AI Generated Art

Some thoughts on AI after reading Ted Chiang’s Why A.I. Isn’t Going to Make Art_

Art, like writing, involves a thousands of decisions at every stage of its production. Generative AI substitutes generic statistically average conclusions for each of those decisions. The result, while readable, is bland.

Some kinds of writing—business memos, especially the kind that require a quick summary of a meeting—sure, you can replace that with AI. But have you really replaced anything of importance? Maybe the meeting, or the email, should have been avoided entirely.

The quality of writing is proportional to the time it takes to create. Nothing is gained when an LLM converts your rough outline notes into long-form prose, only for the reader to apply his own LLM to summarize it back into an outline.

Just make better prompts! say the enthusiasts. And sure, with enough detailed prompts maybe that bland AI-generated novel could resemble something worth reading. But if you supply 100,000 words of prompts to get a 100,000 word novel, are you better off?

Consider my thesis, that ultimately AI-generated content will be prized for its clarity and readability and that human authors will need to compete based on their humanity. That’s true, in the sense that people will want to read authentic prose and that Wikipedia isn’t all there is to worthwhile writing.

I’ll push back against one of the author’s assertions, that language is communication. No, it’s coordination. My heartfelt Mothers Day card is not about communicating my love; it’s about coordinating with my mother so that we are both on the same page regarding my relationship with her.

An auto-complete algorithm might be able to communicate something to people, but it can’t coordinate anymore than the tail light on the car ahead of you is coordinating.

Quotes Worth Repeating

Many novelists have had the experience of being approached by someone convinced that they have a great idea for a novel, which they are willing to share in exchange for a fifty-fifty split of the proceeds. Such a person inadvertently reveals that they think formulating sentences is a nuisance rather than a fundamental part of storytelling in prose. Generative A.I. appeals to people who think they can express themselves in a medium without actually working in that medium.

As the linguist Emily M. Bender has noted, teachers don’t ask students to write essays because the world needs more student essays. The point of writing essays is to strengthen students’ critical-thinking skills; in the same way that lifting weights is useful no matter what sport an athlete plays

The computer scientist François Chollet has proposed the following distinction: skill is how well you perform at a task, while intelligence is how efficiently you gain new skills.

The task that generative A.I. has been most successful at is lowering our expectations, both of the things we read and of ourselves when we write anything for others to read. It is a fundamentally dehumanizing technology because it treats us as less than what we are: creators and apprehenders of meaning. It reduces the amount of intention in the world.