Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

CaptionBot by Microsoft

captionbot.ai

Post reply on HN
Print view View original

141–150 of 170 posts

Re: CaptionBot by Microsoft

#141

Post by jnky »

mc32 wrote: post #139
readams wrote: post #114

Earlier quoted context omitted.

People in general look very much like gorillas. They are close cousins after all. Telling humans and other apes apart can't be very easy for a poor, overworked neural network.

Also why are people offended by gorillas, or apes. Imagine being mislabeled as a bear or a giraffe, would people get as offended? What about bring identified as a parrot?

I get where you're coming from, but black people being compared to apes is kind of an age old racist trope, so it is understandable that people would be unhappy about it.

Being mislabeled as a bear on the other hand does not generally carry the negative connotations as calling people apes, so I suspects that this is why CapitionBot is erring on this side of the classification.

Re: CaptionBot by Microsoft

#142

Post by mc32 »

jnky wrote: post #141
mc32 wrote: post #139

Earlier quoted context omitted.

Also why are people offended by gorillas, or apes. Imagine being mislabeled as a bear or a giraffe, would people get as offended? What about bring identified as a parrot?

I get where you're coming from, but black people being compared to apes is kind of an age old racist trope, so it is understandable that people would be unhappy about it. Being mislabeled as a bear on the other hand does not generally carry the negative connotations as calling people apes, so I suspects that this is why CapitionBot is erring on this side of the classification.

I'm aware people may want to _use_ it as a racist tool --but I fail to understand that transformation. We're all apes[1] the offender and the target/victim. So more than anything, it shows ignorance by the victimizer/bigot.

I suppose it's like calling someone a Neanderthal, but with racial implications. Interestingly, we're learning Neanderthals were not as cognitively lacking as perhaps some supposed.

It's like saying, hey, you are hairy... Right, we're all hairy (bald or not) with very few exceptions. It's a weird thing.

[1] In Indonesian lang. people and apes share the "orang" root. Oran, orangutan, orangnakal, etc.

Re: CaptionBot by Microsoft

#143

Post by nols »

mc32 wrote: post #139
readams wrote: post #114

Earlier quoted context omitted.

People in general look very much like gorillas. They are close cousins after all. Telling humans and other apes apart can't be very easy for a poor, overworked neural network.

Also why are people offended by gorillas, or apes. Imagine being mislabeled as a bear or a giraffe, would people get as offended? What about bring identified as a parrot?

C'mon, there's a long history of calling black people monkeys or apes as a way to remove their humanity. It's willful ignorance to ignore that history.

Re: CaptionBot by Microsoft

#144

Post by taneq »

colllectorof wrote: post #138
jnky wrote: post #98

I found it curious that this Bot is really bad at recognizing apes: chimpanzees and gorillas specifically. I fed it a lot of the images from a Google image search for these animals and more often than not it either doesn't recognize anything or considers them bears. I don't mean to offend, but I'm left wondering if the creators of image recognition services disincentivize their neural nets from recognizing something…

Reading the blog post and the Twitter thread. Wow. 10 years ago it would have been seen as a comical blunder of a stupid AI. Something to be fixed for sure, but not a Serious Social Issue by any means. Nowadays, it apparently warrants several follow-up articles in the mainstream media, "social" commentary and 3,309 retweets. From the blog post: “The bias of the Internet reflects the bias of society,” she said. In som…

The issue here is that race relations in the U.S. are so convoluted and have so much history that it's literally impossible to tell what would be considered "racist" without a broad understanding of the culture and a comprehensive list of past racial slurs and grievances.

Witness the KFC ad (https://www.youtube.com/watch?v=ZaIhf41ctkM) which was broadly labeled as "racist", despite the fact that the 'black people love fried chicken' stereotype is (as far as I know) only a U.S. construction.

Re: CaptionBot by Microsoft

#145

Post by jnky »

mc32 wrote: post #142
jnky wrote: post #141

Earlier quoted context omitted.

I get where you're coming from, but black people being compared to apes is kind of an age old racist trope, so it is understandable that people would be unhappy about it. Being mislabeled as a bear on the other hand does not generally carry the negative connotations as calling people apes, so I suspects that this is why CapitionBot is erring on this side of the classification.

I'm aware people may want to _use_ it as a racist tool --but I fail to understand that transformation. We're all apes[1] the offender and the target/victim. So more than anything, it shows ignorance by the victimizer/bigot. I suppose it's like calling someone a Neanderthal, but with racial implications. Interestingly, we're learning Neanderthals were not as cognitively lacking as perhaps some supposed. It's like sayi…

I don't think there is as much to understand as you make it out to be. People feel insulted when called apes, because it is frequently used as an insult.

It may well be taxonomically correct to say that humans as well as apes are primates and members of the Hominidae family, but that doesn't make it any less insulting.

Just because something is technically correct doesn't mean it isn't dehumanizing and suitable as an insult. I'm sure nobody would like to be called a sack of meat either.

Re: CaptionBot by Microsoft

#149

Post by jnky »

taneq wrote: post #144
colllectorof wrote: post #138

Earlier quoted context omitted.

Reading the blog post and the Twitter thread. Wow. 10 years ago it would have been seen as a comical blunder of a stupid AI. Something to be fixed for sure, but not a Serious Social Issue by any means. Nowadays, it apparently warrants several follow-up articles in the mainstream media, "social" commentary and 3,309 retweets. From the blog post: “The bias of the Internet reflects the bias of society,” she said. In som…

The issue here is that race relations in the U.S. are so convoluted and have so much history that it's literally impossible to tell what would be considered "racist" without a broad understanding of the culture and a comprehensive list of past racial slurs and grievances. Witness the KFC ad ( https://www.youtube.com/watch?v=ZaIhf41ctkM ) which was broadly labeled as "racist", despite the fact that the 'black people l…

> The issue here is that race relations in the U.S. are so convoluted and have so much history that it's literally impossible to tell what would be considered "racist" without a broad understanding of the culture and a comprehensive list of past racial slurs and grievances.

I understand, but isn't the logical conclusion that one cannot make a piece of technology (like an image caption bot) that is unaware of, for instance, such complex racial relations? And if so, should technological progress really be hampered by people's sense of outrage, even in the absence of malicious intent?

Re: CaptionBot by Microsoft

#150

Post by gengkev »

colllectorof wrote: post #138
jnky wrote: post #98

I found it curious that this Bot is really bad at recognizing apes: chimpanzees and gorillas specifically. I fed it a lot of the images from a Google image search for these animals and more often than not it either doesn't recognize anything or considers them bears. I don't mean to offend, but I'm left wondering if the creators of image recognition services disincentivize their neural nets from recognizing something…

Reading the blog post and the Twitter thread. Wow. 10 years ago it would have been seen as a comical blunder of a stupid AI. Something to be fixed for sure, but not a Serious Social Issue by any means. Nowadays, it apparently warrants several follow-up articles in the mainstream media, "social" commentary and 3,309 retweets. From the blog post: “The bias of the Internet reflects the bias of society,” she said. In som…

Yes, I think people overreacted a bit. But part of the backlash was because if these algorithms were better trained on a diverse set of faces, they might not have made that mistake. I think that's a fair criticism.
Post reply on HN