Hello Cyber Cats, its yours truly the Script Kitty, Killer Kat and tonight we’re going to talk about AI image generation and the world around the CyberKat Cafe.
It’s no secret that I’ve been talking a lot about AI on the show, even before the recent surge of activity. AI is going to change a lot, I wont claim to know exactly how fast or how much but its going to have a lasting impact on our society. So lets take a brief look at the AI image generator Midjourney and also expand on the setting of the show.
As long time listeners will know the Cyberkat Cafe takes place in the not so distant cyberpunk future. Killer Kat the notorious hacker broadcasts the show out of an abandoned radio station on the outskirts of Midnight City. Before you accuse me of copying Cyberpunk 2077 I will remind you that the seminal book Neuromancer by William Gibson uses the name first, which was then copied by the Cyberpunk tabletop roll playing game which was then copied by Cyberpunk 2077. But I digress, Midnight city is a dystopian mega city named after the fact that in order to counteract the greenhouse effect caused by excessive carbon emissions sun-reflecting aerosols are released into the sky above the city blanketing it in eternal night.
Killer Kat (Again yours truly) is seen to be wearing dark clothing and a robotic cat mask. I fed prompts into the Midjourney discord bot and it mostly produced mangled images, were I to take a guess as to why I would assume its the fault of Furry art. A genre of art in which the focus is on anthropomorphic animals, due to the cat mask being a central part of the character and the name Killer Kat the AI seems to try and blend cat features into the image to rather horrifying results.
Below are the three best results I was able to generate, take a look for yourself and you can see the influence of the word cat/Kat.
This speaks to a larger problem with current AI technologies inability to recognize the importance of words inside a prompt. While prompt engineering can be used to somewhat mitigate these issues it cannot solve the problem on its own.
Interestingly the issue of copyright around these images has started to heat up as well. Recently the copyright protection for AI generated images used in a graphic novel have been revoked (https://www.reuters.com/legal/ai-created-images-lose-us-copyrights-test-new-technology-2023-02-22/) showing a possible trend towards similar decisions in the future. Due to the fact that the majority of the images used to train the algorithms were taken without the consent of the artists or more importantly licensing fees being paid to the artists. I do not think it is reasonable to allow copyright for AI generated images.
If you want to hear more on this topic let me know and I’ll cover it on a future episode of the Podcast, and let me know in the comments what you think about copyrighting AI generated works. Stay safe out there and until next time remember, Art isn’t real it cant hurt you (yet), This is Killer Kat signing off!