I Don't Want AI's "Help"

The big AI story the last couple weeks has been about the doomsayers and their assessment that this awesome new tool might actually destroy human life in ten or maybe 20 years. They place the chances at something like five or ten percent. Well, that certainly gives you something to think about when you are reading your favorites sites and eating your yogurt! Check last night's scores and contemplate the impending doom of all you have ever loved and held dear. If they said you have a ten percent chance of getting into a horrible accident if you drive to the grocery store, well, I certainly would not do that. Yet, here we are. Somehow, I think that they are seriously exaggerating about the threat, but I have no actual reason for thinking that. But I do, and I don't think it's just a coping mechanism. 

That said, what I'm here to talk about is the second biggest AI story last week, which was that AI solved a mathematics problem of the very highest order, one which had stumped our best mathematicians of the last decade (including Matt Damon). Apparently it took 88 hours, which, frankly, strikes me as a long time in AI minutes, but which nevertheless is nigh miraculous. Hooray? The machine did what no human could do and will now save us a lot of time. But to me, it is a loss. What a glorious thing for humans to have done, to spend their lives in contemplation and reasoning and dialogue in pursuit of uncovering a truth or code that seems to be inherent in the DNA of existence as we know it. Didn't they make a big movie about a mathematician? Yes, A Beautiful Mind I think it was. There will be no movie about this "stunning" breakthrough. It is completely and utterly boring, no? About as exciting as using a calculator to crunch interest rates. 

My own area of interest is writing, and AI can effortlessly produce articles and essays that are pretty damn good. Better than most people can do I suppose. The value in this however is only there if the essay or article is primarily viewed as information. Unique style and voice are essential aspects of great writing. When I refer to someone as a great prose stylist that is the highest praise. Right now AI "style" is good enough but not always great. Maybe it's predictable in its goodness? Yes, I know it will keep improving, and rapidly at that. But do I want to read a machine that writes as well as Christopher Hitchens? No, I want to read Christopher Hitchens. I guess because I know that how great his achievement was in human terms. It's just no fun, is what I'm saying, I guess. 

Currently, many writers are incorporating AI into their own practices. But aside from research I don't know why I would want to do that. I suppose I could give AI a bunch of prompts and it could produce a mini-essay very close to this. But again, why would I want to do that? I like writing, and consider it a valuable and engaging way to use my mind and consciousness -- as do (did?) the aforementioned mathematicians. What will I do with my thoughts if I'm not writing? Well, I can read, and I do read constantly, but it still amounts to consumption. Healthier, yes, than consuming an endless streaming of TikTok videos. But I need to get a dose of production each day. I lose a lot by not using my mind, and my heart I should add, as do when writing. AI conceivably doing it better than I can is an irrelevant factor. 

There is something else that should be considered when supposing that constructing an essay through finely considered prompts can fully replicate and succeed at "human" writing. When you put your prompts in you don't know what you don't know, and you don't yet really know what you want to say. In fact, this idea and this paragraph was not even in my thoughts before I started writing this. You discover what you want to say through the process of writing. Indeed, you usually don't know in any depth or detail what your ideas even are until you try to write them. Printed words are harsh taskmasters, not disinclined to bring you down a peg. 

I have had fun running my essays and poems past ChatGPT, and the sophistication of "their" feedback is truly impressive. I read what they say and think, "They get me, they really do!" I would say they are astute in their observations. So like I said, it's fun. But I have rarely accepted their editorial suggestions. They often seem arbitrary, like "I really should suggest some changes here, so here's some." And as the writer/recipient, I often feel I should take their advice. But whenever I push back and explain why something is a certain way, ChatGPT will always say, yes, that makes sense. If you know your own mind and intentions as I, at my advanced age do, then it works out fine. But young people may be more susceptible and AI, if relied on too heavily, might short circuit their growth process. 

One other thing I have noticed. I use a lot of implicit references of language and thought in my poems, and ChatGPT frequently doesn't get them at all. They pick up Biblical allusions well, but other stuff not so much. For example, I have one poem that uses language from both Bertold Brecht and Tennessee Ernie Ford, a juxtaposition I'm proud of, but ChatGPT didn't notice. Any culturally literate editor certainly would have, since the references are not obscure. Will it get better? Sure, but it will never know what it doesn't know. It doesn't "think" on the level. It can't surprise itself.

Finally, there is the matter of the value of imperfection. Much of what we appreciate in our friends and companions is a result of their idiosyncrasies, which no machine can replicate. Imperfection adds texture, but the kinds of imperfections that appear in AI are little more than mistakes. There is no charming way they mispronounce something or keep forgetting something. As much as it may seem otherwise, AI doesn't possess character. I don't know, maybe I'm wrong about this, since AI is developing exponentially. 

So I really don't see myself using AI much as a writing tool. I would not have wanted this writing process to be either faster or anything other than what it was. And I'll always want to see the byline of human who no had to struggle and push through to say what they wanted to say. Who indeed had to struggle in the process of living live itself to even have something to say. AI doesn't struggle. During those 88 hours need to solve that math problem, they never had to break for coffee or cigarettes and to call someone and vent. The process was frictionless, but I remain a fan of friction. It's called being alive.