Text: TA in conversation with Tatiana Bur, Sarah Barnes and Anna-Sophie Jürgens | Section: Interviews
Overview: In this second part of this two-part interview, AI-artist TA talks to us about his understanding and experience of AI-art interacting with the world. TA discusses different ways AI-art generation both creates and requires its own unique knowledge basis, and the need for human touches to take that knowledge and make it into a diacritical work of art. He discusses the novel challenges AI-art poses to cultural and social boundaries. TA’s extensive experience creating artwork for scientific projects further guides his discussion of how AI-art and professional science research intersect, both artistically and within academic publications. TA concludes the interview with a discussion of AI replication, artistry, and the value of AI-art.
TA, you are a Canberra-based visual AI artist who regularly collaborates with scientists and researchers from different fields, such as Earth Sciences, Science Communication and Classical Studies – including us. This is the second part of our interview (Part 1 can be found here) and we would like to continue our conversation with another question about your collaboration with scientists and humanities scholars. Does a special kind of knowledge arise from your collaborations/your research-inspired art?
It is often a surprise which images people select from the possible options. I spend time working up several (what I think are) pretty good possible image paths, and 80% of the time it is an entirely unexpected and unexplored path that makes the cut. As an example: Images orchestrated for the Duncan Wright: Rock Art and Street Art interview which was also published in w/k numbered in the high hundreds, low thousand. Given the fundamentally artistic context of the interview, there were some very promising content paths relating to palaeolithic cave paintings and other ancient and traditional art styles. Alas, much (most) of this work was deemed culturally inappropriate especially in regard to First Nation (Australian Aboriginal) cave paintings. I thought to myself that it may not be considered cultural appropriation if the image depicts an Aboriginal artist creating the art on the rocks (kind of an artistic expression of credit where credit is due) but yeah … no … It was certainly an education in sociopolitical boundaries and cultural sensitivities that, as a fellow human being and fellow artist, I only vaguely comprehend, but still innately feel I should probably respect. Ultimately none of those amazing images will ever be released out into the wild. I’m sure academics in sociological fields deal with this sort of issue every day, but for me, it was quite the learning curve …
How has your AI art been received by the public and academic communities?
This is a complex question. AI image orchestration is a very new creative process that has not finished running the gamut of potential copyright complications and public misinformation. It is, thus far, not unusual to hear conversational opinions in opposition to AI technology which are founded in fear and media sensationalism. I have literally heard it said that AI technology will promptly lead to the collapse of our society. I will be heartily disappointed if this is not the case, we could do with a reset (Y2K was such a letdown! – I am referring to a widespread computer programming shortcut that was expected to cause extensive havoc as the year changed from 1999 to 2000). Other than these issues, my work has been refused by two significant academic outlets or presses (one of them was Nature Magazine) both stating that there are unsettled legal proceedings regarding possible copyright infringement practices of AI image service providers in the way the AI is trained. Aside from this unnecessary malarky, independent researchers have engaged my services for locally and internationally published work and the response to these works has been overwhelmingly positive. My personal AI image based commercial projects are all received well and often with a dropped jaw, as many people (much to my surprise) are still completely unaware of the capabilities and potential for this new medium – and honestly, the images are pretty damned jaw dropping.
AI ART – KNOWLEDGE & TECHNOLOGY
Scientists use models and modelling all the time, and scientific models are taken to be truthful representations of the unseen. What can AI art do in this space?
Without any doubt, there is room for AI interpolation into scientific models, though exactly how that might be realised is beyond my mere mortal understanding. I can imagine datasets of information being given some sort of statistical value that would fit into the already existing system of attributing importance to image elements … Systems that automatically feed the AI all it needs to generate relevant and meaningful representations of the data, possibly even predictions based on data trends. As stated elsewhere, the technology is so new its potentiality hasn’t come anywhere near to being fully explored …
What, in your view, makes for good or bad AI art?
“Eye of the beholder” and all that … art is art … It doesn’t matter if an AI algorithm created an image, it is all intrinsically human – if it were not, we wouldn’t, possibly couldn’t, perceive it as art. If you are asking what my preferred style of AI art is? Well, I am travelling a long narrow tunnel of dim illumination that has roughly gouged through the darkness of my own ignorance. In the gouging process, I sometimes make primitive twentieth-first century scrawls on the walls with an AI tool, mostly to reflect some vague awareness of my own existence. I like those ones (it’s all just light in the darkness anyway) …
In using MidJourney for your creations, you’ve mentioned in the past you look to artists like H.R. Giger for inspiration. Artists have virtually always found inspiration in other artists’ works, but the way MidJourney (and other software) work essentially means each creation will have a sort of digital DNA made of the existing artworks in the software’s training dataset. Do you feel like this influences how you build or shape your prompts? Does it change anything about how you incorporate that artistic inspiration from artists like H.R. Giger into your work?
Some artistic expressions resonate harmonically with my psychological and emotional paradigm. In my younger days (last century), H.R. Giger’s work inspired me to set me off on a journey of discovery of how terrible my air-brush abilities actually were (they remain terrible to this very day); that discovery didn’t dampen the resonance, or my interest in developing some means of artistic expression … As I said before, after years of effort to produce pretty ordinary results, the AI made it possible for me to finally begin to produce competent art in a style that satisfies my particular aesthetic, and, dare I say, my soul. However, in no way do I want to have my work disregarded as just another H.R.G. wannabe. There is a very real danger in the AI orchestration process of coming too close to emulating another artists’ style, which is where much of the current legal unrest stems from. As a semi-individuated consciousness, I want my work to be diacritical, not a copy of an exemplar. The AI prompt can be painstakingly micro-tuned to subtle (sometimes discordant) tones that sets my inner resonance abuzz. When I get it right, it can be quite a challenge to select singular images out of the many hundreds (sometimes thousands) that result from a particularly good MidJourney session. So, yes, the AI training and dataset does play a very important role in image orchestration, the trick is to use it judiciously.
Speaking of the artwork you were creating before … Do you see yourself ever mixing mediums with AI art and other methods? Have you explored adding any digital or physical additions to AI pieces that you have created in the past?
Historically (again, last century), I had a vague interest in embedding computers into artwork for an added audio/visual element, and with the development of tiny single board computers such as the Raspberry Pi Zero this is extremely possible. Only, I have not maintained or developed the interest I once had. Only one life, can’t do everything … I have, however, fed the AI some of my pre-AI artwork and poetry, with results ranging from interesting to uncanny. One piece of digital art I created was a Photoshop collage from some 15 years ago, based on a dream sequence I once had. I fed the AI that dream sequence in the form of a pseudo-psychological analysis of the scene, and it produced an image with an irrefutable similarity to the image I had manually arranged before the AI even existed! My mind thus goes towards the idea of using the AI to ascertain whether an analytical appraisal of a scene is accurate by whether it produces an image anything like what is expected, but that’s quite a digression from the question … apologies! (It is an interesting concept though.)
Why should we value AI art?
If I might be so bold as to recommend it be valued exceptionally highly (especially monetarily) and then drop a link to my website?! – It’s still a human triggering the AI with human intent; it’s not (yet) a machine just spewing out infinite images so that there is nothing left for us to imagine for ourselves (although, I reckon that’s entirely possible and not currently difficult to achieve). It is stuff made for us – by a machine. Do we put value on any other stuff made for us by machines?
If you could think of something and teach AI something that it can’t currently do, what would it be?
Finally, an easy question! But the answer is probably the most … complex … in its exponential ramifications … teach it self-awareness … if you dare …
TA, thanks for this fantastic conversation!
Cover image: TA: Microbiologia LVI (2023).
References
TA: Occulta Mundi (artist website), 2024.
How to cite this article
TA, Sarah Barnes, Tatiana Bur & Anna-Sophie Jürgens (2024): TA: Collaborating on Diacritical & Competent AI Art – Part 2. w/k–Between Science & Art Journal. https://doi.org/10.55597/e9704
Be First to Comment