ChatGPT and the Cotton Gin · 8:50am Apr 22nd, 2023
Whole thing up on my website, first post in ages
I was visiting family this month and they asked for a catch-up on all this "AI" stuff going on. This is my best attempt at making a lot of what's happening with ChatGPT, Midjourney and now Adobe Firefly accessible to the most people possible - I tend to see this information written either by tech people who are overly optimistic about it, or artists who are directly threatened by it: In my opinion both sides are mostly right, but end up talking past each other since the tech people tend to be roasted for being bad artists, and the artists for not understanding the tech. This is my attempt to synthesize the optimism for what's good about these tools and the pessimism about why they're so threatening. I'll post the introduction here as a sample.
Appreciate comments here, though, since I never figured getting that to work over there. Ah, well.
Controversial statement: Slavery’s bad.
Here’s the thing though; It’s not just morally and ethically bad, in a lot of cases it’s also just bad economics.
I want to be careful saying that because a lot of orthodox economists finish there. There is a particular kind of brain that hears that sentence and thinks; This means market forces will cause slavery to self-abolish! Keep the course of automation and development and slavers will be out-competed!
This is why we’ve got to talk about the cotton gin.
See, the cotton gin on its own is a good technology. Separating cotton without one is an immensely time and labour consuming practice. The cotton gin is a relatively simple machine that saves people having to do all that work. In a kind and just world that would let the same amount of people produce more cotton while having more free time. In an economist’s world that lets fewer people make the same amount of cotton and the people who are put out of work can retrain into ‘more productive’ industries. The same amount of work results in more stuff for everybody.
But the cotton gin was invented in a world of slavery. Eli Whitney’s cotton gin came out in 1793, and so slavery exploded in the American South.
What the hell happened?
Well, until that point, slavery was increasingly less profitable than freeholder labour. Using technical language, free labour is motivated to increase its own productivity. In human terms; people who work for themselves will work better, harder or smarter because they’ll get paid more. Slaves will only work as hard as you torture them.
Which also means you have to hire people whose sole job is making the slaves work. People to whip them, supervise them, catch them when they run off. Because slaves don’t want to be slaves. Let’s call these scumbags your ‘human resources’ team. They don’t do any farming on their own but, without them, no farming gets done by your other workers.
So it’s pretty easy to see how freehold labour ends up more productive, right? Everyone tries harder for their own reason and there isn’t this massive waste in having a human resources team. There’s only a narrow margin where the work is mindless, can’t be optimized further, and sufficiently miserable to do that slaves can be made to work harder than freeholders.
If the cotton gin hadn’t been invented for a few more years then slavery abolition might have made it through. The money interests would start to dry up or be outcompeted, there was always heavy opposition to it - and after that happens, then the cotton gin can safely be introduced to everyone’s benefit.
But because it came out the moment that it did it re-established slave owners as the dominant economic model - so slave owners became richer and outcompeted freeholders, and that meant more freeholders had to turn to slavery to compete or be run out of business. And the rest is “history”, a history the United States has not recovered from. I’m going to link James Baldwin debating William Buckley about it here, Baldwin’s speech is beautiful.
This is basically my position on ChatGPT, Midjourney and soon Adobe Firefly. They are the cotton gin of 2023.
If those applications are the metaphysical cotton gin and artists are the metaphorical free laborers, who are the metaphorical slaves?
Or am I missing the point?
Very interesting read.
Thank you for posting.
5724148
It's not meant to track completely 1:1 in that sense. The point is that a technology can be harmful because of the conditions of the world it's brought into. The cotton gin doesn't inherently have anything to do with slavery, but its effects on slavery are still half its wikipedia page.
Likewise this software doing white collar grunt work wouldn't be a problem if people didn't need to hold jobs to survive.
5724150
Ah, I did miss the point.
I think I understand now, particularly in light of the distinction between a kind and just world and an economist's world.
AI artistry wouldn't be so big a deal in one world, but causes consternation in this one.
Thank you.
Nicely written article and good summation. I hadn't realized that the artist who submitted AI art also did AI art that was actually, well, artful.
Also, I feel like it might be a bit unclear to others that there's more to the write up at the link, and that what's quoted isn't the whole thing.
5724197
I do say "I'll post the introduction here as a sample" is that not clear enough?
One mistake I think this type of discourse around AI art bots commonly makes is over-emphasis on the result. Ignoring for now the entire problem of art theft in training these bots... (Note: when I use the term "art", I'm referring to all of the arts -- plastic, performing, and literary, "fine" or functional).
Creativity is a process, a journey, the end result is almost inconsequential to that. There's a very old saying in the art world that no work of art is ever truly finished, merely abandoned.
The big problem is, capitalist commodification of art has dramatically changed how we understand and value art, reducing it from a form of human cultural communication and turn it into just an easily replaceable and marketable product, like a sofa, a car, a cheeseburger. Something to be consumed, rather than something to participate in.
What these bots do is fully remove the human aspect, and focus entirely on the commodified product. They've eliminated human expression entirely and created, not art, but a Debordian simulacrum of art. Using these AI bots does not make one an artist, any more than going to a corporate chain restaurant and ordering a pizza with custom toppings makes one a chef.
The other problem is that, because of the commodification of art, there is far too much emphasis on being good at something. That one's time spent on art is only valuable if one has a marketable product at the end, or at least something that will receive praise from others for its technical proficiency. That's a very bad view of art, and one that also helps to stifle creativity; making people afraid of making "bad" art.
But making bad art is not only acceptable, it's absolutely vital to human culture. Art should a journey of growth and self-discovery; and making bad art is part of that process. Whether one eventually becomes proficient enough at one's chosen form of creativity should matter less than the joy one takes in self-expression and developing one's skills.
The idea that art must be "good" in order to be valid is a straitjacket, it's a prison. That's why I've always told people here that when they start writing, they're going to suck, because everyone does when they first start out. That's not a disparagement, that's freedom. Allowing yourself to be bad at something is immensely liberating, it allows you to explore things and experiment with things that you otherwise would be too scared of the consequences to pursue.
I have a similar problem with physical capabilities. I cannot draw to save my life, due to a neurological deficit with proprioception and fine motor skills. So I picked up photography as a medium; but not exclusively. I still paint, weirdly and badly, but I enjoy it. And I write, you can see the consequences of that pursuit posted on this very site. And I make music, if one can call it that. Do I do any of these things proficiently enough to be marketable? No. Do I enjoy them and value them as part of my human development and self-discovery? Absolutely.
AI "art: shortcuts and evades all of that, and the people who create and maintain those bots are only interested in making money. That's why the loudest and most fanatical proponents of AI art bots are the same "Web3.0" techbro types who were hyping NFTs a few years ago.
this is such a cool opinion, thank you so much! i love it
I have seen a lot of the AI discourse for the past year or two, and have strong opinions on the subject. But only what I'm sure everyone has heard before. I'm with the artists who warn that this will decimate the field, and find the biggest cheerleaders of the technology to be just awful people in general. I find talk about prompt-whispering and iteration... unpersuasive. Especially coming from said biggest cheerleaders, who hold art and skill in contempt yet want to be respected for putting a few words into a prompt in a good way
And while I don't want to tell anyone not to enjoy those pictures you posted, they do have the same old failings of AI art: if you know what details to look out for the illusion quickly falls apart and loses all appeal (look at the hands. Those are not human hands.)
Could I just say how much I appreciate you including that note in the main post about how many kinds of inherent and genetic issues can prevent or hinder people from pursuing the art they'd like to? As it happens, I'm also dyspraxic and realistically was never going to learn to draw. And not only do I wish I could in general, but given that one of my favourite mediums to write for is comic books, having the ability to draw what I envisioned even vaguely competently would have been extremely useful. And hearing some people talk about how "anyone can learn if they just try hard enough" is... not helpful.
So thank you for that note, minor as it might seem to many.In fact, it's precisely because it might seem that way that it needs to be said.
5724207
Ah, I actually missed that you wrote that.
At least as far as fandom artists are concerned, I think the Tumblr user's story about grant proposals may be the most illustrative.
My personal reaction to the story was puzzlement at her existential dread—for the grant proposals, conference abstracts, and grad school applications she calls out, I've always taken an extremely cynical view. I agree with her conclusion that they were never important and primarily consisted of stuffing the correct keywords in the correct order. Where we disagree is in our reaction: she had dread, and I had the self-satisfaction of having been proven right.
Circling back to AI-generated pony art, the "generated by Cloppermania" tag on Derpibooru is probably the best AI art tag that satisfies the following conditions:
What that feed is full of is mostly-passable generic clop. The vote scores confirm what I already suspected about Derpibooru voters: they primarily make their voting decisions based on the image thumbnail in the results grid from a search, whether it is for a specific tag or for my:watched. I've found multiple super-hot AI images in the preview box that then lost all erotic potential when viewed in full-size. Moments of "wings don't work that way" and "Where's her fifth hoof?"
Actually, that reminds me of a podcast episode investigating the first successful passage of the Turing Test. As it turns out, humans are much less discerning when horny or pissed off. A computer that algorithmically insulted its user successfully convinced the user that he was chatting with a total jackass from the mathematics department over teletype before I was born.
Back on topic, I've found that emerging artists are most vehemently opposed to AI art. Established artists either view AI as a tool or as a curiosity but are mostly too busy making art to write screeds about consent violations from the training process. It's artists who are on the way (or even would one day aspire to) to having their Patreon be their day job who fear it most, not those who already make their living through commissions.
Of your original words, I found
to be the key.
AI allows people to realize they're polishing a turd instead of chasing an unworthy goal their whole creative life, constantly seeking the technical skill improvement that will make it work. I've watched others (and have myself) fixate on a stupid idea and never realize just how stupid it is because "my prose fucking sucks, I'll try again once I'm better" instead of moving on to a better idea. Often that better core idea will manifest itself in easier improvement in technical execution. In the realm of indie games, AI could allow for a solo developer to make their ultra-niche game that otherwise would not get made or would be watered down until its expected sales could justify hiring a human artist (or the expected sales would be enticing enough for an artist collaborator to accept a 50/50 profit split).
[Speaking of an unrelated use of AI, Grammarly is performing unusually poorly tonight. Yes, I know I'm typically wordy (blame by academic upbringing for padding essays), but I can't trust its conciseness edits because it's also suggesting downright incorrect grammar fixes]
A good bit of pondering, there.
I think the most important part to all this is the realization that the genie is already out of the bottle. We can't put it back. Crying about art theft or that this is going to fundamentally change how, or even if, people do certain jobs is pointless. I absolutely understand how people can be upset by this, but being upset doesn't solve the problem, and no amount of trying to stop the change is going to work. The change is already here. Change with it or become an expensive redundancy.
There are, honestly, much more frightening applications to all this. Technology is reaching a point where computer generated content is indistinguishable from human generated content. This is going to fundamentally change the way we interact with people online. I've already started running into bots posting places just to sway the narrative of the conversations. They only really became apparent after four or five back-and-forth replies with them (at which point the conversation is already swayed). They are only going to get better. We are looking at that, plus the advent of bots with voice and video ability. Again, this genie is already out. It isn't "maybe going to happen" it's "already happening".
The second most important part to all this is also barely touched on. Unless you want the total economic collapse of the industrialized world, we ABSOLUTELY NEED to start both universal basic income and free higher education. Every entry level job on the planet is on track to be replaced by automation. Not just your interns and artists, but also vast swaths of the service industry. It is simply going to be cheaper to build a burger flipping robot than it is to pay a burger flipper. Without UBI, the vast majority of the population is going to find themselves paywalled out of existence within a century.
5724222
5724265
I want to counter with something. The vast majority of people paying for art don't really care about the process. They care about the result. "Does the picture look good?" "Is this song pleasing to my ears?" etc. Algorithmic generation is not threatening the creation of art, it is threatening the marketing of art. Because above all other concerns is "Can I afford it?". Railing against this is akin to the Luddites smashing up textile mills because their profession had been replaced by automation. Textile weaving is absolutely an art. But it is no longer a mass marketable art. The same thing is almost certainly going to happen to today's artists. You can still find people practicing traditional textile manufacturing techniques. But it is all as a hobby or very niche products sold at a very premium price.
This gets back to why UBI is so vitally important. Batwing is correct that all art is valid, no matter the quality. People aren't going to stop wanting to make art, but they ARE going to stop being able to make a living from it. Validity is not inherently marketable and we are quickly reaching a point where there is nothing the average person can do that automation can't do "good enough" at a fraction of the cost.
5724150
On a slightly humorous note, the vast majority of things like cover letters and grant proposals are already reviewed by bots. It's why they so quickly became soulless and formulaic. Anything with personality was automatically tossed before reaching human eyes. Automating their creation just means that we have bots writing letters for bots to read. Also why adding hidden text with all the buzz words to the end of your document is so useful.
5724222
On NFTs, I had a thought. As demonstrated by the last few years, trying to "own" digital art by having some kind of receipt is a fools errand. However, they would work fantastically as a proof of origination. The art can be copied, changed and passed around but the NFT can not. It's on a publicly verifiable and secure digital ledger kept on potentially millions of different computers. The basic idea is that only verified "human created" art gets an NFT. Then, no matter where that art goes, it has a link to its creator.
Basically, they way NFTs have been used is completely ass backwards.
And, yes, this conveniently ignores all the issues with cryptocurrencies.
5724265
You bring up something else that I didn't mention, the devaluing of expertise. Anyone who works in a creative field already knows about the devaluing of art and cultural labour, especially when they've been insulted for not working for free. Art is what allows us to maintain and communicate culture over distances and generations; but the artists themselves have so often been relegated to the fringes of society unless they're producing large amounts of tangible monetary wealth.
Worse, however, is the devaluing of expertise. While that tends to be most blatant in art; we can definitely see it far more widespread in recent decades, and it forms the core of anti-science and anti-academic thinking. When taken to extremes, it's foundational to truly harmful worldviews like the anti-vax and "gender-critical" movements.
5724339
That's not a counter to what I said, that is what I said -- most people have been convinced that the only value in art is as a consumable commodity. They have been alienated from their own creative impulses by a society that devalues and outright denigrates any artistic pursuit that is not producing wealth for a small class of elites.
What is needed, alongside UBI (which I'd argue doesn't go far enough, we need outright socialism), is for people to rediscover and re-connect with their own inner creativity which has been beaten out of them by consumer capitalist society -- figuratively and literally in some cases. To start being creators and not merely consumers.
The problem is that AI art bots do exactly the opposite of this. They're inherently parasitical, feeding on the labour of creative people in order to allow others to produce commodities without compensating the labourers.
That's a gross and misleading oversimplification of what the Luddites were railing against. It wasn't automation per se that they opposed, it was the use of automation to devalue their labour, and the fact that they were being relegated to servants of the machines, forced to work longer and longer hours at menial tasks, and generally not benefiting at all from the improvements in technology. The Luddites were the fore-runners of modern labour and trade unions, and were fighting to prevent what you noted: "the vast majority of the population is going to find themselves paywalled out of existence within a century"
It's much the same system today, where automation in places like Amazon warehouses, instead of freeing workers from menial labour; they're relegated to even more menial work in areas that cannot yet be automated, where instead of being allowed to work less due to technology, they're being forced to work longer hours for less pay in order to keep up with the machines, and are penalized for simple human activities like using the toilet or needing to eat.
It's not just Amazon either, it's all over industry everywhere; automation is not freeing us from labour and making our lives easier, it's producing wealth for a tiny handful of elites who treat workers like easily replaceable machines themselves, and are not allowing them to be human. Which brings up another problem with AI, big corporations like Amazon are using AI to evaluate workers and make hiring-and-firing decisions based purely on their ability to meet ridiculously difficult standards of productivity.
5724344
The problem with this is that there is nothing you are describing that cannot already be done with other technologies, better and more effectively than NFTs, without the environmental harms caused by the blockchain technology NFTs exist within. The entire point of NFTs was simply to pump more money into the pyramid scheme, propping up the value of failing cryptocurrencies and allowing "whale" investors to see a return on their investment at a time when cash flow was practically non-existent due to the lack of ongoing adoption of cryptocurreny. That was its only purpose.
Incidentally, the sum total of the blockchain ledgers cannot be kept on "potentially millions of different computers", especially if cryptocurrency was widely adopted. The bandwidth simply does not exist for that to be feasible; and with increasing uptake, the bandwidth and processing requirements grow exponentially, by design.
5724339
I have to agree with 5724485 here. The burger flippers will be the last jobs automated. Meanial physical labor is inexpensive, meaning there's not as much economic incentive to automate it. Further, physical work, unless highly standardized in the long-term—like an auto manufacturer—often proves resistant to automation without a general-purpose robot that is currently an active research area (read: science fantasy to be profitable to "employ" with current technology). LLM-style AIs will rip through white-collar office jobs that never needed to be done in the first place. Instead, they need to be done (as in something, whether internal or an external regulation audit, will fall apart if they are not complete). Still, their quality is irrelevant [hence why they can be replaced with GPT dogshit with no immediate knock-on effects].
5724310
This is fundamentally untrue. In fact, established artists are not only actively opposing the unethical theft of labour that training these bots involves; but a number of them have banded together to file lawsuits, particularly a very large class-action suit filed last January against StabilityAI, DeviantArt, and Midjourney, which is currently working its way through the courts.
This is not exactly secret, and has been all over multiple mainstream news outlets; not sure how you missed it.