A sidenote on one thing brought up in the article, it's worth noting that the implicit association test (IAT) is of highly dubious utility, despite common use.

A sidenote on one thing brought up in the article, it's worth noting that the implicit association test (IAT) is of highly dubious utility, despite common use. While it measures something (as does any test), there are very serious questions about what it actually measures, how reliably it measures it, whether what it measures actually correlates with real world bias, etc. There are very good reasons to think that, however clever the test may sound, and however real implicit bias is (its existence is almost indisputable), the test doesn't accurately measure it.