That's a totally fair point. I feel like there's an empirical question there that I haven't studied and so can't responsibly address. That said, I suspect it's relevant that 1) DEI training is intended to reduce liability rather than to reduce discrimination, which means that in a sense any reduction in discrimination would be a side effect; and 2) my impression is that DEI modules aren't generally designed by social scientists based on research into what reduces future discrimination - they're designed by humanities grads who've studied social issues in a not-rigorously-empirical way. (And that's not to knock them - I'm also a humanities grad who's studies social issues in a not-rigorously-empirical way, that can still be valuable - I just don't think it's fair to expect to produce real-world results that can handle that type of scrutiny.)

More generally, though, I'm skeptical of the ability to reduce unwanted behavior in other people by subjecting them to what amounts to mandatory moral instruction. I believe that people can change their attitudes and their actions if they want to, but I'd be surprised if telling them to do so would have that effect unless they *already* had the motivation. But I do think that clearly defining what's unacceptable and clearly defining the consequences does work, because then it doesn't matter whether someone's motivation is there. And by removing some of the consequences, liability shields strike me as directly counterproductive towards that end. Like, I want workplaces to disallow discrimination. I don't want that to be contingent on somehow successfully converting every individual to any particular belief system, you know?