In defense of GP, you know why we haven't replaced all hiring interviews with standardized tests and statistical models? Gut instinct. People can make a good guess about how they'll get along with an interviewee. Machines can't just yet. And this is important, because while a mediocre person that fits in well with a team is only a minor positive contribution, a bad team fit is a negative contribution even if they're exceptional in hard skills.
> modern evidence-based research and training.
Interviewing process is hard, social-sciences-style hard. Completely bullshit theories and "standard practices". This doesn't rise anywhere near "evidence-based".
Some of my interview is giving them an underspecified problem and seeing whether they can formulate questions (and dare to ask them!) that lead to solving the important part. I don't know how to standardize that, but we definitely need it.
I'm not arguing against gut instinct being a necessary factor. My point is that when it's the only factor, and when people rationalize it as being the only factor, without any quantification, then that's the problem.
How do you define "bad team fit"? Can you make a checklist of things that are associated with "good team fit" and "bad team fit", and aim to assess for those items individually, before coming to a conclusion? You don't need artificial general intelligence to make a spreadsheet and a checklist. If you don't do at least this, how do you even know in your own mind what a good team fit really is? You don't need to have perfect accuracy with this list of criteria, but not having a list at all will be less accurate than any list you come up with in good faith.
> modern evidence-based research and training.
Interviewing process is hard, social-sciences-style hard. Completely bullshit theories and "standard practices". This doesn't rise anywhere near "evidence-based".