Generative AI and what it's doing to us - part 1

Everyone's a bloody artist now

this is a gif. please feel free to use it.

I’ve been watching, from inside and out, what’s happening in the creative industries and what’s happening around them. I have a lot to say, too much for one piece, so I’m breaking it down into three or four.

This first one is mostly about why I hate generative AI. I explain some stuff that you probably already know, but I am aware that I live in a little bubble of my own, and don’t really know who knows what. It is very broad strokes, and I can already hear the stampede of pedants heading my way. Chill, pedants. I’ll get there in the 3000 words I write after these 3000 words. Or the 3000 after that.

First, AI

I really don’t like AI in general, or rather how AI in all its forms has been foisted upon us, every single human in the world, by a comparatively tiny handful of socially awkward tech bros (or sociopaths), without their wondering if they should or not, and without asking us first. I’m fairly sure the answer would have been ‘yeah, socially awkward tech bros, bring it on’, because most humans aren’t good at seeing the bigger picture and they like shiny things and quick solutions, but it would have been nice to have been asked.

AI — machine learning, neural networks, all those things — might one day help solve all the problems in the world, but only if those problems, let’s face it, will generate a profit once solved. Cures for diseases might be found or new meteorites heading for earth, but we can all bet that the finds taking priority will be the money making ones for the venal money worshipping twonks of the world.

click here for a useful primer on what AI is.

We normies (and that’s very much how the ubernerds see us) don’t know how far and how deep AI is being used in the infrastructure of the world, the war machinery, the food systems, the banking structures. They’re all infected and we don’t know by how much. Do we know that the AI scampering around all those systems is more faithful than the LLMs (ChatGPT etc) that hallucinate on the regular? It could all go tits up because a bomb-bedecked drone might read an instruction that is a made up line from a Jane Austen novel and lose its mind.

The AI that we don’t see being used in coding, in research, in infrastructures, the ether, may end up as AGI (artificial general intelligence, which is the singularity, or Skynet or the Matrix with a less snappy name and which is very much a possibility) or worse, AGI that hallucinates Jane Austen lines. That scares the daylights out of me, so, I try not to think about it. The socially awkward tech bros always know what they’re doing, right?

So, while trying not to think about robots wars in empire line dresses, let’s talk about GenAI, generative AI, the area of AI that is omni-in-our-faces all of a sudden, unavoidable if we brush past any kind of computer in our lives.

Generative AI

GenAI is the wing of AI that creates things seemingly out of thin air. It’s the shiny toy part of AI for the normies. It’s the large language models (LLMs) such as ChatGPT, Claude, Grok et al, and the image and video creation models such as Midjourney, Sora, Firefly and a plethora of others, including audio creation models too. Its rise to ubiquity has been fast and its progress in quality exponential. What the GenAI models can produce compared to six months before is always on a much higher level.

click here if you want to know how GenAI works.

I’ve been waiting for some tech bro millionaire to give me a good reason as to why we need generative AI. Why do we need machines to make stuff, the most human of stuff, let’s call it art? I’m not going to discuss here what art is, I already did that here:

We lost the battle

The reasons that are given are these:

1. It speeds things up so you can save money

2. It does the boring bits so you can spend time on what matters

3. It democratises art

4. It helps inarticulate people formulate their ideas, thereby giving everyone a voice

Hmmmm.

1. It speeds things up so you can save money

There are some processes in that most costly of the art forms, film-making, that can be helped along with some AI processing, for instance. But, if films are costing, say, $100,000,000 these days, who cares about another few mill on hiring a few more people? That’s simplifying wildly, of course, but I’ll expand on that in another post. All of my work apps now have AI tools included, whether I want them or not, and I can’t turn them off, and some of them COULD speed up some of my processes (I’ll go into those in another post), but do they really save me time or money?

An example.

You can skip this if you don’t want to know the ins and outs of visual effects and video editing. (TLDR: It didn’t speed things up and it cost more money)

Recently a client came to us and wanted a short film including a shot with a person with a smartphone using the client’s app, that is, a close up of a hand holding the phone with the app appearing in the phone. We do this a lot. Using a stock shot, I track the motion of the phone, which you need to do even if the hand seems to be steady, as it never is. I then “comp” an image of the app into the phone screen. Simple. But then they decided they wanted the shot to show the action of clicking through the app, which means making an animation of the app being gone through and making the person’s finger hover and click in the right places over time.

The deadline is the end of the day, and it’s already lunchtime.

There are several possible solutions to this:

• Film a real hand navigating the app - this is not practical. We don’t have time to hire a hand actor, and none of us are hand models and our offices aren’t prepped for shooting, and we know that there will be changes up until the last minute such as ‘oh, we forgot to mention the final button press!’, and we can’t be shooting hands all day.

• We can cut the hand out of the original stock shot, and animate it so that the finger presses the buttons in the right places on the screen at the right time - this would entail tracking the hand, masking the hand (the mask changes every frame), then cutting the hand out of the base footage (masking and replacing the phone edges and the background where the hand was, by far the hardest process in this), then animating the cut out hand to coordinate with the footage of the app being navigated. You don’t want us to that, because it would take all day and possibly most of tomorrow. For a five second clip.

• We can say, ‘no, don’t be silly’ to the client, you wanted this by the end of the day, didn’t you? Let’s cut to the app screen instead’ - so, we start with the footage and the hand, and then zoom into or cut to the footage of the app, full height on the screen, the hands and background now gone.

• Use GenAI to take that stock shot, and try to recreate it with the finger going to the right places at the right time. As an experiment, we tried this. It took us most of the day trying to get the damn thing to work, to get the finger going to the right place at the right time, to get the hand looking like it belonged to the same person, etc., etc., etc. before we resorted to not using it.

It didn’t save us any time, and it cost the client money.

2. It does the boring bits so you can spend time on what matters

What they mean is cutting out all the thinking time, all the learning time, all the decision making time, all the practice makes perfect time, all the donkey work time, all the time that makes us artists and designers and writers what we are. They erroneously think that “what matters” is the final touches, the fun bit of making something with pretty colours and brush strokes or fancy words. They think that the fun bit is when you send the work out to print, or stick it on instagram, or release the film, because that’s when you get paid. They don’t understand that when I’m slowly and carefully cutting out elements, or writing out my own words and rearranging them to make sense, or painting a line over and over again, because I can’t get it to look how I want… that’s where the art is, that’s what matters.

3. It democratises art

Art is already democratised.

Anybody can already pick up some tools, whether they are paints and paper or an iPad with an Apple pencil and an app, a camera, some sticks and glue, some clay, whatever, and they can make stuff. Have at it. It’s good for you.

However, for the last few years, because of social media, an idea now stalks the world that anyone can be a professional artist, designer, writer, illustrator, animator, film-maker, etc., just because people have seen a bunch of artists (I’ll just use artist as shorthand for all those trades) looking dead clever in their timeline, and want some of that. It didn’t really matter, though, because to be a professional artist requires skill and talent and perseverance and voice, so those people who just wanted to be artists because of the cool, fun angle were never going to make a dent in the world of professional artisting.

This isn’t to say that there isn’t a whole bunch of tasteless crap out there making money… in general, people don’t have taste and will buy any old crap. There’s money to be made out of badly drawn rainbowy unicorny sad looking anime looking girls leaning up against a tree etc., it seems, but what can you do? Sorry. I have earned the right to SOME snobbery, surely.

Now that GenAI exists, though, there’s a new idea out there that anybody who can imagine an idea can be an artist and worse, a “professional artist”. Get the fuck outta here. Writing a prompt, or writing sheets and sheets of prompts MIGHT be a skill, I grant you. Knowing the right words to write to get an image out of one’s AI buddy before one’s credits run out is something learnable, I guess, but it’s a skill on the level of knowing how to play GTA without killing too many pedestrians.

They don’t have the talent, skill, taste, perseverance, or voice to see that the drivel that comes out of the AI is drivel. And they are selling that drivel to the gullible, offering their services as professional illustrators and designers and writers.

To be fair, it’s harder and harder to discern what is and what isn’t AI, so we have to be kind to the gullible.

All of this is muddying the waters for everyone else and in many cases, taking work from professionals. This I’ll expand on further, later.

The internet removed much of the gatekeeping around art. Curators, publishers, agents were no longer entirely in charge of the gates. This has been positive in many ways, allowing new art forms to emerge, especially those that don’t fit into traditional publishing, film, TV, galleries, etc. Now, AI is helping to destroy the last vestiges of that gatekeeping, giving everything a superficial gloss, and because gatekeeping is now a dirty word, people are blind to what that means. What it means is that DRIVEL REIGNS!

Not everyone has the right to be a professional artist just as not everyone has the right to be a doctor or a zookeeper or an astronaut, just because they think it looks cool. Some people really don’t like it being pointed out, but one needs to have innate talent, and then to spend years training and honing that talent, understanding one’s direction and limitations, experimenting with media, thinking about taste, and learning new skills.

And being a professional artist is not cool. Ok, it’s a bit cool, but mostly because we decide what to do with our days, and sometimes it’s just playing. But playing is where we learn, and the learning all goes into the work. Most people are not built for being artists.

Being a professional is not all cool, then. It’s a precarious life, and even with all the talent (innate) and skill (learned) one might have, we are all dependent on the whims of the markets, fashions and tastes of other people to pay for our livelihood. This is true now, and has been true since the beginning of time.

Plus, one of the big important criteria for “being an artist” is the ability to self-critique, endlessly. Most people don’t have that ability. Being a professional in the creative industries means knowing a lot about your whole field and, importantly, knowing enough to know what you don’t know and to investigate.

4. It helps inarticulate people formulate their ideas, thereby giving everyone a voice

If you’re using an LLM to help shape your ideas, then you’re giving up your critical thinking skills and your voice will be that of the LLM. People are starting to notice that every piece of text that comes out of an LLM sounds the same, has the same tone. Well, of course it does, because it is created by AVERAGING all the text in its system.

Stick an “idea” of yours into an LLM and it will expand on your idea with thoughts you should have had yourself but didn’t, make a set of points and bullet points to support the thoughts you should have had yourself, and gives you a voice… its voice.

I saw an “author” (by god, there’s a lot of “authors” these days), the other day complaining, without irony, that she seemed to be losing her writing style, since she started “writing” with an LLM. No shit.

So, the reasons the socially inept weirdos give are, in large part, WRONG.

But I will, as mentioned above, go further into the AI tools that are available in the ART PRODUCTION PIPELINE OF THE WORLD that we can discuss, in another post.

A quick run down of the rest of the reasons why GenAI sucks arse.

environment

While governments are (pretending to be) aiming to go net zero they are simultaneously green-lighting the huge energy guzzling processing centres belonging to the tech bros. GenAI uses a vast amount of electricity, and a vast amount of water to cool the machines run by that electricity. It uses vast amounts of computing machines which need replacing and upgrading, plenty of rare elements going into them chips, and those computing machines need to multiply in number every year. Makes net zero sense to me. [fill in your conspiracy theory here] [no need, it’s just $£€, obvs].

intellectual property

This was the first thing that got the writers and artists’ attention a handful of years ago, when we realised that GenAI models were scraping the net for every single word and picture it could find to train on. Everything any of us had ever written or drawn, even if covered by copyright, if it was on the internet it was red meat to the machines. But that wasn’t enough, and some of them digitised physical books. DRM (digital rights management) means naught to this lot, and a watermarked image? Forget about it.

So, everything that GenAI runs on was stolen from all of us. We weren’t asked, we didn’t give permission.

There have been lawsuits, but what does a couple of billion matter to a company like OpenAI or Meta or Anthropic? They. Don’t. Care. If you have ever managed to listen to Sam Motherfucker Altman (OpenAI) without throwing your device out of the window while screaming ‘Motherfucking Altman!’, you’ll see that he does. not. care. On being asked about intellectual property, other people’s, he just said ‘oh, that, again…’ wearily. Try taking his IP from him. Or Zuckerberg’s. Or Elon’s. See what happens then. They’ll sue your arse.

Regarding intellectual property, i.e. copyright, I wonder how many people understand that the garbage they make with AI that they put on their book covers and their nasty over-engineered flyers and their tshirts and product listings and logos… those things are uncopyrightable, since they weren’t made by a person. Copyright requires for a human to have made a very large proportion of the thing. I read that a couple of people have tried to sue having had their AI designs “stolen” (oh, the irony) but were unsuccessful. Yes, the schadenfreude is quite delightful. We could go “stealing” their stuff for fun, but it’s so goddam ugly, what would we do with it?

time wasting

AI doesn’t just not save time, it is now an active waster of time. Designers of all kinds are now being given things that were made in AI to either copy or defuck. I promise you that in most cases, it’s quicker for a designer to originate something than to unfuck something that was made in AI. Plus the time it takes for us to let them know that none of what they’ve given us is copyrightable or licensable for distribution.

What does use of AI say to artists?

It says “I don’t care, I am uneducated and uncultured, basically, I have no taste. I have no respect for artists.” We read you loud and clear. We always knew but now you’re wearing that message on a t-shirt.

why it could fail… (ever the optimist)

Nothing in the modern world hangs around for long if it doesn’t make money. Big money.

A lot of people are resisting GenAI, and walking away from producers that use it. Hopefully that trend will continue, and the main use of AI will be by hobbyists for fun, i.e. no money.

However, this depends on our being able to recognise it, and on transparency from the producers. As the quality of GenAI improves exponentially, this will get harder and harder, but the optimist in me sees a glint in the darkness that maybe it really is just a shiny new toy that people will tire of one day. Think Tamagotchis.

Later I’ll address some other parts of this whole mess, such as:

• What people misunderstand about digital creation

• What GenAI tools are being used for that you may not know about, things we might be ok with … or not… and where AI is taking work from artists.

• How GenAI is fucking with society: deep fakery and stupidity. Spotting GenAI.

Kumbaya and out.

and the Romance languages version. have at it.