"Creativity" and AI-Generated Imagery

Asking if AI-generated imagery is creative misses the point.

Disparaging it as "useless," "cheating," or "just copying," does no better.

Generative AI technologies, such as Midjourney, are tools. Specifically, software tools, that help with one or more tasks. If using it implies "cheating", then so does using every other piece of software ever built.

Such tools lower barriers by reducing the amount of work required of the end user. Understanding just how broad the definition of "work" really is in this context can be quite unsettling for many people.

Part of why "work" can be so broadly defined in software comes down to this, it isn't just a simplistic, kind of "raw" work. It's work that is directed by the information that's been distilled into the software (ex. formula to calculate interest). For the user, it does two kinds of work.?

1. Makes available the work of others on how to calculate interest.
2. Performs the calculation itself.?

The result is, directed work.

And the "direction" that work can take can be quite sophisticated. For example, it could be the product of mastering the application of mediums like oil, graphite, or charcoal to specific types of surfaces. And in reality, this kind of software tool has been doing this kind of "directed work" for several decades as brushes etc. in your favorite paint program.

Yet, we do not credit these tools for any creativity we may see in an artist's work. Nor do we detract from the artists' creativity because they used these tools.

There is a critical point here, that is actually central to the questions about creativity and work done using generative AI.? The work the artist did to capture the familiarity and ability to make use of these tools was not a creative act. Not generally speaking, unless they were doing new research, or rediscovered some technique. An artist may employ creativity when performing such practice work, but the work can be done without it. The majority of us acquire such things by learning about what's gone before and emulating it.?

Although generative AI represents a new method for distilling and encoding the information that makes up half of software's directed work, the paradigm remains.?

?One can certainly open up a modern paint program, grab a "pencil tool" and see marks on the screen that are reminiscent of what a lead pencil will make on paper. I think we'd agree this, in and of itself, is not creative.

You can also use generative AI and type in "oil painting, horse" and see a reasonable representation of an oil painting of a horse on a canvas. This, too, falls short, in terms of creativity, IMHO.

Yet, both tools can be leveraged to display a person's creativity. While the paint program is a "less evolved" tool, and requires more raw effort and ability to translate the creativity from the person's mind to the screen, they both serve a similar purpose.

To whatever extent a problem, exists, I believe it comes down to people needing to update their internal rules when it comes to imagery in general.

Our brains are constantly looking for shortcuts, unfortunately, this appears to be doubly true for making judgements. "Is this a good artist?" "Are they creative?" Even something as simple as, "do I like this image?" can become a bit confusing as these boundaries are redrawn. In social media circles, these questions have often been boiled down to a question of shared aesthetics and detail counts. Something you get for "free" with generative AI.

Our brains don't like this because it breaks our system of shorthand for deciphering this aspect of the world. So, we look for a reason for our irritation and for many, it would seem, they land on a sense of indignation over "cheating." In an interesting reversal, one side effect appears to be that people are looking for shortcuts for identifying AI-generated images, versus shortcuts to identity those artists who continue to be creative. A correlation that seems to better match what people are feeling when they see these images.

As times goes on, we'll sort out what our signals are for questions of creativity again, but in the meantime, I'd urge people to make allowances for the fact that one can be genuinely creative with these new tools. But how we identify that creativity, and perhaps even craftsmanship, will need reconsidering.

I've included examples of some images I generated since I began, over a year and a half ago. (These aren't meant to "wow," especially when it comes to shared aesthetics and detail counts.)

Look at how short the prompt, "paper cut craft Morgoth," is! Yet, I do remember carefully considering what would be an interesting juxtaposition of weak and strong. So I settled on the flimsiness of the craft paper compared to the head dress of Middle Earth's greatest foe. In my view, this is the beating heart of creativity. As a person with a creative bent and ADHD who was never been able to obtain the abilities needed to transform my thoughts into something so aesthetically pleasing, it's been a godsend.

Some were definitely more complex, in that I had to generate multiple previous images, then combine those in various ways.

At the time of writing, Midjourney 6 is the current version.