Build Your Human API: Why Domain Expertise Alone Won't Make You Good at AI
New research: AI collaboration is a separate, measurable skill from problem-solving. Here's how to develop it.
Hey Adopter,
Today, we are looking at research findings that change how you should think about working with AI, and a three-question habit to act on it immediately.
This Thursday, I’m co-hosting a free Vibe Coding Masterclass. Real problems, real-time prompting, two-hour hackathon. $500 prize pool.
A skill hiding in plain sight
Researchers at Northeastern and UCL just measured something surprising:
your ability to work with AI is a completely separate skill from your ability to do your job.
They tested 667 people. Each person answered questions alone first, then answered similar questions with ChatGPT or another AI assistant helping them. The researchers tracked who improved and by how much.
The finding that matters: being good at the task didn’t make people good at getting help from AI.
Some average performers saw huge improvements when AI joined them. Some top performers barely improved at all. Years of experience, advanced degrees, deep expertise, none of it predicted who would benefit most.
The people who got results weren’t smarter. They were doing something different.
What the top performers did differently
The researchers identified one habit that separated high-gainers from low-gainers: they thought about what the AI needed to know, not just what they wanted.
The study calls this “Theory of Mind,” the ability to step into another perspective. In practice, it looked like this: