20 years after Mark Zuckerbergās infamous āhot-or-notā website, developers have learned absolutely nothing.
Two decades after Mark Zuckerberg created FaceMash, the infamously sexist āhot-or-notā website that served as the precursor to Facebook, a developer has had the bright idea to do the exact same thingāthis time with all the women generated by AI.
A new website, smashorpass.ai, feels like a sick parody of Zuckerbergās shameful beginnings, but is apparently meant as an earnest experiment exploring the capabilities of AI image recommendation. Just like Zuckās original site, āSmash or Passā shows images of women and invites users to rate them with a positive or negative response. The only difference is that all the āwomenā are actually AI generated images, and exhibit many of the telltale signs of the sexist bias common to image-based machine learning systems.
For starters, nearly all of the imaginary women generated by the site have cartoonishly large breasts, and their faces have an unsettling airbrushed quality that is typical of AI generators. Their figures are also often heavily outlined and contrasted with backgrounds, another dead giveaway for AI generated images depicting people. Even more disturbing, some of the images omit faces altogether, depicting headless feminine figures with enormous breasts.
According to the siteās novice developer, Emmet Halm, the site is a āgenerative AI party gameā that requires āno further explanation.ā
āYou know what to do, boys,ā Halm tweeted while introducing the project, inviting men to objectify the female form in a fun and novel way. His tweet debuting the website garnered over 500 retweets and 1,500 likes. In a follow-up tweet, he claimed that the top 3 images on the site all had roughly 16,000 āsmashes.ā
Understandably, AI experts find the project simultaneously horrifying and hilariously tonedeaf. āItās truly disheartening that in the 20 years since FaceMash was launched, technology is still seen as an acceptable way to objectify and gather clicks,ā Sasha Luccioni, an AI researcher at HuggingFace, told Motherboard after using the Smash or Pass website.
One developer, Rona Wang, responded by making a nearly identical parody website that rates menānot based on their looks, but how likely they are to be dangerous predators of women.
The sexist and racist biases exhibited by AI systems have been thoroughly documented, but that hasnāt stopped many AI developers from deploying apps that inherit those biases in new and often harmful ways. In some cases, developers espousing āanti-wokeā beliefs have treated bias against women and marginalized people as a feature of AI, and not a bug. With virtually no evidence, some conservative outrage jockeys have claimed the oppositeāthat AI is āwokeā because popular tools like ChatGPT wonāt say racial slurs.
The developerās initial claims about the siteās capabilities seem to be exaggerated. In a series of tweets, Halm claimed the project is a ārecursively self-improvingā image recommendation engine that uses the data collected from your clicks to determine your preference in AI-generated women. But the currently-existing version of the site doesnāt actually self-improveāusing the site long enough results in many of the images repeating, and Halm says the recursive capability will be added in a future version.
Itās also not gone over well with everyone on social media. One blue-check user responded, āBro wtf is this. The concept of finetuning your aesthetic GenAI image tool is cool but you definitely could have done it with literally any other category to prove the concept, like food, interior design, landscapes, etc.ā
Halm could not be reached for comment.
āIām in the arena trying stuff,ā Halm tweeted. āSome ideas just need to exist.ā
Luccioni points out that no, they absolutely do not.
āThere are huge amounts of nonhuman data that is available and this tool could have been used to generate images of cars, kittens, or plantsāand yet we see machine-generated images of women with big breasts,ā said Luccioni. āAs a woman working in the male-dominated field of AI, this really saddens me.ā
Removed by mod
If I follow your reasoning, blackface isnāt racist because it doesnāt involve real black people?
no, a generated image of a black person would not be racist in itself.
So your understanding of something like racism is that it can only be racism if it involves a live real human? What about racist fiction? I donāt think your idea of sexism or racism holds upā¦
is a generated image of a white person racist?
we havenāt discussed anything about the details of this theoretical image of a black person. in order for it to be racist, the very act of depicting a black individual would have to be a racist act in and of itself.
now if the image somehow reinforces, or perhaps exemplifies, racist stereotypes, then perhaps it would be racist.
The details are literally women being objectified, and blackface. You have intentionally chosen to not read words toā¦ What exactly are you trying to achieve here? Actually read the post and the comment you have responded to and think deeply about your rhetoric and the meaning and time you chose to dedicate.
someone tried to shift the narrative to racism, comparing generating ai images of women to wearing blackface, asking the above poster if they thought that wasnt racist. I donāt think these are in any way equivalent, or even related, so I provided an example i considered comparable, and asserted that that would not be racist.
I donāt try to shift the narrative, I use the same method than Simone De Beauvoir in The Second Sex to highlight a discrimination.
Generating the picture of a black person is not racist (the AI could have bias, but thatās an other subject). But generating pictures of persons in different skin tons on a website called apartheid.ai and making people vote āwhite or blackā, that would be racist.
The problem here is not generating picture of women, but how these pictures are used.
iām sorry, then, i misunderstood - and if, as someone else suggested, you were simply being hyperbolic in bringing up blckface as a comparative example, then iām sorry i didnāt catch that.
We are certainly agreed that the circumstances surrounding and contained within an image impact the harm it does, be it racism or sexism or anything else.
Which begs the questions, do you know that the person you initially responded to was using hyperbole to draw out a point from the person they were responding to? Do you not care for analogy? Why did you choose to specifically not read the comment you were responding to to make a tangential point after being grilled for two additional comments to actually make your point?
The most charitable explanation to all of the above is that youāre here to win an argument, content and rhetoric or implications or literal interpretations of your comments text, be damned, despite otherwise actually (nominally) agreeing in no uncertain terms with the commenter you responded to negatively.
Wait, are you implying that in order for this app to be sexist, the very act of depicting a female individual would have to be a sexist act in and of itself?
Because I donāt think the author of the article is arguing that, nor anyone in here.
no, not at all. but the other person who was arguing me seemed to be saying that they thought I was wrong for saying an ai generated image of a black person wouldnāt automatically be racist.
this app isnāt sexist because it generates images of women, itās sexist (or not, since thatās whatās being debated) because it lets men (people) rate them, and perhaps because it seems to generate female images that overemphasise features that are considered to appeal to the male gaze.
personally, Iām unsure if I consider this app sexist. i would say that rating real women this way definitely is, but is it sexist to ask (your audience, or viewers) if a painting of a woman is attractive? even if itās of a fictional woman? what if the intent is to appeal to people who are axially attracted to women? thereās a lot of pornographic art out there, is it sexist to make these images?
the sexist part here, if anything, seems to be giving people the opportunity to rate the fictional women, and as i said, I think it is sexist to do that to real people, so even if this app isnāt sexist per se, Iād still consider it bad if it encourages people to do that to actual women. but if people only behave that way in the context of the app, then I think itās at worst harmless and possibly even beneficial, if it gives a harmless outlet for some urge which would otherwise be inflicted on real women.
I donāt think this situation is nearly as clear cut as most people seem to be taking it to be, in either direction.
I see what youāre saying. I think a better example to test what you are saying about real vs imaginary people would be if there was a realistic app where you whipped AI generated black people with a virtual whip and made them dance for watermelon.
Would that app be non racist simply because the depicted people are not real?
Would making the app/using the app be non-racist?
Note Iām not trying to say whipping people is equivalent to rating their looks. Obviously itās not. Iām just making a thought experiment to unpack this idea that imaginary interactions canāt be -ist.
oh, iām not denying in the slightest that interactions with imaginary people can still be racist, sexist, whatever. even if you were rating AI generated people of colour according to their looks, or probably nearly any other criteria, that could very well be racist.
In regards to the original topic, I just think thereās too much to unpack to give an easy verdict of sexist or not, at least for myself. But, i donāt know, if it were an app that rewarded you for discriminating (or abusing, definitely) someone, even an entirely fictional person, that would definitely be sexist, or racist, or whatever else.
Sexism in fiction has been discussed for quite long time
But see theyāre made up. If the characters arenāt real, who could possibly suffer the effects of media intentionally objectifying women or otherwise reducing a group of people into caricatured stereotypes- ohhhhhhhhhhhhh
This type of āparty gameā is still at itās core objectifying women. They may be generated images but the whole project is aimed at passing judgement on women you would rate as fuckable or not. Itās encouraging behaviour that makes women feel uncomfortable or unsafe.
This type of objectifying isnāt exclusive to this project. Groups of men will rate and objectify women casually and frequently. Iāve worked in the trades and have been surrounded by such talk from men. The more normalized this type of behaviour is, the easier it is to consider women as less than human. Feeling like a replaceable tool with no sense of self or sense of worth is dehumanizing.
They could have chosen to base this project on just about anything else in our world. We have animals, nature, technology and so much more to try this kind of thing out on. Yet, what seems like another ātech broā idea was focused on hyper sexualizing and objectifying women as if they were just another thing for menās entertainment.
Simply, itās gross behaviour. Just because they are generated images does not make it any less gross or acceptable. People are not objects for another personās amusement and we should not encourage such behaviour.
This is like dunking on DeviantArt because it has artists who make cheesecake pictures of ladies. Iām not saying itās something I personally enjoy, but who am I to tell others what art they should enjoy?
The only way for this to be consistent is if you believe authorial intent or real practical effects on an audience have no bearing on the properties of a piece of media.
As long as itās fiction, itās okay?
Unless the author writes an essay to accompany their piece, I think any conclusion you make about authorial intent is speculative. A beefcake pic of a guy in speedos lifting weights could be sexual, or maybe the artist is doing a study in human musculature? Heck if I know.
Effects on the audience, Iām not sure I understand that. Itās up to the audience to decide whether they like something, or not, or whether they are happy with whatever āeffectsā it has on them. The effect most are interested in is āpleasureā, I think. If one doesnāt like the pics, one is not in the audience for that art.
If one wants to make the argument that folks shouldnāt look at cheesecake or beefcake pics, because they create some sort of problem for the viewer, the onus is on the claimant to win the hearts and minds of the audience. As long as all parties are consenting adults making informed decisions, I donāt see the issue.
I do concur that it could be āsexistā in the same sense that anybody discriminating based on sexual preference is sexist, but Iām not sure that is wrong. Someone who prefers lady types as sexual partners may prefer to look at cheesecake pics of lady types, I guess, and thatās technically sexist because theyāre choosing those pics based on lady characteristics.
Now if you want to argue that such pics have downstream effects on a vulnerable/disempowered population, that would be a different argument.
We have no control over who we are attracted to sexually (or not at all), but we do have control over how we interact with the world. Who you are attracted to cannot be sexist, racist, etc. because there is no intention - it merely is. Being attracted and choosing to objectify someone are two very distinct processes because one involves intention. Discrimination is also an act of intent.
What āsomeoneā is being objectified in this case?
Iām merely explaining why it is not analogous and why attraction cannot be considered bigoted. Anything that involves intent can be criticized for bigotry if it is present.
Thatās fair. Thank you.
That is literally the argument in question about this whole post. š¤¦
Your rant about not being able to do any rhetorical analysis without an author spelling it out for you is really not my problem. Maybe donāt criticize it if you have no practice doing it in the first place.
Your willful misunderstanding of how objectification in fiction can ever be any more problematic than ādiscrimination based on sexual preferencesā is justā¦ Wow.
I can only respond to the complaint you made:
ā¦ not the one you imagine you made.
To be clear, I disagree with this:
To clarify, I donāt think the authorās intent really matters in art. If one is interested in context, then itās a useful context.
In this case, the images have no āauthorā, theyāre a machine output, so Iām not sure how you think authorial intent figures in this.
EDIT: My mistake, Iām mixing up responses. I should further clarify that, in the case of cheesecake/beefcake pics on DeviantArt (the example I gave), there clearly is an author/artist. But ultimately Iām still not sure it matters what their intent is. Do they like drawing lingerie as an artistic subject, or do they like drawing ladies for sexual titillation? Iām not sure there is any moral imperative on the viewer to care.
Iām curious now if vegans get mad at lab grown meat for similar reasons.
Jumping into the feminism community to challenge a fairly core tenant of feminism is a bad take. Iām removing this because it was a comment made clearly in bad faith. Youāre expected to be nice on our instance, do better in the future.
Removed by mod
Not the case at all, enjoy a 14 day ban. Try thinking about what you did wrong.