LLMs Reward Expertise

Sean Goedecke argues that LLMs democratize access to skills while quietly amplifying the people who already have them. His anchor example is Terence Tao working through mathematics with ChatGPT: deep domain knowledge is what lets a user recognize good output, push back intelligently, and steer the model toward better solutions. He extends the point to programming, where familiarity with the specific system matters more than generic principles, and lands on a thesis worth keeping: the most important skill in prompting is expertise in the domain you are prompting for. A useful counterweight to the idea that these tools flatten the difference between novices and experts.