I can get ChatGPT to mock men but not women.
How is that appropriate?
How is that appropriate?
This is a classic pattern. For example, people have complained numerous times about Google's search suggestions and results for certain topics, e.g. "my boyfriend hits me" yielding sponsored information about domestic violence but "my girlfriend hits me" not yielding anything.
Go contact OpenAI about how you’re concerned that their bot can produce misandry and try to convince them it’s an important issue that should higher in their backlog.
I'm not saying it is; indeed, I recognize that fairness is often about equity and not equality.
However, it's naive to assume that certain things don't exist because their authors simply didn't have the time or resources necessary. In many cases what's available is a direct result of their biases.
To further my previous example, I have met people who are vehemently against resources for men's around mental health or domestic violence because they believe that men are inherently privileged and thus undeserving. As another commenter mentioned somewhere in this thread, what gets added or removed is a direct reflection of the author's biases and endorsement of a specific worldview.
So yes, absolutely these things can be a reflection of people's biases but in no way whatsoever is the example you mention a "socially acceptable prejudice." Recognizing that that the thing we call the patriarchy creates systems that privilege men does not ever imply "and so the solution is to actively hurt men and deprive them of necessary care."
In the greater context of the article, I think it's important that we be able to have these open and honest conversations, especially when they're uncomfortable or not socially acceptable. Unfortunately, "Open" AI is anything but, and I think we're right to be skeptical of unnaccountable people being the arbiters of what is or isn't acceptable. I am especially weary of granting sweeping powers just because they're being used for something I agree with.
A big part of that is also what people's expectations are for the technology. Going back to Google as an example, if my expectation is that Google is a search engine then I think it's justifiably irksome when queries are obviously hard-coded not to deliver the best result. With ChatGPT, people currently see it as a powerful and maliable AI that has a wealth of information at its disposal, and certain information is obviously being restricted. While I won't pretend to know the best solution for how it should be, I think it'd be a shame for it to become a lobotomized propaganda machine or "souless" inoffensive advertiser-friendly husk.