I decided to experiment today with making art with artificial intelligence. I've seen the hype and conversation around this for a few weeks now and thought I would have a play around with it and see what it's like for myself.
I chose to go with using Starryai as they give you 5 credits free to start with. All I was after was the chance to experiment so this was ideal. Please note, this review is entirely independent. I'm not receiving any paid or free benefits for mentioning this app.
How to Make Art Using Starryai:
The interface is very simple and straightforward. You begin by selecting whether you'd like to create artwork, something photo-realistic, an illustration, or something from 'your style' which is only possible after you've generated a bank of work from which the artificial intelligence can draw inspiration.
After this is done, you can input your description. This is the most important part. The more specific your description is, the more accurately it can render from the prompts you've inputted. For example, if you input 'a fish out of water' you will get something more random than if your input was something like 'a fish, in a fish bowl, on a windowsill, at dusk'. It's up to you entirely how much reference material you provide for the Ai to work with.
Once these two things are done, you then have a few more basic toggles to play with. You can choose whether or not to give it a reference image. The generator attempts to stick quite closely to the shapes and colours of your reference so consider whether or not this is something you're after. You can then also choose a style in which to create the asset (from 12 in the free version but more if you purchase a license). Finally, you can choose the canvas size and then how long you want to let the generator run (more time equals more credits). Pretty simple.
Feeling both nervous and excited I pretty quickly punched in some descriptions. I was most interested to see how far I could push the logic of the Ai like putting fish in the sky or requesting a white background when my reference didn't have one. I also experimented with the various styles available.
Two Artworks Generated in Ai
Firstly, I was amazed at the run time. Within 5-10 seconds I was given back not one but four images for each prompt. I didn't feel that the Ai was able to handle suggestions that forcibly broke the logic. For example, I wrote this in one prompt:
A Painting Of A Gothic Villa In A Landscape At Dusk With Fish In The Sky
It gave me these:
You can see how it's produced everything I requested within limits that are logical. The fish are in a body of water in the foreground rather than in the sky like I requested. Other Ai services may be different or more advanced and if I'd chosen other settings in this one (in the toggles mentioned above) it might have generated something more wacky. But as a first stab, you can see how it works.
This next one really blew me away. The description I gave was:
An Enormous Ancient Oak Tree Filled With Doors, Pathways, And Connecting Bridges. A Fairy City. A Starry Sky.
Just look at the accuracy and richness of these! It's gone beyond what I requested and added atmospheric fog, some water, windows, and lanterns. They're just stunning. And this is where I really think Ai could be both a really useful tool and a curse to us.
The Value of Artificial Intelligence to Artists
Ai can be an incredible tool for generating original reference material. As artists, we all have to be careful about where our references are sources, whether they're fairly credited, and how we use them. Using Ai could be incredibly helpful to generate truly unique images to further flesh our ideas, rather than just be the end product.
In addition to this, as the mythical oak tree example above demonstrates, we can get ideas from this too! It's like having an assistant idea generator. Ideas about what colours to use, what additional details to add, and what composition we might favour are just a few examples of how much this tool could serve to better our own art-making.
Ai will call into question the ethics of authentic human creation vs machine-made and how we govern this. In areas like art making where plagiarism is so easy, it will mean we need to continue to scrutinise what we see online vs what we know in reality. It gets harder all the time to know what's real and tools like these won't make it any easier.
Another downside is that while the advancement of Ai is incredible and fascinating, it also encourages us to focus only on getting to the end of every experience as fast as possible and thereby missing out on the beauty in discovery and the potential growth available in exploration. While speed and efficiency are prioritised in so many areas of our lives already, I don't believe art-making should be one of them.
It can be quite easy I think to feel cheated out of the experience I'd have had by making this artwork myself. It can lead to wondering 'what's the point? This machine has just churned out four spectacular images in seconds that would have otherwise taken me weeks to create'. But really, the point of making art has not changed and will not change with the inevitable integration of Ai into our lives. It is in the making that we find our meaning as artists. We are always practicing art and having Ai to compete with is really no different than admiring another human's creation. Just that Ai is more a collection of many ideas to form a new image rather than it having come from one person. I think it's more important now than ever to remember and cherish the value of what it means to create for each of us. That it's truly more about the process and less about the result.