This image was generated by an algorithm — how AI art is changing design
The past few years have been staggering in terms of technological development. The explosion of the Metaverse and Web 3.0 have spearheaded a new era of the internet, with companies gearing up to adapt and evolve within these landscapes. It’s no secret that the largest tech companies are stockpiling assets too, with Facebook rebranding to Meta and Microsoft investing $68bn into the Metaverse.
Within these larger markets are even more interesting submarkets; one of which is AI art. Immediately you might think ‘How could a computer make art? Surely the spirit of art lies within human emotion and creativity?’ — and to which, you’d be right. To an extent. AI software such as Midjourney and DALL·E are challenging our preconceived notions of what constitutes art, and how it can be created. I was fortunate enough to gain beta access to Midjourney and try it out for myself.
What is it? How does it work?
In a nutshell, these tools use an autoregressive language model, which is a fancy way of spinning ‘machine learning’. They associate keywords with things (in this instance, images) from feeding on a massive scale and can then reconstruct based upon their understanding of these words.
A look inside Midjourney
At the point of writing, Midjourney has just launched into open beta. Anyone interested is able to sign up and access a trial version immediately. I was able to play around with it for a few weeks with the extra features accessible during the closed beta to get a feel for how this worked and a look inside the budding community.
First of all, it operates entirely within a Discord server. How’s that for flashy? It’s an extremely smart decision by the developers as it makes full use of the community features within Discord. There are a variety of channels within Midjourney, many for image generation, but also a few for prompt crafting, support, discussion and showcases.
Generating an image can be as simple or complex as you like. Of course, the more complex your prompts are, the more likely you are to get something closer to the idea in your head. You can put in pretty much any descriptive word you like and come out with something interesting, but the bot utilises certain words more effectively. Art styles such as ‘photorealism’ and ‘cinematic’ will take priority in the algorithm to create something in line with what you’ve asked for.
When it comes to specific styles, the server offers a variety of themed channels (accessible only with a paid subscription). These are not only a great way to understand which keywords will be most effective, but also allow you to draw inspiration from others. Everyone’s designs are viewable publicly, meaning you can see what prompts were used and the images generated from them. There are ways to privately generate images too, but the public aspect is likely why the future of Midjourney in particular is so exciting.
So what can I do with these images? You can use them however you like, as long as you’re subscribed to the service. Permitting you make under $20,000 from blockchain commercials (to stop this becoming an NFT cesspool), you can redistribute your images for personal or professional use. All at the low cost of $10/month. There are other packages at higher prices, but to have such a powerful tool available at a tiny price is extremely encouraging for the market.
How will this impact the world of design?
As somebody who is both a digital creative and pioneer of emerging technology, this is equally as exciting as it is scary. I can fully understand the criticisms traditional artists have with AI art, but I equally feel that the potential (and even current ability) of this market will overcome them. There’s a great video from The Studio that demonstrates the battle of AI vs human art, and within it you will notice how humans (currently) persevere.
However, this technology is still capable of great things in its infancy, with an infinite ceiling to reach for. Accuracy to the brief, consistency and repeatability are some of the most important aspects of a creative’s toolkit, and are things AI art currently struggles with. Due to this, there’s no reason artists should feel threatened in the short term. But, as technology progresses and these skills are refined, we may start to see AI art more frequently in our daily lives.
Where will this take us? Sure, these bots are churning out a few nice pictures, but what’s next? To circle back to the emergence of the Metaverse — where everything is digital — we could in theory have entire worlds created by AI. Everything from communities to fashion may be generated to be photorealistic within the Metaverse, to the point we’re technically living within a simulation. The future is bright, but let’s just hope we find a way to keep a human aspect within it all.
You can try Midjourney today at https://www.midjourney.com/
