Shamelessly, this will be a shorter update as many artists are either in steady-state grinding away at projects like Jason (this is a good thing), have been traveling to ride out the dog days of summer, or perhaps both. Personally, I am already looking forward to the calm focus that fall brings. Thomas Merton put it best:

‍

"October is a fine and dangerous season in America. It is dry and cool, and full of the memory of last year's fallen leaves. The days are full of a distilled autumn-like sorrow, a sense of the last things passing by. It is a fine time for going back to school, and for starting afresh with all the best ideas you ever had, and letting them think they were original."

‍

But, alas, we aren’t quite there yet. A couple more hots and we’ll be through.

‍

AiR Artists Round Table

‍

Today we are ecstatic with anticipation for our first Current Artist Round Table live stream! Since we are a 100% remote residency, this will be the first time that all of the artists will be in one “room” together, exchanging ideas about AI, contemporary art, and what it’s been like to be an AiR Resident. This one should be excellent! (he said with gusto, likely jinxing it). While unsure of the exact timing of this newsletter send, be sure to either tune in at 6pm EST (twitch.tv/civitai) or look out for the recording which should be posted to Youtube on Monday. 

‍

Jason Bailer Losh

‍

Jason has been both traveling and grinding away at his new oil painting. His focus now has shifted from the conceptual work of planning and designing his new work to simply knocking it out. This takes time, and we are here to give it to him. In case you missed it, I did an instagram reel on his work at the beginning of this week. I’m going to be doing more of these, along with simpler posts of images and of course artists speaking about their own work.

‍

‍

Ira Greenberg

‍

Those of you who saw Ira’s lecture last month will know how deeply entrenched he has been in the emergence of creative coding over the past few decades, authoring the first book on Processing for artists, which has gone through several editions. Through this residency he has started exploring ComfyUI, which has opened up an entire new world of possibility. 

‍

In teaching himself the intricacies of animatediff he has begun applying it to reanimate still images from his archive. The Reticula series was made several years ago using primarily Midjourney, and visualizes highly detailed impossible machines that up until now were frozen in place. These works provided the perfect substrate for his new explorations in motion with his new AI workflow breathing life into these delicate monsters, sending gears spinning and wheels whirling. He’s been sharing these on instagram lately:

‍

‍

‍

‍

He’s been using SD1.5 in his animatediff workflow, and will soon be exploring up-resing these short videos to recapture the full detail of the original works.

‍

Keion Kopper

‍

Keion has had a few hangups trying to train his new model, and isn’t quite satisfied with the LORAs that he’s been getting back. We discussed his technical issues and made a few suggestions for future models including limiting the amount of training images and also focusing the style a bit more.

‍

His goal is to produce one 4’x4’ fossil painting for exhibit at our Art Basel event in December, in partnership with a16z. He also brought up the idea of creating a physical “DEX” out of a raspberry pi that could hold all of his AI permutations. I think this could be a great way to expand the lore around the universe of his fossil pantings, so am keen to see any experiments here. However, the model must be perfected before the DEX can be built, and the painting produced, so he has is work cut out for him. If the models he trains on his own next week still arent up to his standards, we will help connect him with an expert to produce the desired results!

‍

Here is his progress map from his figma board:

‍

‍

Ines Kivimaki

‍

Ines has been playing more directly with language and translations, with her latest video edit featuring moments where her AI deepfake voiceover lapses into Russian, then seeming to self correct back into english. I enjoy imagining these clips as extracts from the interior of an AI vision brain, which operates in a post-linguistic environment. This brain would seamlessly switch codes whenever prompted to do so and therefore “think” out its results in a hybrid panoply of different human languages, all of which are superficial to its core processing. In this emergent reality language is therefore present only as a vestigial appendage catering to the cultures that bore its creation. 

Click this link to watch the clip if it doesn't play below: Figma.

‍

‍

Elon Musk recently spoke with Lex Fridman about his progress with Neuralink. In that conversation he discussed a human’s typical “bitrate” of communication, equating each word to one bit and dividing it by the number of seconds in a day, 86,400. Studies show that humans typically use 15-20,000 words per day, so even if you reduce the denominator by half to account for only waking hours this means that on average the effective bitrate of human communication is sub 1 bit/sec. Since we know computers are capable of transferring and processing data at gigabits per second, Musk makes the assumption that human language is therefore an incredibly inefficient mode of communication by comparison, one that will be vastly superseded by Neuralink's ability to bypass language in its transmission of data. While he may be right when it comes to transferring hard data sets, encounters with linguistic uncertainty like the moments depicted in Inés work reveal the ineffable complexity of human communication. 

‍

I just don’t buy Musk’s basic framework where one word = one bit. This core assumption to me is massively wrong and destabilizes any other assumptions that could come from it. Context aside, a single word has an expansive network of concepts and histories embedded within it. And within a given context (of which there are infinite variations) one word can also have thousands if not millions of meanings and effects on the receiving human. This doesn’t even factor in the reflexive “proprioceptive” effect of speaking a word to another human and considering the act, therefore producing a change of state within the speaker. Regardless of the context one computer bit will always be just one computer bit, whereas words may actually represent millions of bits of compressed data.

‍

Anne Horel

‍

Check out our latest reel describing Anne’s work!

‍

‍

We are currently working on developing a physical item store to launch her AR enabled T-shirts! These will be available in the shop section of our main site. As mentioned in the video, when you scan the QR code you will be able to actually wear the mask on the T-shirt in an AR environment!

‍

More on her work soon, and be sure to catch the replay of our artist round table, which should be available on YouTube by Monday!

‍