Effective Methods for Evaluating AI Skills in Hiring Beyond Self-Reported Confidence

curious if anyone's cracked evaluating AI skills during the hiring process in a way that actually works. we've been looking at this at aisa.to and the gap between self-reported AI confidence and actual measured skill is enormous — 62% of people score below proficient when you assess them through conversation instead of a quiz. has anyone found a screening approach that reliably separates "uses ChatGPT for emails" from "has AI integrated into their workflow"?