It ends up favoring research that may or may not be better than the honestly reviewed alternative, thereby lowering the quality of published papers in journal where reviewers tend to rely on AI.