This text belongs to ‘Artificial’, the AI ??newsletter that Delia Rodríguez sends every Friday.
Dear readers, dear readers: A few months ago I decided to read Bret Easton Ellis from the eighties (Less than Zero, American Psycho), which led me to return to Historias del Kronen and Tesis, already in the mid-nineties and in Spain . I remembered the obsession at the time among communication students with snuff movies, a kind of urban legend about the existence of films where people were murdered just for the pleasure of recording them on video.
In the 80s and 90s, snuff movies did not exist, they were just a literary and cinematographic resource. In 2023, in beautiful Barcelona, ??there is a company where 20% of its workforce of 2,000 workers is on leave, traumatized by repeated exposure to images of extreme violence published on the internet that they must moderate. I imagine the people who have given their testimony to Ignacio Orovio and Gemma Saura of La Vanguardia getting dressed every day, signing in, getting a coffee and sitting at the computer to label “brutal murders, dismemberments, torture, rapes, live suicides, children being abused” for two thousand euros a month.
I recommend reading the full report, but here are some excerpts:
Social networks have robots that carry out a prior selection of the content that must go through human moderation. But they are usually people who give the first warning that something is not right and who make the final decision to eliminate content after examining it. According to Meta data, in the second quarter of 2023, 7.2 million contents were removed for sexual abuse of children; 6.4 million for suicide and self-harm from Facebook; 6.2 million for violence on Instagram; 17.5 million for hate speech on Facebook.
In theory, better technology could help ensure that this soul-toxic work is never done again by anyone. But experts, like lawyer and activist Cori Crider, doubt it: “the work of moderators is necessary and important, and unfortunately I don’t think we can ever completely derive it from artificial intelligence.” At the moment what we are seeing with generative AI is an increase in content. Including, for example, those of artificial child pornography. If the drive to infect the worst of human beings has been there for decades, and the Internet amplified it, let’s see what happens with a new technological leap.
As we said recently in this newsletter about the Almendralejo case of minors stripped naked with AI, legally, a real image is not the same as a synthetic one, and thank goodness. But its effects on our brains can be.
What else happened this week
– Important, and also related to human moderation. In Slovakia they had elections on September 30 and a couple of days before an audio created with AI appeared on Facebook where the leader of the pro-European party and a journalist talked about how to rig the elections by buying votes from the Roma minority. It was false, but apparently it found some holes in the system that caused it to spread more than it should: Meta’s policies only contemplate videos with manipulated voice and not audio, explains Wired. The suspicion of Russian disinformation interference hovers once again. The pro-Russian candidate Robert Fico won, by the way, although a cause-effect relationship cannot be demonstrated.
– Mark Zuckerberg and I still believe in the Metaverse. He has given an interview to Lex Friedman in which the two of them converse, face to face, in a virtual world through their own avatars using technology in Meta tests. The funny thing is that they are incredibly realistic, sometimes indistinguishable from themselves, and let’s remember that very recently we were laughing at their lack of quality. You have to watch the video, and also read this thread, in English, where someone who knows explains why the human brain, in reality, doesn’t need much to be immersed in another world: we are factory built for that.
– Google has presented its new Pixel 8 phones, which, as suspected, incorporate AI to improve images and videos. Magically delete what you don’t want, even a sound within a video, and choose and mix the best shot, for example changing your face in a group photo for another in which you are with your eyes open.
– Keep an eye on the audiobook market. Producing one costs between 3,000 and 4,000 euros, a figure that could be reduced by 75% with voices generated by generative artificial intelligence, almost indistinguishable from real ones, explains Antònia Justícia. “If now not all new audiobook releases can be produced due to the costs involved, it will be possible with voices produced through AI, which will mean an explosion for the sector,” says Maribel Riaza, advertising director at Storytel.
– Tom Hanks has denounced the unauthorized use of a digital version of him in an advertisement.
– The writer and mathematician specializing in artificial intelligence and robotics Carme Torras recommends books to Llucia Ramis. Among his own, La mutació sentimental and Estimades màquines. And among the latest outsiders that she has read, Machines Like Me, by Ian McEwan, and Klara and the Sun, by Kazuo Ishiguro.
– More books: non-human narratives compiled by Jorge Carrión.
– I haven’t seen The Creator yet, but it looks interesting. Justo Barranco says that he goes from some Western humans who face in 2065 to some robots who have taken advantage of the fact that they are not prohibited in New Asia to develop their own civilization. The director, Gareth Edwards, is a techno-optimist: “I like to think that if something is smarter than us, it will be better than us. “Many of the bad things that happen are due to stupidity, not intelligence.”
– Norberto Gallego explains what AI has to do with Adobe’s strategy, which has doubled its value despite spending a year waiting for regulators to authorize the purchase of Figma.
– Two fakes: the Tesla robot that does figure skating and the 1930 photo of the Loch Ness monster.
– Midjourney seems to be quite fine in the details, Dall-E 3 is good at comics and ChatGPT just has to show it an image and ask it to create the prompt capable of generating it.
– Two interesting reports from The Guardian, in English. One, about how AI is affecting the world of perfume, from custom algorithmic fragrances to market research via scans of consumers’ brains. The other, on the use of AI in predicting fashion trends, for example mining images from catwalks and social networks. Its paradoxes are interesting: for example, the popularity of bright yellow clothing may grow on social networks, but that does not mean that consumers want to buy it.
– Rewind, a pendant that records everything you say and hear, and that in combination with your mobile can make you go back to it at any time. In crowdfunding process, for $59. Via Xataka.
Level of IAnxiety this week: hoping that we don’t end up half of humanity moderating the content of the other half while the machines take care of the rest of the jobs. For now, I have booked a Rewind, let’s see when it arrives.
