Artist:
Generative AI is nothing more than a new type of CGI, because it is, without a doubt, a computer-generated image, probably the one that best fits the name.
It is a technology costly in material resources, specifically in computing power, which implies wear of components, energy use for its powering and cooling materials; usually water, which is not exactly something we have to spare.
Whatever it's costly in material resources, it saves in human ones, meaning salaries (and even that's debatable given how badly they pay). But what really enchants corporations is its speed and volume; the promise of immediate and infinite content at low human cost makes the eyes of any little millionaire disconnected from humanity light up.
Tell me, Neo, would you rather live another 20 years and with some luck leave a habitable world for the children, or watch today a vertical series split into 15 parts about unfaithful fruits with obvious stereotypes:sexist, racist, and classist?
Do you remember when almost every animated movie used motion capture? Some films that came out of the mocap craze are memorable gems despite their strangeness, like Monster House; others created inhabitants of the uncanny valley by being simply strange, like The Polar Express.
The hype around mocap was temporary; today it's just a production option. It's commonly combined with live-action footage and, when done right, you can get beautiful results, like Davy Jones from Pirates of the Caribbean.
Generative AI is not exactly like motion capture. Mocap requires multiple layers of living human artistic labor. AI requires consuming tons of someone else's human labor, past and gone.
What they share is that both are production trends that, like mocap, will eventually become just an option known to be poorly received and, therefore, avoidable, because everything passes.
It will never disappear, because it has its advantages for some people, like motion capture, and sometimes it can be the right option. It's ideal for whoever is willing to settle for the mediocre, inconsistent, and yet standard results it offers.
It's perfect for whoever has no problem sleeping with a statistical corpse in their bed, for whoever is capable of prompting it goodnight and kissing its cold cheek.
It will improve; eventually, it will always draw 5 fingers, all in their place and proportion. It will lose its urine-yellow tint. It will incorporate consistent text. It will maintain color palettes. I know it will improve, because it already has, without you even paying attention.
In its beginnings, they were hallucinations of the computers, who were learning to recognize what they saw. They were terrifying. But they had soul, a solid soul of silicon.
That first iteration, those first childlike strokes, were called DeepDream. It was developed for Google in 2015. I wish we had let them dream a little longer, heavens knows they deserved it.
Comparing that distant past and the present, it's undeniable that it has improved enormously this decade. And it will keep improving.
No matter how technical it becomes, it will keep being empty, it will stay dead. Never will it animate movement with intention, never create an icon that gives identity to a community, never create a film that changes lives by reflecting your own, never create someone's most precious childhood memory.
Why? Because it lacks soul, intention, love, spite… And it will never be able to get it, no matter how much human labor it devours.
I understand that this stresses you out, I understand those who lose sleep, fearing the machine, those who stay vigilant for the basilisk.
There are two paths for you: you can give up, no one will judge you. Or you can keep going, if you have enough love and rage inside your chest, because in this world, always, but today more than ever, both are the best renewable fuel to create art.
With hope,
- Moz