Back to Blog

Welcome to the Dawn of AI Art

No items found.

New AI art tools are proving popular and exciting to experiment with. Is it time to Panic or party?

In recent months we witnessed the emergence of AI art tools such as Midjourney, Dall-E, and Imagen. Designers, animators and all kinds of creatives are taking notice, and disputing what any of this means to the industry. Regardless of what we think, the programs are making a case for something we once thought impossible: Artificial Intelligence will soon develop truly compelling art.

As someone who specializes in designing fictional (and real) futures, I was both intrigued and a bit cynical about these tools. Eager to understand better, I jumped in and was soon swept away by this still budding phenomenon. What follows is some of what I’ve learned, as well as some expectations for where this all can lead. It’s exciting, dystopian, inspiring, and a little scary. Buckle up, the future has arrived faster than expected…

It’s officially the dawn of AI art—do we panic, or party? 

AI-designed art? What is this madness?

Walking towards a new creative future. (/imagine {prompt:midjourney})

In early June, I heard the increasing murmurs of a new AI generative art tool called Midjourney. I joined the beta, and within minutes felt as though I was witnessing a disruptive impact that could change everything I know about creativity. This was original art made by a computer. 

While still putting my mind back together, the gossip in the industry grew louder. Within 72hrs, I saw many of my colleagues and friends riding the edge of this shockwave. 

A new horizon. (/imagine {prompt: a futuristic sailboat in front of a utopian sunset})

This technological advance had me scrambling to create images in an all-new way:  just type a description, or “Prompt”, and Midjourney does the rest. In less than a minute, the A.I. returned 4 images. Each of those 4 options could be a final image, or a jumping-off point for another 4 variations. And they were interesting, sometimes incredibly striking. Critically though, it is fast. Ask for an image, and you will receive nearly instant results.

Is an AI really capable of making compelling visuals?

This image looks exactly and not at all like its intended result. (/imagine{prompt:vintage formula 1 poster graphic design minimalism –aspect 9:16})

As you play, you can be relatively casual with your prompts.“Show me this in the style of that” triggers a decent, often imperfect result. Midjourney clearly has its limitations. Results can appear coherent, but upon further inspection dissolve into psychedelic nonsense. Certain physical principles of reality can casually blur into surreal melted madness, often in a way that is absolutely sublime.

As I looked around my creative community, I saw amazing, specific, and detailed results coming from some of the best artists I knew. Did they have some advanced version of the beta? Nope— they had something else.

The robotic creative assistants take fewer cigarette breaks. (/imagine{prompt:tens of cubicles in a retro futuristic corporate office, humanoid robots walking around, realistic, cinematic, warm colors, 1960s, in the style of syd mead --aspect 16:9})

The most successful Midjourney results come from top-notch creatives—people who themselves can often create these kinds of images from scratch. But most importantly, the best results come from creatives with a strong vision. The same vision required to make an exceptional project with your favorite tools, or a team of collaborators, is a prerequisite for success.

Don’t like the results Midjourney is creating? Maybe you need to refine your prompt a little further. Be more descriptive, more specific. Pick better references. Have better taste. Yes—to get the best Midjourney results, you need to be a better director. 

The result from a prompt from the author’s 8yr old daughter. (/imagine{prompt: big rainbow tree})

And quite honestly, even the nightmare client, the “I’ll know it when I see it” agency ass-hat will still get nice results out of Midjourney. I’m out of my technical depth here, but there is something about the way the AI has been “trained”. It’s excellent at composition. Without being asked it generates very pleasing arrangements of detail frequency within a single image. Without being specified, it chooses great color palettes and authentic textures. The best I can tell, Midjourney is the quiet-but-talented senior designer… the one that makes a weak creative director look competent. And when collaborating with a master, it can make magic. 

Is AI coming for our careers?

Hey team, meet Hal, our new freelancer! (/imagine {prompt: medium shot of a t-800 skeleton robot wearing hipster glasses warby parker})

I don’t think tools like Midjourney are going to replace anyone’s job, at least not in the near future. At the same time, I’d be wary of ignoring tools like this… as they become more advanced they will at the very least be time-savers built into our design/animation/3D workflows. AI could be a godsend for any of the grunt-work we encounter. Making seamless textures, roto, even modeling optimized geometry at your request. 

The ideal end result? Creatives can focus on the concept, the big ideas, and pick and choose the technical challenges to take on. From my experience, that bit of distance from the technical weeds creates a newfound optimization of concepts, stories and strategies that resonate much more clearly with your audience. Hopefully every production artist has the opportunity to step back, and think like a director.

Any technological advancement our community has seen, whether better hardware, or easier to learn tools, or faster rendering has rarely eliminated jobs. Instead, it raises the quality of our output, raises our expectations, and raises our creative ambitions.

Fawning over an AI’s creation feels kinda dystopian.

Sure does! 

The super-satisfying feedback loop I have with Midjourney can be discomforting. It inspires me to figure out the best way to speak its language,—to twist my own thinking and writing to become robotic and compatible with its way of thinking. It even encourages me to stereotype entire swaths of creative output to simplify communication. Seriously—just adding ‘Octane Render’ to your prompt will result in slick hard-surface visuals with beautiful glowing blooms. You could use “Artstation trending” as a mindless prompt for “drawn the way the cool kids do it.” and it works…even if you feel gross doing it. 

Happy trees (/imagine {prompt: bob ross in front of a black background smiling at the camera holding a paintbrush in his hand in front of an easel holding a painting of the t-800 terminator endoskeleton})

Wait until we have an infectiously catchy pop-music hit that is later revealed to be entirely composed by an AI. Because it could tell what would resonate/trend/titillate. AI’s ability to read our patterns and get out ahead of us is a bit spooky. It can even get out ahead of a lead Google AI engineer

So, does this really change everything?

Even while being enthralled with the possibilities these tools present, my traditionally cynical mind wonders if this is a fad, the hot new plug-in. But I must say, this feels very different. Midjourney is currently an experiment from an independent research lab. It’s obviously the tip of the iceberg for tools like this with tremendous headroom for improvement. And even in this primitive state, I wouldn’t hesitate to call it production-ready for tasks such as matte-painting or texture generation.

It seems inevitable that these tools will go beyond making still images— moving images, 3d assets, interactive experiences and much more. As any AI is trained on patterns and history, it’s also a warning that human-kind’s taste has gotten pretty… predictable. Movies all seem the same. Disruptive fashion becomes simplified and homogenized for the masses. Even the fringes of music can be packaged into hyper-specific sub-genres. There’s no question an AI will eventually get one step ahead of any of these patterns. 

As machines become more human are we becoming machine-like? (/imagine {prompt: translucent human organ scientific futuristic optimistic utopian})

Content may become entirely on-demand. Not, available on-demand, but CREATED on-demand. “Hey Siri, I’d like to watch a 90’s style thriller starring Timothee Chalamet and Marilyn Monroe, with a Michael Bay car chase, Tony Scott cinematography, and a Shyamalan twist all in 2.35 aspect ratio.”

Ease of creation leads to over-stimulation. (/imagine {prompt: a utopian holographic diagram of a human brain, contained within the silhouette of a human shows all of the information ever known to mankind})

And in that moment, we could see an AI-driven renaissance. More staggering amounts of output than could be made by any entire industry. Hyper-saturation of every kinda of media (and you thought scrolling through Netflix was exhausting). And just when it feels like creative Judgement Day (Terminator, not Christianity) is upon us… we’ll revolt. We will instinctually trend away from things we discern as pattern-based, too AI generated. And in the attempt to push beyond the AI’s patterns, we will have no choice but to adopt an even-wider range of tastes and more genuine human-made unique art than has ever been seen before.

Consider me excited, a little scared, and developing a deep anxiety over whether or not this is all an AI-generated simulation anyways.

See you in the future!

John LePore is a creative leader known for designing technology for fiction (Black Panther, Spider-Man: No Way Home) as well as advanced future products (Hummer EV, Microsoft Hololens). John currently consults with clients, agencies and studios to design the future for Film, Technology and Automotive.