As for my ask of HN, some might say, “Estas Loco”, if in a coding interview a candidate could use GitHub copilot you wouldn’t be able to tell if they themselves could solve the problem on their own.
While there is likely. value in them being able to solve that problem on their own, it is much in the same way that it is valuable to know how the compiler works and what assembly language does and why.
That’s not what you’re interviewing them for though.
You’re interviewing them to see if they can solve your problem, your bosses’ problem, or if you’re your own boss, your customer’s problem.
So if your problems are accurately represented by this interview question that is easily solved with GitHub Copilot why are you hiring a person if you only discover what a machine can do for much cheaper than a human?
If I give a problem to two engineers with the option to use GitHub Copilot if they want like I’d say they can use google to look up libraries, I’m going to hire the one who writes the most beautiful, performant, comprehensible, maintainable, and just downright good code. If he who used Copilot had the better code, I’d pick him. Why should he be faulted for using the tools he has to make the best code? An inexperienced hobby woodworker, working by hand, will make worse projects than the inexperienced hobby woodworker using the right power tools.
The experienced woodworker, will know when to use his tools and when to do things by hand.
“_The competitive advantages the marketplace demands is someone more human, connected, and mature. Someone with passion and energy, capable of seeing things as they are and negotiating multiple priorities as she makes useful decisions without angst. Flexible in the face of change, resilient in the face of confusion. All of these attributes are choices, not talents, and all of them are available to you._”
That might be the goal of the overall interview process but it isn't the goal of whiteboard data structures and algorithms questions.
Tying your interview process to a difficult to reason about tool that's new, not widely used and has a probabilistic output seems like a really bad idea.