Can AI be rascist

[cradleandshoot]

I just read this article today. I need input from all you highly educated folks on this forum. Is AI racist????
Sorry no link but a quick Google search will get you up to speed.
The ACLU has waded in with their opinion.

[Typical Lax Dad]

I just read this article today. I need input from all you highly educated folks on this forum. Is AI racist????
Sorry no link but a quick Google search will get you up to speed.
The ACLU has waded in with their opinion.

Can you post a link to the article?

[jhu72]

... YES!

Racism in racism out! You put data from American culture (Western culture) into an AI engine, you get American / Western culture out.

[cradleandshoot]

I just read this article today. I need input from all you highly educated folks on this forum. Is AI racist????
Sorry no link but a quick Google search will get you up to speed.
The ACLU has waded in with their opinion.

Can you post a link to the article?

Will do when I have the chance.

[Typical Lax Dad]

... YES!

Racism in racism out! You put data from American culture (Western culture) into an AI engine, you get American / Western culture out.

I wanted to see how the article Cradle read differs. This isn’t “racism” but how and who data is collected from when software is developed matters:

https://www.medscape.com/viewarticle/99 ... sachusetts.

The implications stretch to wearable medical devices and the relative accuracy of said devices. It’s a big market.

[NattyBohChamps04]

From 7 years ago: Twitter taught Microsoft’s AI chatbot to be a racist a-hole in less than a day

[Andersen]

I think it makes since to think of AI as more of a plagiarism tool that puts together bits and pieces it gathers than actually creating anything independently.

[PizzaSnake]

Of course it can. We are making a "god" in our own image. Are the tech bros who conceptualize and reify these algorithms and probability distributions racist? Some, undoubtedly.

[youthathletics]

Worth a listen if you want to learn more about it from the guy in charge of leading the OpenAI Devgru team. In a nut shell Ai learns from our behaviors, so if the democrats have anything to do with it, it will racists as hell before you know it.

[MDlaxfan76]

I think the point is that we should not expect, certainly at this stage, for any "AI" to have an independent moral compass.

It will not be able to distinguish morality, absent programming to do so.

Morality requires putting another person's interest ahead of one's own, especially in extremis, and unless that is programmed, it's definitely not a sure thing in a world that is largely focused on self-interest and self-preservation.

On a more basic level, bias is something that humans have developed through evolution to provide self-preservation advantages, enabling fast recognition of danger through various cues of difference. Exposure to difference can mute or moot perception of difference as dangerous, but absent such, the evolutionary impulse is to perceive, and emotionally/physiologically respond to, such as danger.

In conjunction with bias, we also have an evolutionary impulse to work in social groups together to create collective advantage.

It requires a moral ability to overcome this evolutionary tendency to favor groups that 'look like us'...the question is, what will AI "learn" from our behaviors...?

Will it see humanity as one, or will it see groups, as we tend to do absent significant effort?
Will it see itself as part of humanity, or separate?

And what would that mean to the AI?

[PizzaSnake]

I think the point is that we should not expect, certainly at this stage, for any "AI" to have an independent moral compass.

It will not be able to distinguish morality, absent programming to do so.

Morality requires putting another person's interest ahead of one's own, especially in extremis, and unless that is programmed, it's definitely not a sure thing in a world that is largely focused on self-interest and self-preservation.

On a more basic level, bias is something that humans have developed through evolution to provide self-preservation advantages, enabling fast recognition of danger through various cues of difference. Exposure to difference can mute or moot perception of difference as dangerous, but absent such, the evolutionary impulse is to perceive, and emotionally/physiologically respond to, such as danger.

In conjunction with bias, we also have an evolutionary impulse to work in social groups together to create collective advantage.

It requires a moral ability to overcome this evolutionary tendency to favor groups that 'look like us'...the question is, what will AI "learn" from our behaviors...?

Will it see humanity as one, or will it see groups, as we tend to do absent significant effort?
Will it see itself as part of humanity, or separate?

And what would that mean to the AI?

“Morality requires putting another person's interest ahead of one's own, especially in extremis, and unless that is programmed,”

But how to program morality? How would items like the Trolley Problem (I, Robot film had an interesting framing of same) be “programmed”? From what human construct that addresses these questions? Which “religion” or philosophical tradition?

Humans can’t/haven’t been able to come to any consensus of this topic for, well, forever (as far as we can know). Now that we are making a “god” in our own image, who shall be the exemplar? Same old song, just the stakes have gotten higher. I’m going to predict human misery and suffering as s result of this little innovation (like most human artifice). Debating the relative benefit will be left as an exercise for the reader.,,

[MDlaxfan76]

I think the point is that we should not expect, certainly at this stage, for any "AI" to have an independent moral compass.

It will not be able to distinguish morality, absent programming to do so.

Morality requires putting another person's interest ahead of one's own, especially in extremis, and unless that is programmed, it's definitely not a sure thing in a world that is largely focused on self-interest and self-preservation.

On a more basic level, bias is something that humans have developed through evolution to provide self-preservation advantages, enabling fast recognition of danger through various cues of difference. Exposure to difference can mute or moot perception of difference as dangerous, but absent such, the evolutionary impulse is to perceive, and emotionally/physiologically respond to, such as danger.

In conjunction with bias, we also have an evolutionary impulse to work in social groups together to create collective advantage.

It requires a moral ability to overcome this evolutionary tendency to favor groups that 'look like us'...the question is, what will AI "learn" from our behaviors...?

Will it see humanity as one, or will it see groups, as we tend to do absent significant effort?
Will it see itself as part of humanity, or separate?

And what would that mean to the AI?

“Morality requires putting another person's interest ahead of one's own, especially in extremis, and unless that is programmed,”

But how to program morality? How would items like the Trolley Problem (I, Robot film had an interesting framing of same) be “programmed”? From what human construct that addresses these questions? Which “religion” or philosophical tradition?

Humans can’t/haven’t been able to come to any consensus of this topic for, well, forever (as far as we can know). Now that we are making a “god” in our own image, who shall be the exemplar? Same old song, just the stakes have gotten higher. I’m going to predict human misery and suffering as s result of this little innovation (like most human artifice). Debating the relative benefit will be left as an exercise for the reader.,,

yup, that's the problem.

[Farfromgeneva]

I think the point is that we should not expect, certainly at this stage, for any "AI" to have an independent moral compass.

It will not be able to distinguish morality, absent programming to do so.

Morality requires putting another person's interest ahead of one's own, especially in extremis, and unless that is programmed, it's definitely not a sure thing in a world that is largely focused on self-interest and self-preservation.

On a more basic level, bias is something that humans have developed through evolution to provide self-preservation advantages, enabling fast recognition of danger through various cues of difference. Exposure to difference can mute or moot perception of difference as dangerous, but absent such, the evolutionary impulse is to perceive, and emotionally/physiologically respond to, such as danger.

In conjunction with bias, we also have an evolutionary impulse to work in social groups together to create collective advantage.

It requires a moral ability to overcome this evolutionary tendency to favor groups that 'look like us'...the question is, what will AI "learn" from our behaviors...?

Will it see humanity as one, or will it see groups, as we tend to do absent significant effort?
Will it see itself as part of humanity, or separate?

And what would that mean to the AI?

“Morality requires putting another person's interest ahead of one's own, especially in extremis, and unless that is programmed,”

But how to program morality? How would items like the Trolley Problem (I, Robot film had an interesting framing of same) be “programmed”? From what human construct that addresses these questions? Which “religion” or philosophical tradition?

Humans can’t/haven’t been able to come to any consensus of this topic for, well, forever (as far as we can know). Now that we are making a “god” in our own image, who shall be the exemplar? Same old song, just the stakes have gotten higher. I’m going to predict human misery and suffering as s result of this little innovation (like most human artifice). Debating the relative benefit will be left as an exercise for the reader.,,

Buddhism

[youthathletics]

Can AI be racist....you betcha, so much so, they got called out big time. Plenty of other articles if you google.

[jhu72]

... the stuff being called AI is in my opinion not really AI. It is more marketing hype than anything else. The problems being solved / addressed today, with very very few exceptions can be solved without what is being called AI. Traditional programming paradigms / techniques / algorithms can solve the same problems (for the most part) and frequently using less computing power. Until we have a real AGI solution, AI will simply be marketing hype.

[jhu72]

Can AI be racist....you betcha, so much so, they got called out big time. Plenty of other articles if you google.

... this is the result of telling your AI machine to adjust for western culture's racism -- your input data. Don't do the adjustment and you get the original racism in your western culture. If you think this is the biggest problem with AI, you don't understand how AI works.

[PizzaSnake]

... the stuff being called AI is in my opinion not really AI. It is more marketing hype than anything else. The problems being solved / addressed today, with very very few exceptions can be solved without what is being called AI. Traditional programming paradigms / techniques / algorithms can solve the same problems (for the most part) and frequently using less computing power. Until we have a real AGI solution, AI will simply be marketing hype.

The bitcoin fools needed a new object of affection.

[jhu72]

... the stuff being called AI is in my opinion not really AI. It is more marketing hype than anything else. The problems being solved / addressed today, with very very few exceptions can be solved without what is being called AI. Traditional programming paradigms / techniques / algorithms can solve the same problems (for the most part) and frequently using less computing power. Until we have a real AGI solution, AI will simply be marketing hype.

The bitcoin fools needed a new object of affection.

... could be. It certainly seems to have the same flavor.

[Farfromgeneva]

... YES!

Racism in racism out! You put data from American culture (Western culture) into an AI engine, you get American / Western culture out.

Maybe folks need to understand AI = Large Language Models. Garbage in garbage out.

[MDlaxfan76]

AI would likely spell check correctly the title of this thread.