Maybe I worded it badly. I was attempting to make a distinction between moderation rules that are designed to be fair based on some world view VS rules that are automatically generated based on some sample set to appear convincingly fair to us.
In essence the AI is just giving us what the majority of it's test data tells it that we want to hear.
I think the whole point of this is that the first part of your "vs" statement and your second part can have huge overlap if you happen to agree with the worldview that is informing the responses of the AI. It looks fair to you if it spits out the response that fits your view of the world.
But the whole point is, there's no such thing as "objective", especially when you start talking about people. There's no place to stand as a human being that doesn't involve experiencing the world through one's own consciousness, which has an implicit value structure. Everyone has their own little "religion", and when they share that worldview with other people, the religion then becomes that which holds the group together. "Ligament" (that which holds bones together) and "religion" (that which holds groups together) have the same root: "lig" which means "to bind".
AI is revealing this to a large segment of our population that considers themselves "objective" and "secular", and the cognitive dissonance that is coming out as a result is funny and disturbing in equal parts.
> there's no such thing as "objective", especially when you start talking about people
I agree with you.
However I would add to your statement about AI being revealing. One thing that AI I feel is revealing to me at least, is that it's now making me think much more about the distinction "a human search for truth" however flawed that may be, and things that are made simply to appeal to my biases.
Linking this back to the topic at hand, maybe there is no such thing as objective, but there is a difference between a rule written by a human that was at least trying to search for truth VS a rule written by an system that just parrots the most pleasing thing back at us.
And my point is, for those working on this system, it appears to be very difficult for them to see the difference between "trying to search for truth" and "a rule written by a system that just parrots t he most pleasing thing back to us" because it is a venn diagram that completely overlaps, since they think that they are "trying to search for truth" and haven't considered that they might be wrong, because everyone they work with sees the world the same way they do. They are never challenged, until the platforms get exposed to the public.
This is essentially the same situation as people who have never left their state, or their home country. Many things they take for granted and that they consider to be "basic human nature" actually end up being very cultural and very uncommon once one starts traveling outside one's home country. But you are completely unaware of these situations because they never get challenged when you live in a culture that instantiated itself in you as you grew up.
You dont have to go that far, just start from two words: reality and truth. Reality is what is, and truth is how we describe it.
You can see how truth starts to shake right ? It is an intermediary we use to interface with reality but it is imperfect: it uses eyes and ears, languages and words, expression and cultural habits, ink and paper.
For instance, I think it is true that women cannot do all than men can and men cannot do all that women can. It sounds terrifyingly horrible but because it's too hard for me to express reality. I could continue and say the differences are irrelevant to the weight of their voices that should be equal, but then... why ? I used should here, and leave reality to start ideology. I could say that opportunities should be offered in equal proportion, but... how ? Are women/men in that context two discrete categories, or a gradient where people can choose what they want to be? And do they choose wisely at all time ? What is wise ? Is it constant in all time and place ? Should we distinguish men and women ? Should we ignore what makes them different ?
And now, I trapped the AI: it can't care, it doesn't have a daughter, a mom, a wife and a sister like I do so it will not overcompensate patriarchy like I do, maybe, while I really want my daughter to prevail maybe at the cost of someone else's son, so my fair is not their fair...
Searching for truth is important, but once you've reached it, all you have is ink on a paper transcribing in a grammar and syntax what your poor eyes and weak ears perceived of reality, or at least what your ever aging brain remembers of it. It is no different for the AI, limited just as us, but not exactly overlapping the same flaws.
You can define a ton of types of "fair." It's just like "random."
The big "fairs" imo are:
a) majority votes, one person one vote,
b) auctions where everyone starts with the same funds,
c) random picks, where everyone has an equal chance of being chosen.
None of these have anything to do with ideas, they have to do with people. The way to be fair about ideas is to express them without distortion or omission, in a way that would be recognizable to the people who support them. The only way to do this is to leave out all of the extraneous commentary and stick to clear, falsifiable statements.
edit: what generative language models do is parrot a ton of acrimonious political discussions. Ironically, they can't be fair because they have no values.
Instead of deciding the content moderation on a specific item in a category (i.e. black people or women), use the category itself. So moderate the language if it is biased against any race, not just black people; or any gender, not just women.
