In the cracks
Well, that didn’t take long.
This week, on a post of mine, I received a comment that was unmistakably generated by ChatGPT.
It had all of the hallmarks of ChatGPT’s writing style, or lack thereof: Identical structure between paragraphs, restating the words of the prompt verbatim in each topic sentence, and over-reliance on phrases like, “It’s important to remember that...” and “find what works best for you.”
But the thing that struck me most about encountering ChatGPT’s work in the wild was its complete lack of a point of view.
And that, I believe, will be the thing that separates useful applications of ChatGPT from useless ones. If you’re not going to present a point of view, why bother writing anything in the first place?
As ChatGPT put it, “Presenting a point of view in writing is important because it allows the author to convey their perspective, ideas, and beliefs on a topic.”
Mm-hmm. Presenting a point of view gives the writing a point of view. Got it, thank you! I’m having flashbacks to teaching middle school language arts.
So much of what I write would be impossible for ChatGPT to contribute to, at least given my current understanding of its capabilities along with my own capabilities in getting it to do stuff. I write to figure out what I think, and I often find the topics that I’m most excited about “in the cracks” between established lines of thought.
ChatGPT is ideally suited to share those established lines of thought — to express and build upon conventional wisdom. The ideas I’m interested in exploring seem to be invisible to it.
Of course, it could be a failure of my own imagination. I have the same trouble when it comes to implementing good SEO. The search terms that draw people to my website are completely random, and I don’t know how I would change that when I’m always trying to construct a juicy analogy that can be extended in various ways or tell a story that has metaphorical resonances. When I’m writing about sailing or fish sticks, I’m not really writing about sailing or fish sticks — but Google, using its own artificial intelligence, can’t always tell.
Nevertheless, I’m intrigued by the possibilities of AI. Maybe, with practice, I can go beyond the superficial and find the “cracks” that exist in and around the accepted and obvious uses of these tools.
From my limited experimentation with AI and recent encounter with an egregious use of ChatGPT’s output, it’s clear to me that, as the human involved, my role in the collaboration is essential. Garbage in, garbage out, as the saying goes. If I’m not offering a fresh idea, AI certainly isn’t going to come up with one.
As with any new tool, the opportunity is in figuring out how to use it as more than just a toy or a novelty. That’s a test of ingenuity and forward thinking. That’s the work is quietly brewing in various pockets of the world today that will someday be accepted and obvious. Copy-pasting AI output is the equivalent of getting your 1980s Casio calculator to say “BOOBLESS.” That’s a start, but there’s more to be found. The good stuff still lives in the cracks, there for us human beings to discover.