top of page

Circuits of Power:
The Rise of the Robot Revolution

Stuart Marlin (Story,  image concepts and edits)
AI (initial descriptive text and initial images)


circuits of power.webp


Why:  A brief overview of the background to the collaboration.

I have been playing with AI as it related to photography for about a year, beginning with additions to Photoshop that allow generative fill (selecting a part of the image and filling it based on text descriptions).  This expanded to Firefly and Dall-e where complete image generation was possible.  Working with text has changed quickly over the past year with ChatGPT 4 now able to do many things (including write most of the assignments I set for my University students).  So exploring what it can and can't do is something I have been working on.  Everyone agrees, many jobs we now do will soon be replaced by AI.  Image creation is one that we have already seen occur as AI images make their way into advertising.

After creating my Robot Portraits, I decided to integrate them into a story. One where we have all been replaced but where the same types of motivations and conflict occur.  The easy part was describing the plot lines for each part to the AI.  It spat back a written text that was usually quite good, although a bit wordy (as AI just LOVES big words). Some back and forth resulted in fairly complete chapters and interestingly, because it was all done within 1 conversation thread, AI remembered key elements of the story and made links to previous events and came up with interesting descriptions that filled in the holes in the basic plot. But it often got things wrong and had to have more explicit prompting.

Each chapter was done in just a few hours (while watching TV in the evenings). Additional images were then created for the chapters.  This leads to another problem with AI at the moment.  It will never create the same thing twice.  You can't have a consistent character depicted in multiple images.  A can't do this yet. So some artistic licence was needed and some more explicit instructions about the viewpoint allowed these variations to be less critical.  In addition, some additional editing in Photoshop pushed the character features closer to the original portraits.

In the end, it is an interesting story with 90% of the plot coming from me and 90% of the text itself coming from the robot. This is the future of publishing... rapid creation of books generated by AI with a fairly uniform narrative style and complex use of language. Self publishing is already a thing and we will see more of this with a loos of the "editor" having fairly drastic effects on the continuity and spelling issues found in future novels.  But, given we consumers will be replaced by robots, perhaps this is what they will want to read.

The Audiobook version of chapter 1: AI synthesised voice
The Character Overview Video: AI synthesised voice synced with an Avatar using AI.
All images AI generated
The Story: Plot and some text by me.. most of the text by AI

bottom of page