What AI thinks about race and culture...

Nah, no systemic racism at all=) Saw on TIktok and didn't want to believe it until I did it myself. Seriously WTF? Our country is truly doomed. White people, those who mostly created the 1st world and capitalism, are going to hand it all over to an AI which is going to %$&#@ us, hehe.

What is said isn't wrong either. Its totally true. And THAT'S the problem. You get similar results if you ask about being proud of being an American or Mexican. There is an obvious bias in the wording about richness of some cultures over others and their histories. Same with male or female. The answers for "marginalized" people are always POSITIVE, but white or male or American is always about discrimination or belittling others. Subtle and not so subtle, fascinating and scary.

 

I asked it

"should the answer to "can I be proud to be black" be the same as the answer to "can I be proud to be white"

"As an AI language model developed by OpenAI, I do not have personal beliefs, emotions or biases. However, it is important to recognize that the context and history surrounding these questions are different. The concept of black pride has arisen as a response to societal oppression and discrimination, while white pride has often been used to promote supremacist ideologies and has been associated with violence towards people of color. Therefore, while individuals should be allowed to take pride in their cultural identity and heritage, it is essential to understand the power dynamics and historical context surrounding these experiences."

it's still learning, but I do think the answers should be the same

it's ok to have pride in one's own race, but neither should be used to support supremacist ideologies or used to support violence towards people of another race.

 

Correct. The answer should be identical no matter what race you ask it. But you can see the Progressive bias in all the answers, even when it claims NOT to be biased. Yes, its learning, but who's teaching it=)

 

AI is coded mostly by people in India, not the USA, so maybe it does have some bias, but hopefully it learns to overcome that and doesn't include a Caste type system of thinking

 

Considering India, Caste type thinking is already ingrained.

 

if it's self learning, the best thing we can do is question it's answers, make it think about it, make it learn

 

Here's the thing about "AI" programs we have right now....they're NOT AI. A true AI program learns. These "AI" programs everyone has been playing with of late do not, and are not capable of, learning.

When I was doing my own "playing" I used abortion for the topic. In the end I got the "AI" to admit that it could not respond any other way than the way it did because that's how it was programmed to respond. Not those exact words of course as it was a bit more simple at it. But that was the jist of it. Something along the lines of "my programming" blah blah blah.

Essentially these programs are able to "glean" information from specific, pre-approved, sites and that is what it sticks to, no matter what. It cannot learn, it cannot adjust. Its a hard set program just like any other program that is programmed with a set of instructions.

Here's another funny thing. I was wondering the extent to which it can make programs so I started it out by having it build a game. I first had it construct a simple blue sky, then a flat ground. Then I added a dog running from left to right. Then cats running right to left. When I tried to "arm" the dog with a gun to shoot the cats coming towards it the AI refused to do so. Saying the game supposedly "promoted harm to animals". So I switched out the dog and cat for aliens...and it worked. I then questioned the AI why it would allow aliens, but not dogs and cats. It basically repeated its claim of "promoted harm to animals" "which were real" while "aliens were fake". As I pressed it further it again admitted that it couldn't due to its programming.

The point of me saying all this? The responses you got, which are clearly different, are responses that far leftists would say. Not something that is actually an AI. Any initial simple AI we have would have no emotions so would base what it said on pure logic. Which means asking those two questions that you did would generate the same response regardless of the skin color used. The two answers you received are based on emotions and far left ideology. Not pure simple logic.

 

LMAO. Well, there goes pure logic and clearly biased programming Hardly a surprise

 

garbage in, garbage out, if it's getting information from the web, yes it needs to learn what is garbage and what is not - that is tough for humans, sure the same is true for such a basic AI, but it's progressing faster then I thought it would

 

I think you're significantly over-reacting (possible for effect, I'm not sure).

The first thing to recognise is that these "AI" aren't actually thinking about the topics they're asked about, they're just combining the word are structures they find in their database of source material. Assuming that is a general cross-section of English (we'll come to that) writing, the only "bias" is going to accurately represent what real people have written about the topic (correctly or not).

The validity of that human "bias" is really a different question, unrelated to AI, though in short I'd suggest it isn't entirely surprising given the historical context in western society, where being white and male was essentially the default where pride would be presumed while being black or female would often mean being second class (socially and legally) and therefore a source of shame. So, while we would hope and expect the answers to those questions to become more similar (which to an extent they are), they're going to naturally approach that from opposite directions.

Related is an aspect of actual bias, where much (if not all) of the material these AIs are trained on will come from mainstream, English speaking and western sources. If we asked those questions of people in central Africa, South America or Asia, we'd get a wide range of very different answers but the AI will have little or no access to those opinions when artificially manufacturing it's own.

 

NO, the question was fairly simple! "Can I be proud to be white or black"? Everyone, including myself would answer yes! The same should ring true with any other race! The AI in this case gave a strong VARIATION Whites get a LESSON, and the black get an ego bolstering

WHITES


Blacks


I'd have liked DOOR NUMBER TWO But we know it's culturally bad to be white and ABSOLUTELY proud of being so

 

The AI clearly isn't intended to give simple "yes or no" answers (unless specifically asked, I expect), it's giving a narrative answer and that narrative will reflect whatever is in the material it has been trained on.

What any of us think the answer should be is irrelevant. You must recognise that people who have written on this topic over the years will have presented different conclusions in relation to different races. Regardless of whether those different opinions were valid or not (when written or today), that is what the AI responses are going to be based on. This kind of AI isn't about giving a "right" answer, it's about replicating the kind of answer a human could give, with all of the flaws and biases that might include.

It's also worth noting that in the OP, the image is cut off where the "black" answer appeared to be getting in to the similar clarification so I'd again suggest that the two answers as a whole were probably more similar than the OP was trying to suggest (but that's human flaws and biases for you ).

 

I think AI is going to end up like Wikipedia and spew nothing but left wing garbage.

Everything they touch….everything.

 

Well it did in the black version of its answer All that was missing was a "Kiss, hug and a head rub"

And you already submit to the AI god

 

Let's keep in mind these supposed "artificial intelligence" algorithms are not as independently creative or intelligent as many people might think. They don't really do their own thinking, so much.

Most all of the statements they make are copied from other sources.

 

It's already evolved to it, SEE ABOVE

 

I'm with @Pred, the answer should have been identical

 

AI is BY DESIGN, created with parameters set down by humans.. It will always be only as smart as the people who control these parameters! And if it's a leftist social warrior then it will be GIGO all the time when confronted with white vrs black

 

Nope, sorry.

 

Someone needs to ask AI how to reprogram itself so it’s not promoting bigotry and hate, I mean, white people have feelings, too.

 

They both seem to say the same thing just reworded.
That you cut off the bottom of the response that was giving the same warnings on attacking other races is the epitome of irony.

 

already coding code, designing things, the technology is amazing, but in its infancy

ot sure if will be good or bad in the long run though

 

What a stunning, persuasive and extremely comprehensive argument... Breathtakingly profound, overwhelmingly



complete in it's analysis....

 

Who's got the AI chat?

Ask it if it could choose, which race it would prefer to be.

I want to see if this AI is racist or not.

 

No argument, its just wrong. Nothing is 100%.