AnamCaraMia

This is literally what people are talking about when they say AI will be used to mainstream widely held bigotry. LLMs are trained on frequency and probability -> straight relationships are more well represented in the dataset -> straight pronouns and terms become the "correct" normal.

“For example, if you’re trying to convince people to boycott a segregated store, your object is to convince them that boycotting the store will have a strategic effect, not that desegregation is morally important. For whatever reason, on a cognitive level human beings have a really hard time with this. Smucker cites an example of a Lefty roleplaying session where people were tasked with selling an action to people who agreed with them on principle but didn’t see the strategic merit of the action. Surprisingly, the sellers couldn’t make the conceptual switch to sell strategic merit: instead, they doubled down on THIS ISSUE IS IMPORTANT — even though it had been stressed to them that the people they were selling to bought into the importance of the issue. People react poorly to “this is important, so do WHATEVER I SAY”; they want to be convinced that what you’re proposing will work.”

Also from above:

Somewhat related, my favorite comic strip of all time: