OK, OK, I get it—ChatGPT is indeed impressive in what it can do. But I must confess that I’m amused by all the people doing prompt engineering, that is, coming up with ever more elaborate ways to elicit certain types of output from ChatGPT. I can understand this when it comes to DALL-E, Midjourney, and Stable Diffusion. After all, one of the main selling points of those systems is that they enable the creation of art (of a sort) by people who are good with words but can’t paint or draw to save their lives.
But if you’re good enough with words to construct extremely complicated prompts for ChatGPT, why not devote some (or even all!) of that mental effort to figuring out what you want to say and how you want to say it. The part of writing that both requires the most attention and (in my opinion) is most rewarding is coming up with insightful ideas, apt metaphors and analogies, and stylish turns of phrase.
This post is an example of that. I had no need nor desire to have an AI expand an outline, or write it in the form of an academic paper, or create a limerick out of it. I just made a mental connection between the more elaborate forms of ChatGPT prompt engineering and Dustin Hoffman staying awake for multiple nights to prepare for his performance in the infamous dental torture scene in the movie Marathon Man—an extreme method technique that supposedly prompted his costar Laurence Olivier to ask, “My dear boy, why don’t you just try acting?” And the rest followed from that . . . .