Modules VTT Tokens Artwork A.I. Contact

Statement On How I Use Artificial Intelligence In My Work (December 2023)

I work almost exclusively in traditional media. I use computers for digitising, scaling, and cleaning up images, but I do not use computers as a medium for producing art.

Knowing that it is a sensitive issue, I want to be clear: I do not use, and have not ever used, AI as part of my artistic process for any commissioned work.

Outisde of commissioned work, I do use AI to produce images that I use for developing reference images, for practicing my art, and for generating concepts that I use in the RPGs I run. For instance, I use generative AI (MidJourney, specifically) to produce faces to practice drawing and painting people, rather than using images of real people. The artwork is produced by me, inspired/influenced by images generated by AI.

As a baseline, I follow the guidelines for the responsible use of AI put forward by the Government of Canada (https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai.html). I am Canadian, and after looking at various similar documents, it seems, to me, to be a reasonable and well thought out.

Collaborating with an AI (February 2023)

Midjourney (https://www.midjourney.com/) is an Artificial Intelligence program which produces an image based on a text prompt. For instance, if the user give it the prompt “a hockey team sitting in a maple tree”, it will produce a series of images that, according to its algorithms, approximate a hockey team sitting in a maple tree. The user can also set the level of detail, the aspect ratio, and other aspects of the image, including asking the program to produce an image in the style of specific artist or medium.


Midjourney’s results for “a hockey team sitting in a maple tree”.

The program will have varying levels of “success” depending on the images it has been trained on, the prompt given, and the algorithm.

There are a number of programs that produce these images, most famously DALL-E (and DALL-E 2) which was developed by Open AI (https://openai.com/), but also Dream (https://app.wombo.art/), Pixelz (https://pixelz.ai/), and others. There are differences in the levels of advancement, as well as the copyright provided to the end user for images the AI produces. It is worth noting that there are significant copyright issues resulting from these AI image generators. The software can be used to generate images that infringe on intellectual property, as well as produce harmful and negative images. As more and more artists experiment with this technology I expect that more issues will be uncovered and resolved. Some worthwhile articles on the issues with AI generated art can be found at Muddy Colors (https://www.muddycolors.com/2022/08/robots-vs-lawyers/) and TechCrunch (https://techcrunch.com/2022/07/22/commercial-image-generating-ai-raises-all-sorts-of-thorny-legal-issues/).

I have chosen to collaborate with the Midjourney AI for a few pieces, using it as a jumping off point for some prints and some paintings. As prompts I am using descriptions of work I have done or have planned and the medium I intend to work in. From the initial four images it produces I am choosing one, and then producing a work which follows from the result. In some cases, I will try multiple similar prompts with minor changes to see how it affects the output. My intention is to use it the software as inspiration, to get help me look beyond my first idea, and to look at ideas from a different perspective. I am asking the Midjourney AI to produce work in different mediums including ink, gouache, watercolour, linocut prints, and woodblock prints. In some cases, I am asking the algorithm to produce a realistic image and developing a painting from that, for example, asking it to produce a landscape and then painting that. I am also experimenting with different periods, styles and schools of art, for instance add qualifiers to the prompt such as gothic, academic, and 18th century British landscape.

The following are a few of the pieces that have emerged from this collaboration, so far: