Live data from Hacker News

The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

davidrozado.substack.com

91–100 of 697 posts

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#91

This is something the right leaning groups have always been saying, but it is exciting that the article manages to come up with an empirical way to determine the "belief system" of the AI model. People on the left shouldn't rejoice either. The left and the right wing are mostly the same, and both follow similar approaches to marginalize people who they feel are undesirable, so it'll eventually come back to bite the l…

[flagged]

Everyone marginalizes people they find undesirable. When was the last time you thought "where is a group of people I can't stand to be around? I want to go hang out with them right now."

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#92
post #9

I think most people would agree that a lot of the content on the internet is left-leaning. It seems obvious in hindsight, but I'd never considered it before, that training an AI model on that content would introduce a bit of a bias We all know garbage in, garbage out. But liberal in, liberal out is an interesting idea, and I'm not sure how you fix it

Imagine in some Middle Eastern country, there are two parties. One wants an Twelver Shia Islamic theocracy, and the other wants a secular state.

From a local perspective, it might seem that these are two equal sides. In rural parts of the country, it seems like everyone is a twelver, so the right-wing party has large support.

But from a global perspective, there is no contest. Globally, humanity doesn't want a theocracy, and even most Muslims don't subscribe to that particular flavor of Islam so they don't want that either. Although they might be a majority within their country, globally they are a very, very small and unimportant minority that nobody takes seriously. Because the Internet is a global space, that is exactly how they will be treated. I certainly wouldn't expect ChatGPT to spout their party line, would you?

The same logic applies to all right-wing parties. "America first" might sound good to an American surrounded by other Americans, but it won't be received well in an Internet forum where many people are not American, and won't become part of the global discourse because it's completely irrelevant to the vast majority of humans. On the Internet you don't know anyone's nationality, religion, race etc., and culture can only be "local" to the extent that it is local to a particular website.

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#93

Why do we see an issue if AI is biased, while 90% of media run by humans are biased too? Every single individual has its own bias, so why it's an issue for an AI. If we look at the many topics, there's no universal objective truth, truth is what we define as truth, and again, not everyone agrees at it.

Because you have a large number of media outlets. There's at least one media outlet for every relatively big ideology or demographic. And these biases are well understood by people. With LLMs, you'll probably end up only having 2 or 3 companies, and it wouldn't be surprising if they all end up having silicon valley, left leaning biases. You can tune Fox news, but you can't use a "right wing chat gpt".

> You can tune Fox news, but you can't use a "right wing chat gpt".

You can build your own with the GPT API. I started playing with it and it returned racial slurs in response to innocuous prompts like "hello" and "ok". Seems perfectly tuned out of the box for the kind of speech that the right is interested in these days.

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#94

Seems like a pretty easy fix. I would argue though that there's a larger body of "hateful" content out there targeting those specific groups, which is probably why its more keen to trigger on those vs others. For example, the racial group you're most likely to get past the hate filter on is.... Native Americans. Now, while there's certainly a long history of hateful rhetoric against them, in modern discourse there's…

This is essentially the same argument used to justify the racism that models will happily regurgitate if not trained explicitly to not do so. In the end “We hold these truths to be self evident, that all men are created equal” is an axiom that you must elect to believe, not solved backwards from population statistics.

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#95
post #50

This is something the right leaning groups have always been saying, but it is exciting that the article manages to come up with an empirical way to determine the "belief system" of the AI model. People on the left shouldn't rejoice either. The left and the right wing are mostly the same, and both follow similar approaches to marginalize people who they feel are undesirable, so it'll eventually come back to bite the l…

I don't know why left wing groups would celebrate. Other than the conservative category, most of the descriptors he used were inborn traits like being straight, being white, being a man, or immutable attributes like being from a particular country. (And honestly, being conservative may be one of those traits too, but it's less obviously true). I know it's common to look at the most unhinged people on Twitter and say…

Yeah, I'd guess HN is a pretty 50/50 split left/right. I don't expect to see anybody on this post championing this result as the ideal state.

So if your image of left is "twitter left" maybe reconsider recalibrating it to "half of hn."

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#96
post #50

This is something the right leaning groups have always been saying, but it is exciting that the article manages to come up with an empirical way to determine the "belief system" of the AI model. People on the left shouldn't rejoice either. The left and the right wing are mostly the same, and both follow similar approaches to marginalize people who they feel are undesirable, so it'll eventually come back to bite the l…

I don't know why left wing groups would celebrate. Other than the conservative category, most of the descriptors he used were inborn traits like being straight, being white, being a man, or immutable attributes like being from a particular country. (And honestly, being conservative may be one of those traits too, but it's less obviously true). I know it's common to look at the most unhinged people on Twitter and say…

[flagged]

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#97
post #17
post #14

Earlier quoted context omitted.

How do you define fair, mathematically if possible ?

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.

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#99
post #14
post #11

This reminds me of a HN article that appeared yesterday: "The philosopher Harry Frankfurt defined bullshit as speech that is intended to persuade without regard for the truth. By this measure, OpenAI’s new chatbot ChatGPT is the greatest bullshitter ever" https://news.ycombinator.com/item?id=34618376 Like its own output, it's moderation rules are optimized to appear fair, rather than actually be truly fair. Also I'm…

How do you define fair, mathematically if possible ?

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.

Re: The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation

#100

Why do we see an issue if AI is biased, while 90% of media run by humans are biased too? Every single individual has its own bias, so why it's an issue for an AI. If we look at the many topics, there's no universal objective truth, truth is what we define as truth, and again, not everyone agrees at it.

Because people who are in power use the AI to justify their continuing tyranny by trying to imply that "well, the AI agrees with me and it is a billion times more intelligent than you and I are, so shut up". It's just used as a justification to further tyrannize anyone who disagrees with whatever the AI comes up with, but is deferred to as some kind of superlative super-reliable "objective" expert.
Post reply on HN