AI can't tell you what good looks like

Hilary Gridley was Head of Product at WHOOP and now writes writerbuilder. Pointing AI at your whole Google Drive won’t teach it your taste, she says, because the Drive is full of junk. A small, curated set of good work will. In recruiting, that means writing down what a great hire looked like before you ask a tool to find more.

Karan Dhir has fifteen years of career behind him and writes about the career bets that compound. His top bet is going deep on one narrow thing. He says the premium on depth has gone up in the AI era, for the reason above. In recruiting, the deep part is knowing which person fits this team, this year. A tool can sort a pile of CVs. Someone still has to know what to look for.

Eric Porres writes Prompted by Eric. Out of the box, he says, an AI agent is “a capable amnesiac.” It doesn’t know your name or your voice, so he writes it an onboarding manual. Molly Graham asks the question that manual has to answer, in Lessons: “What does good actually look like?” If a manager can’t spell it out, the AI can’t pick it up either.

At work I went back through years of recruiting searches. The searches that ended in a hire and the ones that didn’t took nearly the same amount of activity. Effort alone wasn’t the signal. I’ve also trained the whole firm on AI, with regular tips and office hours. The tools are the easy part to share. Knowing what good looks like is the part worth sharpening.

If you can, that’s the brief for any AI tool you try, and for anyone new who joins your team. If you can’t, that’s the place to start, before any tool gets involved. Write the five lines, then show them to a colleague and see if they agree. Reply with yours… I’d love to see them.