Blockface and Birdsongs
Goodlinks 2023/10/09: Chinese Deepfake Influencers / NextGen Photo-Editing / Edgelords go Image Synthesis / Gaslighting Algorithms / Horror of Doomscrolling / Lucid Dreaming Headset / and much more
A Forthcoming Book Turns a New Leaf On Remarkable Photographs of Trees From Around the World. I love photographing trees, like this weird guy i shot at Wannsee Berlin over the summer, and this sounds like a book right down my alley.
Guess a Font: My first try placed me as a "Comic Sans Companion" and the game continued with "Hmm… you seem relatively new to the world of typography. Everyone has to start somewhere and you’re starting from the very beginning." There you have it.
Birdsongs Visualizer: Sound Circles takes DFT spectrograms of birdsongs and maps them onto a circle. Birds are cool.
FontShop is being shut down and integrated into MyFonts. It's the end of an era and FontShop, founded by typographer Erik Spiekermann (who was the boss of my former boss years before i became a graphic designer), was foundational for new typography in the 90s and their own FontFonts are among the best fonts you can find, with FF DIN and FF Meta being some of my all time favorites. RIP Fontshop.
This is awesome. Everybody should have a BlockFace - A Stamp Kit to Explore Typography: "BlockFace is an easy-to-use modular typography printing kit designed by graphic artist Will Mower. Combine the stamps to construct letterforms, design typographic posters and cards, and explore the joys of printmaking. (...) BlockFace is an open-source project, meaning you can freely download the files and guides and create your own kit as long as you have access to a 3D printer and/or laser-cutting machine."
Users find that Facebook’s new AI stickers can generate Elmo with a knife and Bing Is Generating Images of SpongeBob Doing 9/11. I think the Facebook-sticker thing to be kind of unnerving as these stickers are directly used in communication processes, while Bing generating edgy shitposter images to be somewhat okay, questions about intellectual property aside. I should be free to produce image synthesis of edgy imagery that is not that different from other edgy culture jamming popart, besides the fact that these image synthesis algos come from labor exploiting companies. But Facebook providing a tool in a messaging service which lets me produce stickers from SuperMario with an AK-47 is another story and i don’t like the idea of Facebook providing communication tools featuring SuperMario with an AK-47.
4chan Uses Bing to Flood the Internet With Racist Images, additional reporting from Ars Technica and Bellingcat. Edgelord shitposters use image synthesis to up the ante, which is related to the above, but interesting mostly for a network which is already famous for employing what they call "Weaponized Autism" -- a large network of "trolls" with lots of free time to narrowly focus on one "mission" to solve, mostly racist flooding attacks on activist networks --, and which is now maximizing their "productivity" with artificial intelligence. This will become even more critical when they start targeting journalists' real time reporting on news events with fake imagery, which they will do, i can promise that.
Generative AI Is the Newest Tool in the Dictator's Handbook, link to the report: Advances in Artificial Intelligence Are Amplifying a Crisis for Human Rights Online. "Over the past year, the new technology was utilized in at least 16 countries to sow doubt, smear opponents, or influence public debate."
Alexa says the 2020 election was stolen. What does it mean for 2024? It means that chatbots and AI-devices stay an unreliable source for information for the foreseeable time. I can't wait for the ultracustomized, ideologically aligned open source AI-models driving the smart speakers of the future. Like a customized Fox News, but more intimate and made to match every of your media dietal desires. Gonna be a blast.
Duke University launched a journal for Citical AI with interdisciplinary articles looking at the tech from a humanities perspective. All articles are free to read online.
‘Dream’ AI Girlfriend Randomly Turns Into Nude Jennifer Lopez, Has Four Legs. Synthetic porn services generate mutants and nonconsensual nudes, while the "platform is resolute in prioritizing ethics and safety". Sure buddy.
Texts from my ex. That's what you want, an AI trained on your Whatsapp-texts to train a virtual version of a nagging voice from your ex-partner. Privacy nightmares and juicy psychological effects guaranteed.
Important paper by Anthropic on Decomposing Language Models Into Understandable Components. One big roadblock to AI safety is that AI-systems are a so-called black box and we don't understand what AI is doing, even when we can look at individual neurons. Anthropic now identifies groups of neurons sharing a similar language feature, e.g. being legal language, hexcode, DNA sequences, nutrition information and so on. These features can be much easier identified and steered. In simple terms: Anthropic invented a way to identify "brain regions" in Language Models.
Max Tegmarks paper Language Models Represent Space and Time claims that AI models learn models of the world in training and identify "individual 'space neurons' and 'time neurons' that reliably encode spatial and temporal coordinates". This to me seems more related to Anthropics language features mentioned above than the models humans learn by interacting with the world and which are not language related.
Who moved my chips? Life in an AI entrepreneurs’ houseshare: The same grifters who brought us Influencer Houses now give us "Hacker Houses" where young devs are exploited to develop AGI or something. "When residents hang out in the kitchen, they talk about the latest ai research or get tips on how to raise capital." At least the dotcom bubble was built in garages. Party like it's 1999.
Another brain-to-image-model: DREAM: Visual Decoding from Reversing Human Visual System: Crucually, "for each subject, the model is retrained because the brain activity varies across subjects", meaning it’s gonna take a while until you can finetune a image-neuromodel on your brainwaves via earplugs and your smartphone. But we're getting there.
Adobe gave a preview of Project Stardust, their AI-enhanced photo editing tech, which combines all kinds of AI-techniques into one editing tool. Meanwhile, Google announces AI-powered photo-editing features for new Pixel phones.
Ryan Broderick writes about this: The AI Wants Us To Forget It’s Still Out There: "An AI for altering the records of our lives. Something you can quickly do and forget about. Which makes you wonder what happens when we can’t remember if we used the AI on our photos or not? What if a future upgrade tweaks our photos and we don’t even know?"
And Rob Horning in The egg of experience comments on Roderick: "Why should we treat images as 'records of our lives' rather than potentially something else, as whims or jokes or wish-fulfillments? Why would anyone assume that looking at an image is enough to make us remember precisely what it captured? Is AI 'horrifying' because it reminds us yet again that what we already take to be real is a brittle construct, that human memory is not computer memory, that the social relations necessary to sustain meanings across time can’t be automated?"
I'm not sure if AI is horrifying, but i did write about wishful mnemonics in Teenage Kicks in context of the TikTok-teenage filters and AI-anachronisms, and extrapolating from that, i think that a liquid personal archive we can edit on a whim is such a new cultural development we simply don't know how it's effects will look like. It's very possible that the ubiquity of photo editing will make us very aware that photos are not really representative of reality, but it may also be that it makes our archives irrelevant for documenting the world, when these documentations become fully editable without effort.
The Messenger signs a deal to “eliminate bias” and flag “clickbait” with AI. The AI by company seekr flags Breitbart, Daily Mail, and ZeroHedge as "very high in reliability", and Bloomberg, NYT, BBC, Guardian as "very low". Case closed.
Wanted: High Performers for the Last Job You’ll Ever Have imagines a company in which high performance employees agree to train mimetic AI-models on their every move and automatize themselves, while staying employed and cashing in. Sounds outlandish, but i think this is not a bad idea in a capitalism-eats-itself sense? And to go even further: I bet this will happen sooner than you think in stock trading with traders selling AI-models trained on their performance to the highest bidder. Think an AI-model trained on the investment patterns of Warren Buffett. I want to see this, and if only to see it all go up in flames.
Deepfakes of Chinese influencers are livestreaming 24/7: "all the human workers have to do is input basic information such as the name and price of the product being sold, proofread the generated script, and watch the digital influencer go live. A more advanced version of the technology can spot live comments and find matching answers in its database to answer in real time, so it looks as if the AI streamer is actively communicating with the audience. It can even adjust its marketing strategy based on the number of viewers, Sima says."
In the paper Art or Artifice? Large Language Models and the False Promise of Creativity AIs fail to pass a creativity test and I'm not impressed, because at one point in the future, they will pass this test too. But even then those models still are not creative, because creativity is a mental process which's output we may be able to simulate, but we can't synthesize the process itself (yet).
This is bad: 23andMe User Data Stolen in Targeted Attack on Ashkenazi Jews: "Attackers gathered the data by guessing the login credentials of a group of users and then scraping more people’s information from a feature known as DNA Relatives. Users opt into sharing their information through DNA Relatives for others to see. Hackers posted an initial data sample on the platform BreachForums earlier this week, claiming that it contained 1 million data points exclusively about Ashkenazi Jews. There also seem to be hundreds of thousands of users of Chinese descent impacted by the leak."
I've mentioned sometimes that i am against all screens in schools for lower grade and this article is a good summary of some of the reasons why i am against this: The case for continuing to write by hand.
Elle Cordovas conversation with a gaslighting algorithm. She's just so good.
Enable 3rd party cookies or use another browser
Lawyer Who Successfully Sued Alex Jones Says His Next Target Is Elon Musk and more from the man with no balls: Elon Musk’s X is a black hole of value: "X is worth around $8 billion" and clownman paid 44 billions for it. Lol.
Tibor Martini: "There is a viral thread on Twitter listing a lot of critical points about Elon Musk. For some reason, the tweet loses likes again and again in a mysterious way – so many that someone even programmed a deleted Twitter Like Tracker."
X rolls out new ad format that can't be reported, blocked, so basically, you now can pay Musk to run undisclosed ads. Mashable calls this 'ad format' 'weird', but it's not 'weird', it's unethical and illegal.
After removing headlines from link-posts to keep people hooked to the platform, Musk now also wants to Hide Reply, Retweet And Like Numbers. This is not necessarily bad, as removing those kind of engagement numbers has shown to improve mental health for Instagram users, when they experimented with this. I'm very sure Musk is not concerned about the mental health of the users of his platform, and the motivation for this is unclear.
However, with all these changes, here's Control Panel for Twitter, which does away with a lot of them, and more important: The Moral Case for No Longer Engaging With Elon Musk’s X: Accounts on TwiX post violent content and one of them recently made fun of posting videos of a murder. "In July, he posted that the site had paid him more than $16,000. Musk interacts with him often."
And now, this: Misinformation about Israel and Hamas is spreading on social media and Verified accounts spread fake news release about a Biden $8 billion aid package to Israel: "Many of the misleadingly labeled videos were shared by verified users on X, who are eligible for monetization of their content."
All that being said, here's my Mastodon: https://sigmoid.social/@rawx and here's my Bluesky: https://bsky.app/profile/rawx.bsky.social. I still use TwiX to follow the musings of a bunch of people, but my personal engagement there is dropping significantly.
Why we overshare on Instagram, TikTok, and other social media, a good piece on the tension between sharing intimate moments with people you care about and the different audiences on platforms and their expectations from your posts.
I was, and maybe still am, an oversharer, but i also come from a time when blogging was mainly about "say everything".
Oversharing back then was not just welcome, it was a foundational part of blogging. Just look at the history of the first blog Links.org and it's writer Justin Hall, who produced a mini-documentary in 2015: Overshare: the links.net story . In times of a myriad of public facing platforms and the dark forest web of encrypted messaging and chatservers, this subject has just grown in complexity and relevance.
Social media traffic to top news sites craters: Publishers are in deep shit after cratering to social media platforms for traffic. Now platforms are devaluing news and they face a threat by generative AI, which is bad. Ryan Broderick also writes about this: This Is What Replaces News On Facebook.
An Apocalyptic Meditation on Doomscrolling: "Doomscrolling is a way of 'weirding' crisis media in order to secretly, unconsciously transform it into a subgenre of horror".
Roughly ten years ago, when the psychological effects of social media modes became obvious and transformed all discourses into what we see today, it became clear to me that all that outrage and culture wars was not just terrible, but engaging with it and reading an infinite scroll of terrible stuff literally turns you on. I also always suspected a connection to narratives of the horror genre, but i never came around to write it down. This is a piece about that connection.
The article also links to a highly interesting paper about how Bad News Has Wings: Dread Risk Mediates Social Amplification in Risk Communication. In experiments, the researchers have shown how bad news gets more and more loaded with negative statements the longer it travels to the network, no matter if you reintroduced the original message in the chain of communication. I think this is related to attention economic incentives, which make us one up each other constantly and at the end it's all about doom and death. This is the inner workings of memetic outrage.
Meta in Myanmar, Part III: The Inside View. Part 3 of Erin Kissane's detailed account of Facebooks engagement in Myanmar and how it supported genocide. I linked to Part I: The Setup and Part II: The Crisis in the last roundup.
NASA satellites reveal restoration power of beavers. NASA Satellite imagery reveals that rewildered Beavers regreen whole landscapes. Beavers are cool.
The Octopus Abattoir Must be Stopped: "Spanish seafood company Nueva Pescanova (...) wants to build the world’s first factory farm for the common octopus (or Octopus vulgaris) in the port of Las Palmas, in the Canary Islands. There, the cephalopods would be slaughtered en masse for the world’s dinner tables, mainly by being thrown into 'baths of ice flakes and water that would gradually drop to -3C'."
Octopusses are famous for their rich cognitive experience and intelligence, remembering people and holding a grudge, fostering and executing complex escape plans and building cities. I'm pretty confident they are conscious beings, and adding them to the industrial mass slaughter we already do every day to complex life forms is a bad idea. Our kids will hate us for this.
Scientists Found Microplastics Deep Inside a Cave Closed to the Public for Decades: "A Missouri cave that virtually nobody has visited since 1993 is contaminated by high levels of plastic pollution, scientists found." This gives you an impression about how high plastic is concentrated in ocean water.
Fortnite Influencers Push Shell’s Propaganda on Kids: Instead of concentrating on reducing fossil fuel emissions, Shell paid "at least six Twitch streamers and three influencers with a combined tens of millions of followers were tapped to push an oil-focused Fortnite map."
Scientists stunned by planet’s record September heat: "Global temperatures soared to a new record in September by a huge margin, stunning scientists and leading one to describe it as 'absolutely gobsmackingly bananas'."
Stunning Footage of Neurons Forming Inside a Chick Embryo Wins Nikon’s Small World in Motion: "Featuring colorful beams of light streaming across the frame, the video magnifies axons, which connect neurons, as they 'traverse the nervous system before eventually forming synapses'."
We got the beat: How we perceive rhythm involves neurological processes that control movement: "Our sense of rhythm developed (at least in part) as an outgrowth of monitoring and anticipating our own footsteps as we walk or run."
AI startup Prophetic aims to build headset that controls your dreams, which is a slightly misleading headline: They are trying to build hardware which activates a state of lucid dreaming, where you can consciously control your dream. I'm sure their monetization strategy will not include advertising in any way, shape or form.
‘Lord of the Flies’ Remake Producer Calls Luca Guadagnino's Film Scary & Fresh. I didn't like Guadagninos Bones and all but i loved his remake of Argentos Suspiria, so I'm looking forward to this.
Trailers worth watching: John Woos Silent Night, Scavengers Reign, Night Swim, Hurricane Season, Thanksgiving, Wingwomen
Can't wait for the superweird, Soderbergh-produced scifi flick Divinity, and here's a Kool Keith Commercial for its soundtrack.
David Lynch Directing Style Explained — How Does Lynch Make Us Dream?
Famous Roland TR-808 Drum Patterns 1982-1985
The Language of Lava Lamps. I loved my Lava Lamp when i had one and i wish i was "the owner of fifty lava lamps" who lived through a "lava period" like Nora Claire Miller writing in The Paris Review.
Bandcamp was recently sold to a music marketing company and things are going as expected: Bandcamp union describes workers locked out of systems, coming layoffs. I signed the petition to Keep Bandcamp Great: Tell Songtradr to recognize and negotiate with Bandcamp United immediately, and you should, too.