You’re Not Hiring a Calculator…
Imagine you're hiring an accountant, and you worry about making sure that they can't use a calculator in the interview, because all you're going to ask them is to add numbers in their head, so using a calculator would be cheating. Pointless, right?
It seems candidates and companies are locked in an arms race of AI sophistication, especially in tech: As it turns out, AI coding assistants are really good at the types of puzzles hiring managers like to use in tech assessments: "You're given two lists. Both are sorted. What's the fastest way to find their collective median?" and the likes.
So now we have candidates developing tools that sit in your browser and feed interview questions straight into ChatGPT and hiring managers wondering if it's time to bring back the in-person (as in, live, not via Zoom) interview.
That's misguided. We can't celebrate the productivity gains that AI enables in those who know how to use it, and then freak out when candidates use AI in the hiring process. Why not design the whole process so that only those who can produce great results with AI pass?
Take writing. Yes, AI can generate lucid and passable responses. But let's say you're actually hiring for a position where someone has to produce writing. Marketing copy for your website, for example. Why bother hiring someone? Couldn't you just ask ChatGPT to "Write my marketing copy, please?" Why? No, really. Why aren't you saving all that money and instead spend just five minutes each day prompting the AI for the writing you need? Or hire a part-time teenager to copy-paste your prompts into ChatGPT and copy-paste the answer to your website. Maybe because there's skill, taste, and discernment required beyond that? (I know for sure that I couldn't just prompt my way to a Pulitzer prize.)
So then, in hiring, make it an "open-book" exam. AI explicitly allowed. Just raise the bar for the outcome you want to see. It instantly defeats the point of using AI to "cheat." So ask for spicy takes, strong opinions, a human explanation for why this is good and that is bad. (Try asking ChatGPT something like "which is the worst frontend framework" and you get some "on the hand, on the other hand, to be fair, in the end..." wishy-washy position. A real human will be happy to go off on a fun rant.)
