Contents
I have my feet planted in two different worlds.
In my working life I’m driven by productivity, efficiency, and continuous improvement. Identify the target, streamline the process, eliminate the waste. In my creative life I change gears. I slow down. I navigate ideas. I idle within the moment.
Roads have represented efficiency, power, and control since Roman times. Reforming the landscape to dictate the shortest path and control access to spaces deemed worthy by commerce, politics, religion, and power. Freeways are similarly constructed for maximal productivity. They take precedence over our slower, unique, local ways.
Technology now takes us on a similar journey. We are driven to work faster and faster, to move from idea to execution instantaneously. We hand over our organic individuality to algorithms and machine intelligence.
Where will this new FreeWay lead?
Why did you become interested in this project?
I became interested in this project when thinking about how I need to split my time and attention across so many different responsibilities. I need to be careful of how much time I can commit to each so that I don't become overwhelmed. I made images of freeways from an outside perspective. Freeways were the symbol of society's pulse and rhythm that I can see but feel like I'm apart. They were a symbol for the relentless drive for productivity our society demands. So, I came into the project thinking about these constant streams of activity, feeling like I could see them speeding past me on all sides, but in order to fully commit to one, I would lose the others.
How and why did you incorporate AI ?
Through experimentation with text to image AI, Midjourney, Dall-E2 and Stable Diffusion, I came to think about how these new AIs would speed the production of image making. These AIs are trained on billions of images scraped from the web. Images are categorized, key-worded and uncatalogued. The AIs are taught to understand the content of the images, they then deconstruct the images and compare them to other similar images to understand the key content of each. They are taught to understand syntax of language, and how to then create entirely new images using the reference libraries of source images. Different AIs have been trained to provide different 'feels' and have slightly different 'abilities'. Midjourney is far more artistic and able to interpret loose concepts, whereas Dall-E2 is quite literal and photo-realistic.
I saw clear parallels between my thoughts around speed and efficiency in society and how these AIs may change image making. My project then became more about my feelings about our push for instantaneous outcomes.
What was your process?
I took my source images that had been shot in a minimal, graphic style. I photographed elements of freeways as if they were components in a machine - free of people, cars and any movement. My inspirations were the work of painter, Jeffrey Smart, architect and AI early adopter Matthew Kudless, and Finnish photographer Timo Lemmetti.
I then took these source images, randomly paired them and along with a text prompt 'Freeway to a Productive Utopia'. I used Midjourney due to the ability to use images as 'conceptual' prompts. I would receive four images back from the AI from this query. I would repeat the process to produce 6 new images. I would then repeat the process using these 6 images to create 3 new images, then 1 final image. I repeated this process 4 times to generated approximately 200 images. I was interested in how much my unique photographs would inform the generated images, and at what point the AI's understanding of the text prompt would take over.
What are your thoughts about AI in image making?
AI image making will continue to develop in terms of quality and application. It's inevitable that this will replace some level of creative endeavor particularly in the commercial world. Today they being used to rapidly generate concepts and ideas that can be used to stimulate work. They are already being introduced into stock photography libraries, they can generate images of people and products that don't exist. Entire classes of creatives in commercial industries will be replaced by AI, and new careers in prompt engineering will emerge.
It's possible that market dominance will lead to a concentration of applications with specific algorithms and source material. It could be that all creativity generated by these systems are via the proprietary technology of Meta, Amazon or Google. Along with all the limits of content moderation, parameters and source libraries that would come along with that.
It's also possible that AIs may become completely personalized to us, learning our preferences and styles as we develop, and becoming our own unique creative assistants free of any limits imposed by commercial interests.
What are your thoughts about intellectual property?
Will images generated by AI by the owned by the user entering their prompt/image, the programmer who devised the algorithms, the artists whose work has been 'sourced' to train the AI, or the company who owns the AI? Who owns the copyright or moral rights? Can the image be sold by the user, can the image be freely copied by another? If an AI designs a product, who owns the patent?
Can artists demand that their images are removed from the source material, and that references to their style or work cannot be used in prompts?
Our laws have not been designed with a artificial intelligence in mind. In some jurisdictions, copyright can only be owned by a human, in others patents have been granted to AI. Media has reported on artists winning competition with AI generated images.
There is a minefield that needs to be worked through around these issues, and the rate of development and change will outstrip our attempts to codify them.
The world has changed, and AI is now with us. Will our experience be like the introduction of the camera, or of photoshop, or of the moving image and give creatives new tools? Will AI commodify creative industries and spawn new industries to replace them?
Select any image to go to the online print store.
