So I am going to learn the wrong lesson here and say that it probably makes sense to delay automated rejections so as not to arouse the suspicions of job applicants, as that seems to be the mistake which cost this employer $300,000+ in damages. Of course the better lesson is to understand the implications of using AI & Automation, which fundamentally wants to
restore a human recruiter to be the one doing the rejecting. How this is going to help with de-biasing might now be beside the point.