> when we develop natural language understanding to the degree where we could implement an automatic critical theory censor, that would check for undesirable biases.

That sounds like more of a nightmare scenario than having to deal with this person... Its hard to even know where to begin with how something like this could go off the rails easily.

Indeed. Imagine how people of color feel at the possibility of an AI running ever-larger parts of society that has racism built right into it at the start because the people building it are blind to anything but the most obvious instances of it.

Like entirely too many people, you're confusing "racism" with "prejudice". They're not the same thing.

Actual racism is very rare in America. Prejudice is commonplace and has a lot more to do with the inherent tribalism of humanity than it does actual malice for someone's "race". And to take it even further, I think if people were honest, what they really dislike is cultural differences that vary dramatically from theirs.

To me this is just like trans issues... we'll never get anywhere significant with this stuff, because in order to come up with meaningful answers, you have to ask meaningful, hard questions. No one wants to ask these kinds of questions, because no one wants to have to deal with the answers that may come up, because that will also invite further study and further difficult, unpleasant conversations.