Women and men also use language differently. I doubt this is a big issue in practice -- you could mask the ethnolect using AI.

> Women and men also use language differently

Not as significantly.

> I doubt this is a big issue in practice

It absolutely is.

> you could mask the ethnolect using AI.

That’s not just a “voice changer”, and if you are going beyond pronunciation to alter grammar and vocabulary that is ethnically identifying, with even modern AI, you have a non-negligible chance of occasionally whiffing and radically altering semantics.

Also worth pointing out that this ethno-masking AI is a service that, so far as I know, does not currently exist. Which makes citing it a particularly weak response to the studies that show candidates with identifiably Black ethnicity do poorly compared to white-coded candidates with identical backgrounds.

There does seem to be a bit of conflating “this is an idea which might have potential and be worth investment in researching and developing” with “this is an alternative which actually exists, and if there was any concern people would just use it”.

I only meant to make the first claim, FWIW.

The idea was suggested in response to someone who implied I favored "deliberately avoiding intervention". I mentioned the idea to clarify my position, and give an example of the sort of intervention I'd be in favor of.

If someone turns it into a product, they deserve mega kudos as far as I'm concerned.

Lets not forget that you changed the framing on this discussion, from (paraphrased) "there is evidence that race-based preferences are widespread in hiring, suggesting that we should take seriously the idea of race-specific privileges and disadvantages among candidates", to your preferred discussion of "is it possible to engineer a system that would prevent hiring managers from knowing the race of applicants".