AI and art production

I started a new job this month.

As I started interviewing for jobs over the summer, everything in tech seemed to be in AI. At first, this was an immense turn-off. But then it spurred an investigation, and a feeling that we are defining a new way of thinking about ourselves as humans, new metaphors for how to describe our world. And it's a movement I want to be part of - there are downsides and great advances possible, and staying in the mix and advocating for the positives and paying attention to the externalities is worth doing.

And in the second week of the role, I strongly felt that “Artificial Intelligence” is just more compute.

Two tasks stood out.

First, a personal information gathering task that will be familiar to many parents of all eras.

I knew I needed some sort of health release form for one or both of my sons to participate in athletics at their school. My previous approach - go to the website, start clicking links, look for the form.

In a previous generation, perhaps I would have called the office. On this day, I used an ai tool. Rather than just asking the ai a general question, I specified the domain: “Using cfsd16.org, tell me what forms are required for a 4th and 6th grade boy to participate in soccer and link me to the form.” After reviewing the answer summary, I clicked on the link and verified, but it certainly was a quicker procedure, and it was not anything frightening. It gave me more text and context than I maybe needed, but thoroughly answered my question, and I didn't click 10 links to get to it.

And it was just a different way to answer the question.

It was not in a file drawer, or a card catalog. It was not searching through a website navigation and trying to understand a model of UI.

I can see how websites have proliferated to provide answers to humans. It now makes more sense to me, and seems less disturbing, that in the same way web pages have slowly become optimized for search engines rather than human consumption, they will soon be annotated in ways that are more easily understood by information engines / crawls. The people who previously had businesses typesetting yellow pages have long gone away - other things will go away too.

The second task was more specific to my line of work. And also on the subject of careers going away.

I am a graphic designer by trade - I have worked in the industry for 20 years, and much of it has been in storyboard design, concept design, brand design: fields that are prominent in the narrative of “AI replacing humans”.

Fortunately, my career has been built on using whatever existing technology there is to get the result I want. And these tools do that incredibly quickly, enabling me to take approaches that I might have disregarded for lack of time. Truthfully, the things that I am tasked with and produce in a day now might have been explorations with multiple designers over the course of multiple days 20 years ago. And that has brought its challenges, but also excitement - it’s really phenomenal to see amazing output. Since I graduated from college, I have had a simple goal that has always been met: I want people to give me money to make pretty things on my computer.

The overall task: I was charged with making an icon set that extended to a look and feel for an emerging product. I attempted to integrate AI in two ways, one more successful than the other.

The project started for me as it typically does: constructing a mental model for the vibe the site might convey, using the name it was given and what it was trying to do for people.

As part of my process, I research competitors to avoid anything that was too similar in the market. I think about trends that are currently in the market, where they come from and what else was in that milieu that might feel similarly good but be distinct or be combined with something else to have a twist.

For this, I attempted to get AI to take my reference screenshots for me. This is a grunt task, and I would not miss leaving out the screenshotting and categorizing of what is discovered. It’s a task that might be assigned to a jr designer, and a task that does help spur ideas - thinking while your hands are busy. But it’s also something that could disappear from human tasks and not be missed.

However, AI could not complete that yet - it wasn’t capable of doing more than an image search of the specified topic (current ai tools) not screenshotting (yet!). And with the image search, it only came up with logos of the most prominent 3 ai apps, none of the new ones. Of those, there was no UI in the image indexed, just logos.

It can’t do that grunt work yet, and more time might be spent in formatting the prompt than just doing one’s own search and screenshotting. Sigh.

Moving on, there was some wood grain in the latest design exploration of the product I was designing. The wood reminded me of the use of wood in electronics in the 80’s - a trend that has never quite been revisited. That led me to industrial design, 80’s typographic annuals, and eventually also Bauhaus.

From these diverse influences, I had an image in my head of how my previously conceived icon concept representing an image being generated might expand into a larger graphic that illustrated the same theme but in a Bauhaus style. This image may or may not be used in the final branding, but it felt like a way forward, an inspiring direction to push.

As the Bauhaus references met up with my idea for how to picture images res-ing in as a loader in my head, I turned to Midjourney to do a task I might have previously accomplished by making a vector illustration and applying a texture.This was just quicker, and randomized the eventual result in only the smallest way - the extra interweaving of the hills recalls the source reference and I might have dropped it for simplicity. I was prompting to get a result in my head, and weeding out the ones that weren’t it.

The final result felt in the world of the reference image, but it was a net new artifact, and helped lead a path that would diverge further from the historic reference. Nothing in this process made me feel like I was stealing other's work any more than any previous process I have taken.

“Nothing is original. Steal from anywhere that resonates with inspiration or fuels your imagination. Devour old films, new films, music, books, paintings, photographs, poems, dreams, random conversations, architecture, bridges, street signs, trees, clouds, bodies of water, light and shadows. Select only things to steal from that speak directly to your soul. If you do this, your work (and theft) will be authentic. Authenticity is invaluable; originality is non-existent. And don’t bother concealing your thievery - celebrate it if you feel like it. In any case, always remember what Jean-Luc Godard said: “It’s not where you take things from - it’s where you take them to."

-Jim Jarmusch

The nature of how the image generations are reached is really interesting - computers work with mathematical equations. So much of what consists in becoming a trained designer, and indeed human ideas of beauty, has scaffolding based in mathematical equations. The golden ratio. The rule of thirds. Perspective. Generating many images and seeing how compositions echo each other first makes you wonder about the source references, but then you look beyond that and think about what they have in common. They are "good" because they have a strong point of focus, foreground and background, and content is grouped in proportional sections of the image. Are these "owned" or a principle we inherit from the collective body of creative practice?

This said, it is possible to steal in the bad sense. I could ape an artist's style. But that would not only be unethical, but bad business. I'm in the business of creating new things.

And both before and after AI, fewer and fewer humans are needed for art production (as with all kinds of production).

I can be sad for those lost jobs. I work in a field that none of my family knew was a field when I was growing up. My uncle, a teacher like most of my family, laughed when I said I was studying design, and asked what I would do for money. There are inevitably jobs my kids will do that are not on my radar.

I have been reading "Age of Wonder", a book about science in the romantic era. While in my lifetime I can remember when jobs were dedicated to things like the creation of annual report books, and books filled with phone number directories for the people and businesses were printed every year, I cannot recall a world of artistic professions anything like what was described. In the first section, about Joseph Banks, the first contact with Tahiti, and the development of the field of Botany, I noted that two illustrators were taken on the journey, their skills essential to the voyage. In the second section, on William Hershel and his foundational astronomy discoveries, I noted that he supported his amateur astronomy habit with a professional musician job. And his indigent sister was trained as a singer in order to bring in an income.

We no longer live in a world where musician is considered a great option for a steady wage. The age of mechanical reproduction long removed that - when everyone had a record player the march began. When cameras arose, painter was no longer a profession so much as a calling. And now, perhaps, that's where graphic design is heading.

There will still be design, but there will be fewer professionals, and those professionals will be wielding tools rather than directing other individual craftspeople.

As something is lost, we can hope that something is gained. I pin hopes on the scientific and medical advances that this amount of compute may enable. And feel that we will be living in a different way of interacting with our environment in 5 years, though its not quite clear how dramatic it will be from here.