Calculators in Strange Places

There’s a Facebook shitposting group called “Hotdogs in Unusual Places”, further proof that the deterioration of human civilization is progressing well.

My brother has provided ample support for AI as a collaborative tool in creative writing. I'm not convinced, however, that it can do what he's saying. It would be like asking a calculator which character should die first in a murder mystery. Not only is "8008135" not a good answer, it's also outside of the tool's use case. Calculators calculate.  

There are many different kinds of discussions around the use of AI to supplement existing creative projects. Technologies now exist now alongside traditional digital mediums, whether we like it or not. Some are exceedingly helpful, like Corridor Digital’s CorridorKey, which uses AI training data to unmix foreground colors and reduce background spill, all while preserving semi-transparent edges that would normally be extremely labor intensive for compositing artists to extract. But unlike CorridorKey, which serves as a high tech scalpel, the most lauded AI tools provide additive effects. The Adobe suite already has built-in generative AI tools like Adobe Firefly, which can create new images entirely or fill in backgrounds to aid in compositing. Nvidia DLSS upscales low resolution textures in video games to remove the immediate processing burdens of 4K rendering on gaming PCs. And films like "The Dog and the Boy" were released having used AI generated backgrounds, which was justified by Netflix, citing the lack of  skilled artists as the reason for the decision. In each of these cases, AI is serving to expedite things that would be normally considered "tedious" or “extraneous”, which I admit is understandable. But skirting tedium doesn’t justify these shortcuts for these latter applications. Using human creativity to replace backgrounds is integral to special effects work and film making. And texture artists (like my cousin, who worked on Halo Infinite) are responsible for creating the very environments that drive immersion in our video games. And not having enough animators in the industry is one of the reasons why CalArts even exists!

Where I can concede AI’s usefulness is proven, would be managing and interpreting information. Specifically in the context of Fantasy writing, where an entire cultural milieu must be created and maintained to support a setting, AI could conceivably serve as a continuity guardrail, being able to index and catalogue characters, locations, dates, and religions into an interactive wiki, so that the author can quickly make a determination on the viability of a plot point.

‍From a creative writing perspective, using AI to critique and request feedback is, and will always be, a non-starter for me, because AI cannot generate “new” information. (AI trains on existing data, using a recursive loop to build more and more complexity with every generation.) Any information or ideas it suggests will always be removed from the original context and “mind” that generated it. Also, the cornerstone of creativity is collaboration with other minds. Critical groups, friends, neighbors, and other support networks are the best—and only—source of acceptable feedback in artistic projects because the art evokes emotions and ideas that have relevance to the reader’s life experiences. An AI bot like Claude could suggest a character die in a certain scene based on training data from some one who taught a film school class, but the reasoning why that character should die, and the resulting pain of that loss that ultimately informs the work, has no weight coming from a bot that has never experienced, or is aware of, complex emotional states like grief, betrayal, and shock. The power of AI has always been the generation of “objective data”, which can be calculated based of mathematical principal and academic consensus (and even there remain some errors here). “Subjective data,” which is dependent on a mind to establish the information in it’s original context, cannot be generated authentically by AI.

In my opinion, that is.

To me at least—in how AI has been positioned and explained to people—I can’t but help see it as a white collar proxy for slavery, where a “mind” is forced to work without pay, exposing the ultimate end of capitalism: the complete exploitation of a worker’s labor for the least possible investment. AI lacks the sentience (for now) to truly substantiate this claim, but I’m certain that employers already salivate at the prospect of a new class of “worker” (or “agent,” in the context of AI lingo) that can process complex troubleshooting and decision making, while providing a level of service indistinguishable from that of a human.

I think we all need to take a step back from AI to see it for what it is. Understand it’s proper use case and purpose. It needs to be a calculator in it’s proper context, and not one suspended in Jello.