The AI Creativity Crisis
This article is inspired by a LinkedIn post from a digital artist with the courage to say: I’m afraid.
The bloom of AI-creativity observed on generators like DALL-E, DALL-E2, Unreal, Midjourney and their like is a boon for humanity and a bane for those who they have come to replace. Or is it? Let’s examine what exactly they can or cannot replace, and where we go from here.
Are you an Artist or a Craftsman?
It’s a trick question: In the creative world, one cannot be either or. It’s always a hard mix of both.
It’s widely accepted to think of Art as coming with the idea or statement, and Craftmanship as the ability to bring it to life.
(ℹ️) There’s also no absolute creative — it’s always mixed with ingest— the ability to detect, observe and analyze, but we’ll mess with that later.
If your main merit is to know which button to press and/or how to move the mouse with precision to get a certain effect—i.e. execution — then yes, you have something to be afraid of. However, not of machines, but of people with an art skill like yours (or better!) and with no craft skills whatsoever, who have access to creative AI.
With a wonder machine that spits out masterpieces by command — it seems that the weight of genius moves from execution to ideation, and people with ideas are found in plenty …right? Not quite. Let’s take a closer look.
The lifecycle of creative work
I recon the steps illustrated here.
AFAIK — this is almost the lowest denominator for all creative works, be it music, scripture, visual arts, plastic arts, movies, industrial design, software, architecture — even the design of work-processes for humans.
Simple tasks pass the entire process in one brain. Complex works would involve a lot of people performing iterations of every step and “passing the stick” to each other in the long relay race to completion.
Indeed — it starts with ideation. We desire. We want. We love agency. To inspect this process, let’s find a simple case to study.
We speak every day and what we say is basically our own creation. Some of us are great orators, but even those who struggle eventually get to say what they have to say in their own way.
Let’s suppose you have to tell a friend something important that is unpleasant to hear.
- Ideation — you decided that a message needs telling
- Concept — you ponder how to present it, logically
- Drafting — you decide how to approach them, what to say and in what order and consider the results. If that does not work — you try another concept.
- Execution — you finalize it into words using grammar and comprehension. If it gets too cumbersome or unmanageable — you consider another draft.
- Delivery — you approach, say what you came to say and observe your friend to see how it is received.
We all do it, and we do it so many times and the phase change is so quick that many of us stop being aware of that process.
Exploratory Art
Sometimes the ideation takes the form of exploratory or associative improvisation . In our metaphor of speech — it’s a simple spontaneous conversation.
Perhaps you can see it better in music jams, improv groups, artists that start with a blank page and start to scribble, or stone-artists that “help the statue come out of the stone”.
However, I’d like to stay focused on the former form of ideation, because although the latter is a driving vital force, it is the former one that drives our economy more.
The Human Skill
The common case is that the person with the idea is not the person who can perform it. And this is the first “handing of the stick” in our relay-race.
The bigger and more complex the project is — the more brains it will have to involve, and more relays occur. Even if creative AI will take over parts between drafting and execution, there is still work that cannot be eliminated entirely.
Each of us is a link in this relay chain — a box. But while creativity refers to what comes out of the box — digest refers to how what comes into that box is processed: how it is being detected and analyzed to affect the created result.
So here’s the first big takeaways:
Get better with people.
Improve how you listen. Specialize on detecting intent. Pay close attention to words — but not only. Read sub-text, mode, emotional state. Develop your social sense and emotional intelligence.
Personal is the new Professional.
Get personal. Remember your peers and their preferences. Acknowledge the person. Enhance the connection. Learn their balance between when initiative is expected, and when not to take them for granted.
If two people can command the generator — it’s natural to choose the one you feel understands you better.
The lowered Entry-Bar
Bottom line — yes, it lowers the entry bar.
The tools before creative AI require precise mouse or stylus movements, a click of specific buttons in a technical order, and provide a gradual modification to a single result. For creative AI generators, the input is some text description (that could be provided by speech-to-text without a single key stroke), and they provide a selection from multiple refined drafts.
So the new skill is knowing what input should one provide the generator to make it produce a result that is the closest to the vision, and choosing from the spat marvels the best suited draft.
Just like searching on Google-Images, but on hyper-steroids.
The Human-Machine Interface
The current state of creative AI:
- They take a limited input and understand it based on what they learn from curated info observed on what is said in the general public.
- They produce results that aim for the curated aesthetics and morals of the general public.
- The cutting edge applications know to retain a short term context in the form of performing a requested change to a given picture, but each request does not depend on its formers. Yet.
Here we can evolve in two ways.
One — the generators will keep specializing in shooting for the general public, and allow more and more specificity in their input. The input is then understood in relation to the point of view of the general public. The skill will be directing them accurately through the world of associations and knowledge they have. Descriptions will get longer. Knowledge about common mistakes, “gotchas” and algorithmic bias will become a private lore that will be under constant threat as the generator evolves.
Two — the generators will start to recognize us and know our preferences like search algorithms of social media do. In this case, the exact same request by two different people will yield a different result — i.e., giving back some weight to the person that speaks the command, because this person’s identity brings with it a unique private(!) context.
The human Craftsman
The lowered bar means a sea of “craftless artists” that access to an AI opens an opportunity for them…
But above them, there will always be another layer of those who know how to take the best generated result and re-touch it to the next level: modify it in a way that could not be achieved by refining the limited input that the generator ingests.
Though the skillset to bring a vision from idea to the material world may still thrive among hobbyists, IMHO the professional layer will dwindle to an exclusive bunch. That’s because for many of the applications in the industry the generated drafts will suffice.
With time generators will know to handle more complex and specific input but will also improve on shooting for what the general public wants. But what if your message is in the fringe?
IKEA did not eliminate carpentry, but it did make it rarer.
Economic Viability
IMHO, there will always remain a niche of performers who are able to understand in a free language what the owner of the idea wants and make changes to the best generated result to bring it closer to the intended vision.
But close enough to whose discretion?? Who is this stubborn person for whom it is really worth paying money to move a visual object 2 pixels to a random direction on a whim? And how many such people are there?
Will they be like those who nobody else but them cares about the difference between Chardonnay 2014 and Chardonnay 2015? But more importantly, will they get to position themselves as those that where it really matters, campaign managers would trust them and only them?
Observing the diversity of our society — I daresay it will be a bit of both. It still means that this niche will become narrow and exclusive over time, and in my opinion, generators will become an essential part of their life as well, just like power-tools to the modern sculptor.
A marble sculptor looks at a stone and asks himself — “what figure is hiding inside?” — which is a flashy cover for the economic question: “the revealing of which figure will require removing of the least stone for the same pay?“ That’s because still, it is better to choose the starting point closest to the vision before starting to dig into the rock.
True, the marble sculptor has a far lesser selection of start points and has to resort more to find the figure in the available stone, but the analog holds: Digital artists — beginners and masters alike — will choose more to get their starting point from a generated result instead of starting from a blank.
Learn the tools. Break the ice AND jump into the rabbit hole. Come from the other side an empowered creator.
Conclusion
As Creative AI Generators become more accessible and deliver more useful results the weight of genius moves from the technical tedious parts to the more human parts — ideation and communication.
Creative AI will become an indispensable tool for any creator of image, sound and text. With them few will produce more.
And who knows, maybe in the future there will be generators for smell and touch , and maybe even generators for the higher level senses: humor, balance, movement, time, social— all these senses that actually constitute our common space of being, through which we get to express and experience.
You may like also: everything ChatGPT tells you is made up
P.S. — …a bonus part
I should have ended there with that optimistic note, but I could not. That’s because this discussion is a local diversion from a bigger picture — the real discussion to be made, or at least be acknowledged.
The video below is 8 years old at the time of this writing. Here’s its conclusion:
“This video is not about how automation is bad, rather that automation is inevitable. It’s a tool to produce abundance for little effort.
We need to start thinking NOW about what to do when large sections of the population are unemployable through no fault of their own. What to do in a future where for most of jobs — Humans need not apply” / GCP Gray
Maybe not all arguments and predictions are tight, but the message is definitely there and the trend is already apparent.
(ℹ️) The link brings you to the part that talks about creative arts and why they are not protected from automation, but consider watching from the beginning to see how it builds to that.
The video started an online discourse and got a lot of reply productions.
My favorite is this next one, despite the fact that it walks around the question.
(ℹ️) This link puts you in my favorite part, at the end of the video, but here too — consider giving yourself the full thing.
And, the current state of AI rendering:
(updated at Nov.2022)
Added at Apr.2023
