Tough love for Artists : the future of AI Art

Don’t want to be surprised again by the machine, like 3 years ago ?
Want a complete reality check on the current state of AI Art ?
Stick to the end of the post, and you’ll get support after the tough love.

The field of AI is slowly but surely eroding humanity’s drive to make Art. It’s happening in an underhanded way, and it’s important that we don’t let it backdoor us.

This post is meant as a serious wake-up call to anyone who thinks we’ve seen the end of the fight for Art.

(it’s also post 2 of a 2 part series on the future of AI Art. Feel free to check out part 1 — it’ll help you understand the concepts !)

post 1 was : Things that AI could do, but AI researchers have no idea

Here is post 2 :

Any groundbreaking, original piece requires an artist to do drafts, practice runs, and studies. Specifically for that piece. That is because the challenge is so high, there is a learning process inside of the creation process.

Mostly, AI was missing this up to now. But now, AI Researchers are starting to fill in the blanks with Reasoning models (we’ve covered this in part 1).
If you’re familiar with AI topics, you may have the following objections :

1- “The models that think aren’t the ones that make images. LLMs don’t generate images.” ; This is true, but the trend in today’s AI is unifying everything together. Soon the image and text (and audio, and everything else) models, will be fully all-in-one.

2- “Unlike artists, diffusion models aren’t iterating over their generations” (diffusion steps don’t count). This is also starting to not be true anymore. Numerous papers have adopted the “create-reflect on the creation-create again”framework.

But why are they doing this ? Is it because they know what it takes to make real Art ?No… They have no idea.
Like someone carrying kindling into the forest “to keep warm on the walk”, AI researchers are once more ushering in a disaster for Art.

Great artists are mostly single-minded and are extremely goal-driven. One way to look at human learning is :”when we define a goal that matters to us, we then learn through something called Reinforcement Learning”

Giving AI human-like goals (attention seeking, meaning discovery, self-overcoming, originality, curiosity…) could lead to more human-like AI art. And this can be done with Reinforcement Learning.

Sadly, it’s happening…

Like carrying kindling into the forest “to keep warm on the walk”, AI researchers are once more ushering in a disaster for Art.

This paper shows them dangerously inching towards a generalization of the Reinforcement learning objective for AI Art.
What does this jargon mean ? It means, they’re getting closer to giving the AI original objectives, beyond mere “aesthetics” or “prompt matching”.

All you need is a creative-minded AI researcher to give the AI another goal than to make something aesthetic, or “be a helpful companion”. So we artists need to brace for it, and keep our practice tight. Read until the end to find out how.

Without a physical body, there is no instinct, no purpose, no struggle… No Art.
And guess what ? AI is stuck on the computer !

Sadly, not for long…

Simulations are easily available, and robots are getting really good.

AI Researchers just haven’t realized the importance of this yet. Because AI researchers with a real grasp of art, basically don’t exist.

But they don’t need to know the disaster they are once again causing…They are like people carrying kindling into the forest “to keep warm on the walk”.

One more area of worry is consciousness research. For some reason AI researchers want to make AI conscious. And in their belief system, one of the main requirements is embodiment. This interest makes one more reason to think AI Researchers will get everything onto physical bodies — including AI Art.

Now that you’ve got your tough love, it’s time you get some support.

How do we defeat the machine ? How do we ensure we keep the meaning in our lives in the face of an unwise world ?

Here are the tenants of our fight at ArtPostAI :

• understand the enemy, and its weaknesses (as an artist, you WANT an AI researcher snitching on the AI for you ; that’s me !)

• galvanize the troops (I recently talked to an artist who was ready to quit a year or so ago. My content allowed them to stay on track. And that means I’ve achieved my goal.)

• Redefine the art of tomorrow : we aren’t powerless. We get to decide through our own effort, what the art of tomorrow will look like. Will it be computer-bound, novelty-chasing, anxious, tunnel-vision art ? Or will it be meaningful, slow and profound, highly soul-nourishing art ? It depends on our level of motivation.

Support this mission ? Talk to your friends about the ArtPostAI Newsletter. It really, really helps !

See you next time,

ArtPostAI