the last few years, I have gone from being vaguely curious about artificial intelligence to structuring big chunks of my life around it.

I measure my days in experiments: a new prompt here, a quick prototype there, a “what if I asked the model to do this?” in between meetings. I catch myself reaching for an AI tool for things that used to feel obviously human: outlining ideas, naming projects, even arguing with myself about tradeoffs.

Part of this is sheer excitement. Modern models are the first general-purpose tool that feels like a collaborator rather than a program. When it works well, I get the feeling of pairing with someone who has read everything and judges nothing. That is dangerously addictive.

But obsession always has a cost.

If I am not careful, I can outsource too much thinking. It is easy to accept the first reasonable answer instead of wrestling with the problem long enough to actually learn something. It is tempting to turn every task into a prompt engineering problem, even when the fastest path would be to close the laptop and go for a walk.

So I am trying to draw better boundaries:

• Use AI to explore option space quickly, but make the final calls myself.
• Use it to explain things back to me, then re-explain them in my own words.
• Use it to remove friction from the boring parts of work, not from the meaningful parts.

I am still absolutely obsessed with AI. I do not plan to change that. But I do want the obsession to serve my curiosity, my craft, and the people I work with — not the other way around.

This post is a small snapshot of that tension. If you are reading this, there is a good chance you feel some version of it too.