A recent observation: the people truly using AI well are researchers. The software engineers who've gotten good at AI are, in a sense, people who are good at research.
I know some top-tier finance professors who are already using Claude Code for their research. Not just within their own discipline — they've used AI collaboration to produce cross-disciplinary ideas spanning literature, philosophy, physics, and math. Limited by their engineering background, there's only so much they can do on the implementation side. But as AI adopters, their speed and depth surprised me.
Meanwhile, I've heard plenty of engineers say: "I tried it a month or two ago, didn't think it was great, and haven't gone back."
This got me thinking: what's the most important trait for becoming a first-wave AI adopter?
Probably not technical skill. Some patterns I've observed:
- Open-mindedness — Willing to keep trying, rather than writing it off after one bad experience
- Sensitivity to change — Not being told "AI matters," but feeling it yourself — sensing that something is shifting
- Structured thinking — Able to decompose fuzzy requirements into steps you can collaborate on with AI
- Trust — Willing to hand over some control and see what happens
- Low ego — Not dismissing the tool because "AI can't write as well as I can"
The reason those finance professors picked it up fast may be precisely because they don't carry the baggage of "I already know how to code." Their metric for evaluating AI isn't "can it write code as good as mine" but "can it help me do things I couldn't do alone."
And their profession is inherently about facing the unknown, forming hypotheses, and validating quickly. Engineering training leans toward deterministic thinking — code should be correct, tests should pass. AI output is inherently uncertain, and that might conflict with some engineers' instincts. Researchers are more comfortable with uncertainty — when an experiment doesn't pan out, they're inclined to adjust the hypothesis rather than dismiss the tool.
This relates to the 10x piece I wrote earlier, but from a different angle. That one argued "using AI well requires judgment." This one is saying: before judgment, there might be another layer — willingness to enter the game. Some people's judgment is perfectly fine; they just haven't started yet.