The issue I take with the concerns raised in the paper regarding:<p>"A language model that has been trained on such data will pick up these kinds of problematic associations."<p>"Such data" is basically everyday, normal discourse, and some of the "problematic associations" are training that includes phrases like "woman doctor", "both genders", etc. While I get the point, this itself is a biased interpretation of discourse and would be worrisome imho to have people filtering models with their own biases vs the language as it's used by the vast majority of people.
Obviously, bear in mind this is a(n early) draft. That means that _it’s not finished_. You really wouldn’t want to read the first cut of Lord of the Rings (yeah, I know some of you would).