Aren't these kind of mutually exclusive, at least directionally? If the interview is meaningful you'd expect it to predict job performance. If it can't predict job performance then it is kind of useless.
I guess you could play some word games here to occupy a middle ground ("the coding interview is kind of useful, it measures something, just not job performance exactly") but I can't think of a formulation where this doesn't sound pretty silly.
Nothing you do in an interview like this resembles day to day work in this field.
A wrong answer with good thinking is better than a correct answer with no explanation.
Oftentimes the explanation is correct, even if there's some mistake in the code (probably because the explanation is easier to generate than the correct code, an artifact of being a high tech parrot)
Humans think differently from LLMs so it makes sense to interpret the same signal in different ways.
Words are hard!
Have a great day.
I just see, in what you're doing, a wild lack of self awareness. You're criticizing me for doing to someone else a milder version of what you're trying to do to me now; I'm genuinely confused how you can't see that, or how you could possibly stand the hypocrisy if you do understand that.
Let's define the interview as useful if the passing candidate can do the job.
Sounds reasonable.
ChatGPT can pass the interview and can't do the job.
The interview is not able to predict the poor working performance of ChatGPT and it's therefore useless.
Some of the companies I worked for hired ex fang people as if it was a mark of quality, but that hasn't always worked out well. There is plenty of people getting out of fangs having just done mediocre work for a big paycheck.
The technical term for this is "construct validity", that the test results are related to something you want to learn about.
> The interview is not able to predict the poor working performance of ChatGPT and it's therefore useless.
This doesn't follow; the interview doesn't need to be able to exclude ChatGPT because ChatGPT doesn't interview for jobs. It's perfectly possible that the same test shows high validity on humans and low validity on ChatGPT.
What part of that take do you not understand? It's a really easy concept to grasp, and even if you don't agree with it, I would expect at least that a research scientist (according to your bio) would be able to grok the concepts almost immediately...