My usage of and views on artificial intelligence

schizoidnightmares

Founder
Administrator
Messages
22
Reaction score
0
Points
106
Gender
Male
Since last year I've used AI "artwork" text-to-image models such as Stable Diffusion to produce placeholder images for Absurd Realms, particularly across the website. I am very consistent to label these AI-generated images as placeholders as they are not intended to be the final product. It is not my intention to replace artists (i.e. human beings), but merely to substitute them temporarily in lieu of quality artwork.

Flesh Run placeholder
I do not consider AI-generated "art" (I'll stop using quotes from here on out) really art. While I wager AI art holds potentially some artistic merits, its essential lack of human touch disqualifies it. What it lacks is a consistent, unique, and identifiable style with creative intent. No matter what you input, it doesn't really intuitively know what you want as an output. It has no artistic instincts. It is inferior in every way to a professional human artist, except in quantity and cost. Sometimes, it can take dozens, even hundreds, of generations to get a "good enough" result.

Still, it is superior than having no placeholder at all. It can provide something visually interesting, even if essentially soulless. I am not against AI for the use of substitution. I am against its use if intended to outright replace merited human artists, whom I consider irreplaceable.

While I do pay for access to AI tools, I would never pay an AI artist to generate an output for me. By paying for access to AI tools, I am just paying to use the processing power of a company's hardware, or in the case of custom models — software as well.

The image in this post is from the placeholder for "Flesh Run," the first short story in Schizoid Nightmares Anthology I. All of my AI-generated placeholders are released to the public domain because I do not consider such work worthy of any copyright protection. Not a single pixel was determined or altered by me in those generations, so why should I claim any credit for its creation? The best I could claim is the prompt used to generate it, or the cost of using the hardware/software behind it. If any credit is to be assigned it is to the training data used to create the text-to-image model and the developers behind the model.

ElevenLabs, which provides the most decent AI-generated voices (IMO), does not permit its customers to release AI-generations into the public domain or even under Creative Commons Attribution. With a commercial licence, they do allow its customers to use generations for commercial uses indefinitely, but this does not extend to others — making permissive licenses problematic. I even contacted them for clarity and they confirmed I cannot sublicense generations, even if I have a commercial licence. I don't agree with the ethics behind this policy, but it is what it is.

Whether or not a company (or individual) can actually claim ownership over entirely AI-generated work is highly suspect. I'm not convinced such a policy would hold up in a U.S. court, but I'm not willing to test these waters or run afoul of ElevenLabs as I still find their software useful for proofreading (reading off text).

I regularly use Grammarly to check for spelling, punctuation, wording/phrasing, and grammar suggestions. However, I do not believe in using AI to actually generate sentences for me. That would defeat the whole purpose of writing for me. I determine the flow and style of my stories, and I would not want anyone, or anything, taking that over for me. Maybe someone could create a text-to-text generative model based on my writings and try to "replace" my imagination. I would find that interesting, but beyond playing around with ideas, I would not find it useful (to clarify, I wouldn't be against others doing this — I release my work under permissive licensing after all).
 
Last edited:
I've cancelled my subscription with ElevenLabs today. Found out they depreciate voices. "Matthew" was by far my preferred proofreading voice, which they're depreciating this month. I could clone the voice, but it's not worth the hassle. Cancelling my subscription will save me quite a bit of moolah anyway and I've found it frustrating and arrogant on their part for a while that they claim copyright over their AI generations. My advice to anyone using the service in the future is to only use cloned voices from the beginning, so they cannot (presumably) phase those out.
 
I've started using Siri (Voice 4) as included in macOS to read over my writings for me. I've found it is quite decent, albeit inferior to the prior "Matthew" voice I used.
 
Back
Top