I'm 100% on board with the "would I like to work with this person" criterion, but I wonder how well it fares now, in the age that, at least on the surface of it, wages a crusade against all and every biases. If I am biased against a particular group, I probably wouldn't like to work with a person from that group, would I? How does the "would I like" criterion hold against this?
I completely agree it can be misused - the best way to counteract biases in interviewing is have a standardized process.
So for example, we don't actually phrase it as "Would I like to work with this person?" we just ask for feedback on "General Behavior and Attitude"
If they pass the other components of the interview but fail that one, we usually send them on to a second interview and directly look to confirm or refute the poor behavior / attitude assessment.
We also ask recruiters specifically to reach out between interviews for "small talk" to discern behavioral findings.
In terms of training interviewers, I couch it in terms of decorum and "putting your hands at 10 and 2 on the wheel" and that it's a box-ticking exercise, not a psychological study - that is, 90% or more of the subpopulation of qualified candidates with sustained work history know how to "behave professionally", and this isn't a differentiator.
