Truck with tags with AI-generated typography
There has been a lot of talk lately about Artificial Intelligence and how advances in this technological field are beginning to revolutionize the world. Platforms like DALL·E or Midjourney, which are capable of generating any type of image by following only a description by words, or ChatGPT, which manages all the information at its disposal to find answers to our questions and configure all types of texts, will represent a big change, especially in the creative world.
How can this affect the field of graffiti and art? We chatted a bit with the trasgraffiti artist Alsino Skowronnek, specialized in the matter.
We take this opportunity to invite you to the opening of Excavation Trail, the next exhibition by Alsino Skowronek in Barcelona, on January 24 at Piramidón Centre D’art Contemporani.
How would you summarize, for someone who is not familiar with the concept, what is AI, and what does this technology mean in the field of image generation?
For me, Artificial Intelligence is a broad term for a set of methods that teach computers to do things and exhibit what we would call “intelligent” behavior. It is essentially learning from data. Instead of programming a computer based on a pre-defined recipe to accomplish a goal (e.g. predicting the price of an apartment by a formula or model), the computer figures out itself what the relevant characteristics of the data are and applies a set of self-learned rules to solve the task, e.g. it recognizes that there is a correlation between the price of the apartment and its size and the number of bedrooms, the location, etc.
“People often associate “Artificial Intelligence” with robots, flying cars and the end of the human free will, which is yet to happen.”
How do you get a machine to do that?
This knowledge is derived from usually thousands of data points. AI essentially means that computers have gained the ability to “learn” from data and develop their own logical systems based on which they can report a result. This is why, in general, I prefer to use the term “Machine Learning” rather than “Artificial Intelligence”. The latter sometimes obfuscates the actual mechanics at work and has more of a sci-fi connotation to it. People often associate “Artificial Intelligence” with robots, flying cars and the end of the human free will, which is yet to happen.
Are there different applications of AI or as you call it “machine learning”?
Of course, there are a lot of different flavors and models in Machine Learning, each suited for a different type of setting: From data analytics to generative models that create different types of media. Recently, there has been a huge push in capacity of “generative models” such as ChatGPT, Dalle-2, Midjourney and Stable Diffusion to synthesize all different types of media, from text to images to videos. These new models offer exciting new capabilities to work with media and generate new content.
“these technologies (…) make it so easy to create content by simply describing it with a simple text prompt or by recycling old data.”
What is the future of all this?
As we are currently seeing, these technologies are quickly being adopted in many fields, especially where images are important, such as design, advertising, architecture and art, because they make it so easy to create content by simply describing it with a simple text prompt or by recycling old data. The important discussions however will be, how these models are trained, controlled, used and which power dynamics are at play here. As these models are trained on “real data”, this means that their output is directly related to somebody’s work in the real world. The next few years will bring a lot of important discussions to the table of how we will live and govern these new technologies.
What role does AI play in your work?
For me, AI (or Machine Learning) represents a set of new tools to work with in my practice, especially in the creation of new paintings. It allows me to “paint with data”, i.e. I can use already existing imagery and bend and combine it to my own needs until I get what feels right. The new generative models are a fun way to try out new ideas and sketch prototypes, but for me, they are always only the starting point to my work, hardly ever the end product.
“A few years ago I started using them to generate new graffiti tags or new abstract train panels and pieces, based on image material from the “Dirty Hands 2” movie or old NYC oldschool writers.”
What does AI offer you that you can’t get through other processes?
AI to me offers interesting aesthetics that are very inspiring, because they allow you to look into the inside of a machine. How does a computer make sense of an image, and how does it “see” it? What features does it learn when generating content, and what does it not account for? These questions also make you think about the way we learn as humans and how we derive meaning from our own experience (or data). What are the similarities, and what are the differences? AI models are very interesting in that sense, as they offer you a small window into the human brain. They are also just a lot of fun to play around with, especially when you tweak them in ways that they were not intended for.
Can you give an example?
A few years ago I started using them to generate new graffiti tags or new abstract train panels and pieces, based on image material from the “Dirty Hands 2” movie or old NYC oldschool writers. I am especially interested in the in-between output of the models, where the machine has already started to recognize certain features in an image, but is still not fully making sense of things. It somehow reminds me of how we sometimes cannot completely comprehend things by observing them, but still derive great value from them through inspiration. The use of machine learning in my work is also a nice playground for me to dive deeper further into the underlying technologies and to start discussions with the outside world about the different technological and ethical discussions surrounding AI.
How do you think AI will affect the art world?
I think artists have always been at the fore of using and abusing frontier technologies, and ask questions about our relationships with them. So on the one hand, AI can be seen as just another material category or tool for artists to explore and transmit their own observations about the world.
“I think there is a lot to question about this new age of AI, and art will hopefully make a contribution in questioning these issues.”
What changes will we see?
As such, we will probably see more and more artwork that has been created using these technologies. This will become very common, however, and it will probably lose the “wow-factor” quite soon as we are accepting the computers’ new capabilities. I think most artists are already realizing that it is possible to create previously unimaginable things with models such as Dalle-2 and the effect on their ability to express their own points of view about the world through these new technologies. This new reality will certainly shift the boundary of what we consider “magic” further out. However, in the long term, this in itself will not be enough in the arts.
What role will artists play then?
The subject and message of art still lies with the artists. Therefore, on the other hand, I think that AI will become increasingly the subject of art itself: We are living in an accelerating digital age, where computers are taking on more and important roles. We are quickly passing a point-of-no-return where we are accepting that machine learning models are making important decisions about our health, wellbeing and social connections amongst others. And we do not necessarily understand how these decision-making processes work. Besides that, much of the AI power is accumulating in the hands of a few big companies and used without our consent. I think there is a lot to question about this new age of AI, and art will hopefully make a contribution in questioning these issues.
“If done right, it can essentially help to question common patterns and conventions of writing, both on the structural and on the content dimension.”
How do you think AI will affect the world of graffiti?
Graffiti in my experience is a very slowly changing field. Since its inception in the late 1970s, not much has changed. Of course, today people are using better spray paint, caps and other hardware, but writers are essentially doing the same thing as before and not much has changed on the “software” side of things. So in that sense graffiti is a pretty traditionalist phenomenon I would argue, with sprinkles of contemporary culture here and there. Nevertheless, I do think that AI can make an interesting contribution in the field: If done right, it can essentially help to question common patterns and conventions of writing, both on the structural and on the content dimension. In the best scenario, this can trigger an “unlearning” and subsequent “re-learning” of the basic elements of writing: letters.
What does this mean?
If a writer really digs deep, this could lead to completely new and unexpected results in letter shapes and style trajectories. But this requires the individual writer to work out her own process in incorporating AI into her own style development. Together with a couple of good people, I am currently working on an interesting project in this regard, which is essentially a research process into how AI can be used in a graffiti setting. More on this soon. Stay tuned.
»Excavation trail« is the first solo show of Alsino Skowronnek in Spain. It shows a selection of paintings and digital work, inspired by both the aesthetics of Artificial Intelligence (AI) as well as codes found in the symbolic languages of graffiti. AI generated content does not suffice as the final output, but is merely a transitionary tool to decouple image structure from content, freely associate and recombine it with textual fragments born out of the lived experience of the artist. Alsino’s works establish a direct connection between the interfaces and data of the digital plane and the more analogical plastic arts. As a result of this dialogue, an abstraction is born that reflects on learning processes and shifting patterns in human-computer interaction.
Tuesday, 24 January 2023
From 19h – 22h
Piramidón Centre d’Art Contemporani
Carrer del Concili de Trento, 313, 08020 Barcelona