This has been making the rounds in the AI/Art infospace, that a group of testers of OpenAI’s upcoming Sora video gen model broke ranks and leaked access to the model publicly when they weren’t supposed to, and posted this open letter on Huggingface in support of the effort (signable version here).
ARTISTS ARE NOT YOUR UNPAID R&D ☠️ we are not your: free bug testers, PR puppets, training data, validation tokens ☠️
Hundreds of artists provide unpaid labor through bug testing, feedback and experimental work for the program for a $150B valued company. While hundreds contribute for free, a select few will be chosen through a competition to have their Sora-created films screened — offering minimal compensation which pales in comparison to the substantial PR and marketing value OpenAI receives.
▌║█║▌║█║▌║ DENORMALIZE BILLION DOLLAR BRANDS EXPLOITING ARTISTS FOR UNPAID R&D AND PR ║▌║█║▌║█║▌
Furthermore, every output needs to be approved by the OpenAI team before sharing. This early access program appears to be less about creative expression and critique, and more about PR and advertisement.
I guess I feel mildly sympathetic about “sticking it to the man” in a general way, but maybe I feel mostly confused on the specifics here. Not to sound too cynical, but were the people behind this letter not aware of what the nature of the agreement must have been when they presumably signed it?
None of what’s being described by them seems all that new or different for software companies, which routinely leverage unpaid labor out of their user base (ahem, social media anyone?), and *of course* inviting artists for early access is part of PR & advertising. That’s just a given for me on something like this. I can understand not wanting to be a part of that for sure, but I just end up wondering: why sign up then in the first place?
All that said, let’s find ways to pay artists for helping build these technologies, and let’s actually actively listen to their feedback (instead of, for example, banning them), because artists bring a whole other holistic and humanist set of skills to AI development that are so well-represented by the overwhelmingly technical, academic, PhD, research, math and engineering-types who currently populate the halls of AI power.
This subject keeps coming up. I looked at it and I guess I don’t really care that much. 8000 books in a year is probably a lot less than whatever Amazon prints for print on demand customers. $5000 is too much to spend on pretty much anything for seat-of-the-pants small publishing. The AI angle doesn’t seem all that strong here to me personally, as I keep thinking that running any of my weirdo AI-hybrid books through a purely AI editor-ing system will absolutely destroy the novelty and nuance of the way I pieced everything together in my very human “bursting at the seams” style of editing.
I’ve been experimenting with the new-ish Photoshop Remove tool. And no, I didn’t completely quit Adobe like I hoped/promised, but I did deeply downgrade from the all apps pla to the Photography 20GB plan, and bought the Affinity suite of 3 apps at half price on Black Friday in order to fill in any other potential gaps of functionality as I work on other projects.
For this I selected most of the original painting in the middle with the Remove tool, waited a few seconds and it spit that out, which I actually think is a very interesting piece in its own right, but also when paired like this with the original source material, to explore these sort of parallel latent versions of a given work, or even a historical moment. Reality isn’t going to be the same anymore.
Here’s a historical riff, removing the flag and pole from the famous Iwo Jima photo:
Outside comic or ironic uses though, the possibilities here get quite dark quite quickly. Consider this flung together xample from the famous Tianamen Square standoff moment – Tankman? What Tankman?
Similarly dark, but in a different direction: shooter, what shooter?
At the same time, I could see this tool and its deep hyperreality effects as being a potent tool for iconoclastic political critique in a world gone completely mad… Don’t like current political situation? Here’s a filter you can use to live in a parallel reality where that person/party/place/thing never even reaches your digital eyes or ears…
Plus, I think it would be pointless to try to regulate this. Where would you even start? To ask for that would be to invite tech companies to implement ever greater filtering and restrictions, and put more and more content decisions into the hands of AIs tied to corporate profit-engines. I mean, that’s going to happen anyway, but at least let’s not go begging for it. At least not until we’ve had a chance to kick the tires a bit before it gets nerfed.
Here is a satirical gen AI image set I made for a fictional event called “CEO AID” in which the top health insurance executives from the US recently banded together to perform an amazing cover version of 1985’s classic song, “We Are The World” in order to raise funds at Christmas time to pay for their private security details.
These images were made in Recraft, which is one of the front-runner image gen models right now, imo.
I finished this very version 1 initial prototype of a robotic drawing arm made from willow branches from my garden, and using syringe hydraulic actuators (no electronics or motors). Terrible picture again, sorry. One day, I’ll do some better quality photos of all these things.
This is one style of a SCARA arm used in robotics, where the S and C stand for selective and compliance, in that it can move horizontally in xy axes, but not vertically in z. In this case, the movement is provided via a human operator pulling and pushing the water-filled syringes and tubing to actuate the joints of each arm segment. (Here’s some related concept art I did for this using AI before/during the actual build.)
I spoke about this and some of my other low tech human drawing bots (or “biobots“) in this episode of the Silicon Synapse show. I’m trying to find that mystical crack in the world between human-whatever and AI-whatever where something stops being a human-ish thing, and becomes a “mere mechanical process” such that it becomes according to the bureaucratic gods who deem it their role to interpret such omens… UNCOPYRIGHTABLE!!!
I, of course, have a lot more to say on these topics, but am trying to document at least minimally some of the highways and byways that I have traversed this past year while investigating these issues creatively. I’ll also do a demo drawing with this machine in the coming days and post that here.
Unfortunately, this initial design is pretty limited in its range of motion, which is okay because I was more just trying to understand how the mechanisms work, and now I have something adequate to experiment with which I think will be able to tell me through experimentation proper dimensions and alignment for an updated version with a greater range of motion.
Since Embodied marketed Moxie as a companion and development toy for children, there’s concern about kids potentially suffering an emotional toll after the robot abruptly becomes inoperable. Embodied has responded by promising to provide a guide for telling children about Moxie’s demise.
In my on-going survey of simple mechanical devices, including toys, one repeating form that comes up often is that of the jumping jack toy, where you can pull on a string or stick on the bottom of the toy, and it articulates usually the legs and arms of the toy figure to move up and down.
I don’t pretend to be an expert in any of this, which is precisely why I am experimenting with making my own working versions of these simple forms and mechanisms: so that I can learn first-hand how they function, and gain experience combining different components into novel configurations. I sort of consider things like this to be very basic precursors to robots…
Here are two Humpty Dumpty themed jumping jacks that I made with my kid, where the limbs are attached via brass fasteners, which I had completely forgotten about as even existing, and turns out are a lot of fun to play with.
Front:
Back:
This site has some nice clear diagrams of how these mechanisms ought to be arranged. Mine is a little messy with the twine and hot glue, so these might be easier to understand.
A reader wrote in recently with the following question, which they gave me permission to post a reply to here, as I thought other readers might have similar questions. I also think it could be funny to have an “Ask An AI Guy” advice column. If anyone else has AI-related questions they want to ask and have me answer publicly, drop me an email at the contact form here.
The question:
I came across your article in Newsweek about using AI to write books. I found it incredibly interesting how open you were about using it as your creative partner in a sense. I’m currently enjoying using chatgpt to help form my ideas into slightly more coherent plot points. And it gives amazing feedback on my writing, possibly better than what a technical editor could give. But I can’t help feeling like a fraud. There’s a nagging feeling like, shouldn’t I be able to do this on my own? Before AI, didn’t writers have nothing but themselves to brainstorm, write, and rewrite? Have you faced this issue? Do you have any thoughts on this? Thanks so much for any insights!
I answered part of this question I think in the Register interview here, but will recap that part briefly: working with AI has made me a better writer. Simple as that. It’s made me more objective about what’s a decent piece of writing, and does a given piece answer the particular need its intended to fill, and does my argumentation take the reader logically from A to B to satisfy that need?
“Before AI, didn’t writers have nothing but themselves…”
First, I would say that’s not true on its own: writers have always had other writers, editors, and other readers. I think the idea that writing is this heroic totally solitary activity is a bit incomplete, as it has always had a very social side.
Second, whatever happened ‘before AI’, we’re no longer living in that mythical before-time, just like we don’t live in a time before computers, etc. While doing things without it is entirely valuable and worth mastering for mastery’s sake, we now live *with* AI. One way to live in a time with AI is certainly to totally reject it. Another is to partially reject it for certain things, and use it for others. I think it’s just about finding what that fit is for you, the writer, and for your audience. For example, I wouldn’t want AI to take over the “fun” parts of writing – but your mileage may vary as to what is actually fun for you, and what can benefit from outside eyes – even if those are digital eyes.
Despite hundreds (thousands, really) of people calling me a fraud for using AI to tell new kinds of stories, it’s a feeling that I have never once myself actually shared. Sorry, I just don’t feel like a fraud. I don’t feel sorry, or cowed or threatened even when thousands of people tell me I’m wrong and bad. Maybe that makes me an asshole, but I don’t think so. It means really I’m able to listen to the intuitive voice and the creative light that guides invisibly my work, and much of my life, and unswervingly devote myself to that, even if others can’t see it or don’t quite get it. That light might tell some people to restrict how they use certain technologies, but mine tells me to find out by doing, to dive in and see, and talk about it with others. Following that light has not served me wrong so far – and if anything, it has made my life better and my work richer. If I stopped following that, that’s where I think I would get into trouble and start feeling like a fraud, because I will have given up on what’s actually actively true for me.
I guess my point is: if the journey is authentic, then so will be the end-product.
As unexplained aerial phenomena (UAP) form a major part of the inter-locking narratives of the AI Lore books, I thought I’d post a round-up of the books that are the most relevant for the events unfolding in New Jersey, and evidently around the globe IRL. In chronological order according to publication date (maybe I’m forgetting some – highly possible):
Having to type out “Willow SCARA Robot” a number of times for this post made my brain finally collapse those together into a nice name for this machine, “Will Scarlet” (will-scarlet). It seems to fit somehow as a moniker for this particular configutation of bits and bobs that becomes, through effort, a machine for creating “art.” A sort of nascent physicalized proto-AI without anything “artificial” really about it. It’s all ‘real’ – and all really controlled by me, using two syringes attached to tubing, one held in each hand.
Here are some process photos of conducting will-scarlet to make a piece of “art” using acrylic paint markers on canvas. But more important than creating “Art” or any semblance of it, simply trying to mechanically, and even scientifically observe what are the capabilities and limitations of this particular device. That was the goal of this initial exploratory session, and its documentation. Some highlights, but I’ll leave most of it on the Imgur linked above. There are also some raw unedited short videos you can watch there too (but which always refuse to embed in my blog posts).
Here is will-scarlet drawing the types of arcs and bows that his range of motion will allow him. I had really no idea what to expect on building this first prototype, so just winged my placement of different components, their relative lengths, etc.
And my components are all as primitive as possible. The arm segments and their pivots are all from willow branches that I grow here locally for basketry. So this robot is kind of an industrial byproduct of the basketmaking trade, in a way.
I’ve written about this in prior posts, but part of my exploration here is to make “biobots” which are as “natural” as possible – whatever that means – or that intentionally incorporate organic (and eventually living) elements. But beyond that, or alongside it on the third rail, I also want to figure out what kinds of “thinking machines” could survive a Butlerian jihad, a la Dune universe. Or like a robotically augmented art studio that could survive some kind of EMP/SHTF scenario. Not so much because I’m worried about those things beyond a kind of thematic motif, but because it’s a ghost world/Uncanny Valley of art that doesn’t seem too well-explored. I found plenty of people coming in from the opposite direction of turning to Arduino code and electronics and motors to produce art-making robots. But I didn’t find anything really like I’ve been doing: strip out the motors, pull out the code and electronics to the greatest extent possible, and put back the human at the helm. A more fluidly organic humanist cyborg augmented. Something that doesn’t rely on someone else’s cloud, API, and infrastructure: art robots for the here and now.
This is as far as I was able to get in one session. I don’t find it super-beautiful yet as a finished piece (even if that’s not expressly the point, I’d like to get it there), but was a good experiment. The plan is to continue layering on top. Hopefully after I can rig up some better controllers, as having one in each hand is rather awkward to smoothly control.
In fact, in the image above, you can see two sections of black marker in the topmost layer. At right is me using it with one in each hand to draw: angular, constrained by having to work the plungers up and down each using only the hand already holding they syringe body. Then my kid came home, and we each controlled one syringe kind of … independently together. And those line tracks are on the left side, and appear as much more fluid. So that’s very interesting.
In the current setup, since the range of motion is so constrained (make a note of this), I end up having to flip the canvas all around on the backing board in order to reach new areas. Not the worst experience in the world, as constraints like this tend to force you to discover new creative ways to get around them… But still, it will be nice to do a bigger better version of will-scarlet – possibly/obviously to be named robin-hood – where I’ve mastered a bit more the placement and sizing of components, allowing for a free range of motion, and a better manual method for controlling it.
ChatGPT suggests turning this into a four-bar linkage, which isn’t a bad idea, but will look less like a “cool wooden arm.” I’ll probably try that anyway. I also want to do a version with the arm in this configuration – minus the motors, ofc.
Also, going back to range of motion, I noticed that as the machine worked for a while, its range changed, and in one case became rather limited. Some of the tubing had popped off, and we’d lost fluid (water) in the hydraulics. It really reduced how the arm segment most affected could move. Once I got that sorted and refilled, it improved again.
Plus there was an issue where, toward the end of one of its poles of available movement, the paint marker would lift off the paper. Looking closer at the machine, it was obvious that it was caused by the irregularity of the willow branches of which the thing is composed. I’m not using machined metal tubing and smooth turning bearings, etc. in all this. I’m using pieces of my garden. So, since that’s an intentional choice, I see those irregularities as a good thing, as “proof of life” rather than as evidence of some failure that needs to be eradcated by ever more regular objects. This is what embodiment looks like. This is how we can tell it’s really “Art” with a capital A – not strictly the painted canvases that the machine outputs, but the entire socio-technical assemblage from which it is born, and through which it depicts in a very clear but also roundabout way, some essential thing about my life in this time and pleace.
