The Indirect Approach
AI for Poets
I rarely take the advice of AI. Rather, I find that my reactions to the answers to the questions I pose, whether positive, negative, or, as is usually the case, tangential, often provide me with the means of solving the problem at hand. Indeed, when I do this, I often use the same tools that I employed before I started to use large language models, albeit in a somewhat different way.
Last week, for example, I asked ChatGPT to help me reduce the results of an N-Gram query to numbers a humble historian might be able to manipulate. The robot offered me several options, all of which required programming skills well beyond my ken. (My last foray into the world of coding, which I made in 1984, took the form of a ten-line program written, in BASIC, on a Commodore 64.) Thus, rather than attempting to wrestle with Python, I enlisted the aid of a faithful friend I have worked with for more than a decade, the advanced search page on the website of the Hathi Trust.
Now, one may think that I could have saved myself a bit of time and trouble by skipping the middleman. After all, I was already wise in the ways of the Pachydermal Prince of content queries. (Can you tell how much I love that service?) My mind, alas, does not work that way.
If someone had asked me to explain both the Hathi website and the N-Gram Viewer, I would have told him that both sat atop the mountain of digitized volumes accumulated by Google. However, while I was trying to make sense of the tiny numbers (such as 0.0072213953%) that I found on one of the curves on my screen, this foundational fact fell through a crack in my consciousness.
So, there you have it, I made trillions of electrons dance to obtain the sort of service that, in the twilight years of the Analog Age, would have been provided by a deck of creativity cards (such as the Whack Pack) or, a few years before that, a brainstorming session led by a Californian in a leisure suit.
