For AI to stop making our double chins disappear


The artificial intelligences used to make pretty selfies continue to create distorted images of us, to match unrealistic ideals of beauty. In this week’s #Rule30 newsletter, journalist Lucie Ronfaut wonders how far this artificial glamorization of our images will go.

I don’t know if you are aware, but it seems that the robots have won. In any case, they have invaded my news feeds in recent days. On the one hand, technophiles marveled at the prowess of ChatGPT, a tool developed by the OpenAI company (already responsible for the Dall-E drawing robot), capable of answering questions and text commands in an automatic way. As often when we talk about artificial intelligence, it is sometimes stunning, sometimes scary, and in both cases it raises serious ethical questions. The same goes for another tool: Lensa AI.

Lensa AI is an application that has existed since 2018, and which allows you to edit your photos and then publish them on social networks. It has recently gained popularity thanks to a new feature, dubbed “ magic avatars “. Against 8 dollars, you upload a dozen photos of yourself, then you get a selection of avatars in different artistic styles. To do this, the application uses Stable Diffusion, a machine learning model capable of generating images from text commands (an example offered by the official site: “the spirit of a Tamagotchi that haunts the streets of Paris “). It has been available in open-source since the beginning of the year, leading to a wave of services offering to generate content ranging from personalized Pokémon to pornographic drawings.

Stable Diffusion quickly became the subject of much criticism, especially in the artistic community, which accused it of theft. Another point of tension is that (alas, classic) of cybersecurity. You could answer me that you are not naïve or naïve, that in any case your personal data has been fraudulently exploited for a long time, and that at least this exploitation is fun. Certainly ! But the discomfort I feel in front of Lensa is more about another subject, so far little discussed: that of her beauty.

In fact, these avatars are intended to make you feel beautiful or beautiful, so that you then want to share them online. On Twitter, many are happy that Lensa gave them back or gave them back hot, that is to say more canon than in reality. I find this idea quite disturbing. Without wanting to play it to me as a philosophy teacher in the final year, what is beauty? For Lensa, the answer will be different depending on your starting appearance. For example, when using the application as a woman, there is a risk of being represented half-naked and with a generous chest. Equally disturbing, author Audrey Gordon, who is fat, noticed that the app systematically slimmed her down, refusing to represent her double chin.

“Lensa really insists on portraying me as a thin person, when I provided her with particularly glorious examples of my double chin. What sadness, what boredom. (Audrey Gordon is an author)

“We are training artificial intelligences to flatter us”

Any artist will tell you that portraiture is a complex exercise. Do we have to represent the person faithfully? Let our perception speak for itself? Make him happy, by improving characteristics considered to be faults? To this must be added the context of our society and its sexist, racist biases, etc., which influence the production of content and its reception. In the 1970s, feminist critic Laura Mulvey popularized the concept of male gaze: how social and political inequalities between men and women are expressed in the representation of genders in cinema, which will itself reinforce these inequalities.

Except that here, we are not talking about an artist and their human sensibility, but about an AI (artificial intelligence) and its automatic reinterpretation of the world. So what perception are we talking about? That of the developers who designed it? That of the database on which it was trained? That of a person who secretly hopes to be improved » by a lark mirror? “We are training artificial intelligences to flatter us “Alerts investor Hunter Walk, former Google and Second Life, in a blog post. “It’s kind of a vicious circle in which the best, fake version of ourselves wins the competition. In the long run, we risk having models who prefer to tell us what we want to hear, rather than the truth.I don’t know if AIs will replace artists (and, honestly, I doubt it). But I know that, already, they are promised a bright future in areas where we need rapid content production, such as advertising, special effects or the publication of dispatches. We should therefore worry about the representations of AIs, think about who influences them. And besides, if we can, and want to, modify them. Should we fear the gauze AI, or our ego?

The data transmitted through this form is intended for PressTiC Numerama, in its capacity as data controller. These data are processed with your consent for the purpose of sending you by e-mail news and information relating to the editorial content published on this site. You can oppose these e-mails at any time by clicking on the unsubscribe links present in each of them. For more information, you can consult our entire personal data processing policy.

You have a right of access, rectification, erasure, limitation, portability and opposition for legitimate reasons to personal data concerning you. To exercise one of these rights, please make your request via our dedicated rights exercise request form.

The press review of the week

At the post (1)

Ultia is a popular French streamer on the Twitch platform. Unfortunately, she is also the regular target of very violent harassment campaigns, for having publicly denounced the sexist remarks of one of her colleagues last year. In an interview given to Mediapart, she announces to file a complaint against her attackers, and denounces (once again) the lack of moderation on the major online platforms. Watch it here.

At the post (2)

French YouTuber Norman Thavaud was taken into police custody on Monday, questioned about corruption of minors and rape potentially involving several teenage girls and young women. The creator had already been accused in 2020 of psychological influence by a Quebec fan, 16 years old at the time of the alleged facts. You can read the article from Releasewho disclosed the information, or that of Numeramafor more details on this case.

Chaos

I try not to devote too much space in this newsletter to Elon Musk and his progressive dismantling of Twitter. Nevertheless, I wanted to share with you this editorial by US Senator Scott Wiener, known for his work in favor of LGBT rights in the United States, and violently harassed because of this commitment. He details the multiplication of attacks against him since the change of direction at Twitter: “ Elon Musk can have fun causing chaos, we pay the price“, he concludes. It is to be read (in English) here.

Cartridge

Last week, video game industry pioneer Gerald Lawson was honored with the Google Doodle. Numerama looks back at the career of the American engineer, who died in 2011, who participated in the design of the first video game cartridge system. He was also known for his commitment to black people in the video game industry. To learn more about his career, visit Numerama.

Something to read/watch/listen to/play

unnamed (22)

I like the month of December in theory (listen to my Christmas playlist on repeat! Eating gingerbread until my stomach hurts!), but a little less in practice (eight hours of sunshine per day! The pressure of having succeeded in the year, and to do even better next year!). So I’m in the mood to consume sweet, melancholic fiction. Proofread The department store concierge is exactly what I need in this strange time, and maybe you too.

In this one-shot manga (which tells a complete story), we follow the uncertain steps of Akino, a new employee of a luxury department store whose specialty is to address animals, with a particularity: most have disappeared or are endangered. As a concierge, she must respond to the slightest wishes of her furry and feathered clients. A ferret tries to impress his boss, a sea mink wants to spoil his daughter for her birthday, a Honshu wolf hesitates to propose to her… It’s funny, comforting, and also a little bitter. Like a month of December, or a world that we will no longer know.

The Department Store Concierge, by Tsuchika Nishimura, Le Lézard Noir editions

The data transmitted through this form is intended for PressTiC Numerama, in its capacity as data controller. These data are processed with your consent for the purpose of sending you by e-mail news and information relating to the editorial content published on this site. You can oppose these e-mails at any time by clicking on the unsubscribe links present in each of them. For more information, you can consult our entire personal data processing policy.

You have a right of access, rectification, erasure, limitation, portability and opposition for legitimate reasons to personal data concerning you. To exercise one of these rights, please make your request via our dedicated rights exercise request form.


Understand everything about experimenting with OpenAI, ChatGPT



Source link -100