The tendency of people to over-rely on automated recommendations, even when they are wrong.
In more detail
It undermines human oversight of AI decisions. Training, interface design and requiring reviewers to justify agreement are proposed countermeasures.