Can a robot do your creative work better and faster than you?

Summer Reading: Does an AI Art Creator Respond to a Client’s Brief as Effectively as a Human Painter? Tim Gibson took some of his creative business and took on robotics.

First published September 13, 2022

Robots come after our jobs. We knew they were coming after the truck drivers, checkouts, and call center employees, but I thought I’d have more time to prepare as a creator. Now, artificial intelligence (AI) art programs such as Dall-E 2 And Medjourney They storm my little ivory photographer’s tower.

Is the danger real? If so, are artists the first creative infantrymen to be cut down on the front lines? Or are AI Art Generators just a new tool in the design toolkit?

Just a few years ago, illustration and design were listed alongside nursing and social work as the industries most likely to be affected by AI-related job losses. They may have had another 20 years. But the life expectancy of illustrations has been shaken by the invasive disruptors that are AI art generators. In an effort to gauge the current sophistication of these systems, and to get a sense of their competitiveness against a human counterpart, I’ve run some completely biased and unscientific tests of my own.

I give explanatory summaries of the AI ​​art generator from Dall-e 2 that I’ve already completed, and then we’ll measure the results semi-objectively against the real-world artwork.

First, maybe it’s worth comparing and contrasting the traditional method of getting artwork for a beer label/poster/marketing campaign against the new “instant” AI system. Either way, someone in the company decides they need clarification to express their thoughts. Then, they might gather a few smart thinkers, brainstorm some ideas internally, and decide what the artwork needs to accomplish, who it should appeal to, some specific content ideas, and maybe an idea of ​​the art style (Impressionist painting vs. 50s comic book, etc.).

Here the two paths diverge.

If you are using a human artist, you will first try to select the right person with good artistic alignment for the style and energy needed. After some negotiation over budget and overuse of the term “thriller,” you’ll sign a contract and release it. If the illustrator is broad-minded, they can “yes, and” the client’s ideas, spin ideas in new (better?) directions or carefully and collaboratively improve the brief before they even pick up the pencils.

Within a day or three, the client may recover some rough artwork, and depending on how close they are to the client’s needs and desire for the artwork, things may swing a few times to narrow down a final, polished work that will reasonably satisfy all involved. Good things take time, but the process can take anywhere from two to four weeks.

However, for a non-human illustrator, the client takes the abstract and reduces the underlying topic into short, visually descriptive sentences and feeds that to the AI. The AI, in literal seconds, spits out some hot visual shots. It should be noted that although AI systems themselves are often built in a way that prevents them from creating explicitly opposing artifacts (some quick button terms cannot be used in a prompt etc), they also cannot push back or forward, Bad ideas – which, let’s be honest, customers sometimes give.

If the AI ​​results are incorrect, the customer can either change their written input to redirect the AI ​​or request variations on an artwork that is closest to what they want, hoping that serendipity will bring the work closer to the desired goal. Experimenting with a different AI became a viable option.

Now, in addition to being an illustrator, I’m also a brand guy and art director who often selects and mentors other illustrators for clients, so I’ll do my best to wear that hat (not my fearful photographer hat) because I’m giving the Dall-e 2 AI an “exciting opportunity for exposure” by Completion of some of my old abstracts.

HĀPI DAZE

Author’s origin

original prompt: “View from New Zealand bush to a lake and Miter Peak, fresh daylight, two men sitting by the lake on deck chairs, 1930’s tourism poster style”

Photo: Supplied
Dall-E version 2

AI score: 4/5 stars. I had to be more specific with my placement of things and replace the ’30s tourism poster’ with colorful directions and the term ‘screen print’ to escape the style of photography he was pushing. I assumed the people on the deck chairs were 100% wearing hats (which was interesting), and he did a great job of the New Zealand specific looking bush and general scenery. They feel a little “school project” to me, but some are headed in interesting directions.

Leader level

Author’s origin

original prompt: “Giant Gundam-style robot, in silhouette with small glowing bright dots, standing in a large, dark sci-fi corridor, flashing lasers and lights on the walls, ceiling and floor, Synthwave art style.”

Dall-E version 2

AI score: 3/5 stars. I think being able to give specific art style and content references helped, but I (standing in as art director/client) didn’t manage to get certain notes or elements in the piece. It felt a bit like working with a very talented illustrator who doesn’t read their emails properly. I have no idea where all the glowing crosses came from – maybe from the synthetic Wave of Justice album covers?

Tsuyu Saison

Photo: Supplied
Author’s origin

original prompt: “A swirl of orange koi fish swims down a flooded cobblestone street, with a reflection of a businessman with an umbrella overlooking traditional Tokyo. Japanese print in Ukiyo-e style.”

Dall-E version 2

AI score: 1/5 stars. I felt like I was bumping into the limitations of AI that humans don’t have — the ability to combine multiple ideas and elements into a single illustration. I’ve reformulated and edited this prompt several times, and it’s only when I’ve simplified the requests by removing certain content that the results start to look better. Really not fit for purpose.

Cat pajamas

original prompt: “Lots of cartoon cats, standing on their hind feet like people, some in costume. Black and white, linear, illustration style like Matt Groening. Cat in the middle wearing pajamas”

Dall-E version 2

AI score: 1/5 stars. Things got a lot more fun when I eventually added: “The cats are having a fun masquerade” to the command prompt, but overall it’s either devoid of life or a full-blown horror show and just didn’t find the sweet spot in between.

perennial herbs

(photo: attached)

original prompt: “Giant monster made of hopping vines and leaning on 3 classic cars, in a jump field, the sky is dark and full of stars and moon, Tim Gibson style.”

Dall-E version 2

AI score: 3/5 stars. For chuckles and chuckles, I’ve added myself as a style reference, in the same way you might use “Keith Haring” or “Rita Angus.” The Dall-E score suggests the possibility that despite my arbitrary identification of my data to giant tech companies through years of phone, social media, and smart home interactions, Open AI/Dall-E doesn’t know who I am. Which comforts, humiliates and insults you all at once.

Ego aside, he’s been confused as to where the scene/setting ends and the monster begins, but even so, there are some interesting stylistic variations he suggested, that would be fun to explore.

BOSSA NOVA T-Shirt (SUPER EXCLUSIVE NEVER PRINTED)

Author’s origin

original prompt: “A delightful toucan sits on Carmen Miranda’s fruit headdress and throws pieces of bright fruit everywhere with its wings, causing real chaos.”

Dall-E version 2

AI score: 3/5 stars. Real photos of Toucans wreaking havoc are high quality deep fakes. The AI ​​couldn’t pull off the complexity of all the quick requirements, and it didn’t seem to get the reference to Carmen Miranda, or her fruity outfit. In the end I replaced this phrase with “fruit hat headdress” to get better results. Lots of fun developments though, with some unwanted anthropomorphism thrown into the mix.

The life-size toucan doodling with the figures of the toucan’s head veers into fever dream territory, which is difficult for most commercial illustrations to do.

So where is the AI ​​art?

Can artificial intelligence draw? Well, and depending on the AI, it can produce gorgeous paintings or photo results. Can artificial intelligence generate ideas? Yes, but like a really annoying employee who only listens periodically in meetings and goes in weird shades when he’s holding up whiteboard markers.

And artificial intelligence is very fast in producing work. It might not be what you asked for, and it probably wouldn’t have upgraded your original abstract like a good human creation, but you can burn a ton of artwork ideas in a fraction of a normal creative post. Free.

Do illustrators need to put down their pencils and cancel their Adobe subscriptions? Do we need to pick organizations outsourcing their creative work to Big Tech? Well sort of. Some publications have already been publicly shamed for saving art budgets by bypassing humans in favor of artificial intelligence. It should be noted that some of these choices have been made by journalists and editors – one step closer to meeting similar AI challenges -. What is this quote? “First, they came for the socialists, and I didn’t speak out because I wasn’t a socialist…”

Thanks to the AI ​​revolution, photography (already full of insufficient content, designs, and criticism) could find itself completely eliminated. Most of the marketing you see about these art generators are examples of great execution of simple ideas: “a blue apple,” for example, or “a storm in a teacup.” I would be surprised if the big stock photo market didn’t license, subscribe to, or even buy an AI art generator and use it to slowly replace human photographers. Just like Uber’s very clear intent to inevitably replace human drivers once self-driving cars become viable.

Then there Medjourneya replacement for Dall-e 2 which would be intimidating if I were still a fantasy/movie concept artist.

Creative agencies and small startups will already be incorporating this stuff into their early exploration creative work, in the same way that mood boards are currently being used.

However, for now, I’d say using AI in the final work is a bit like working with an awesome kid. Great if you are just following bliss and are comfortable about what you end up with. For those who need to design by commission, or at least include multiple requests from people in the creative work, human is still the answer. But this technological innovation happened with surprising speed. So ask me again in a month.

Like a lot of industries, tech companies in this space won’t slow down, and smaller companies that employ creative employees and freelancers will think about speed and cost-effectiveness. The only thing that will have any real impact on the continued existence of creative professionals is how we as a society—and as individuals—choose to value human creativity. Go hug a painter, today.

Leave a Reply

Your email address will not be published. Required fields are marked *