Polytropy

I have been working on a post that could have been the result of the following prompt:

Write on Gödel’s Incompleteness Theorem and AI, using such thinkers as

• Yuval Noah Harari (b. 1976),
• William Deresiewicz (b. 1964),
• Annie Dillard (b. 1945),
• Roger Penrose (b. 1931),
• Robert Pirsig (1928–2017),
• George Orwell (1903–50),
• E.B. White (1899–1985),
• Michael Attaleiates (c. 1022–80), and
• Plato (fl. 4th cent. b.c.e.).

Not until I had finished a first draft did I actually know that all of those people would feature. My real prompt had been more like,

In the style of David Pierce, write on Gödel’s Incompleteness Theorem and AI, as discussed by Roger Penrose in his “Précis of The Emperor’s New Mind.”

So instructed, could an LLM have come up with the connections that I did? Well, sure. Anything that has happened, could have happened, even in some other way. The real question is whether I would want AI to write my next post.

The present post consists of things I wanted to say at the beginning of that other post, after I had a first draft.

I happen to be on holiday in the usual place, described last September (2024) in

• “Tests” (for anxiety, or moral worth, or mathematical truth, or a true prophet, or knowledge of Euripides);
• “Motivated Reasoning in the Cyropaedia of Xenophon” (excessive pursuit of pleasure is a vice, and it would be a great pleasure to pursue and destroy the fleeing enemy; therefore, we should behave like cowards).

My wife and I have been reading and discussing, at her suggestion, chapter by chapter, one of the books that I talk about all the time. The author is on the list in the first prompt above.

The seminar has been very pleasant and useful for me. Though I have read Zen and the Art of Motorcycle Maintenance (1974) many times since high school, I have not got total recall. My thoughts have focused on certain passages, when others may be as worthy of note.

Now I want to note one passage, whose point I always thought was clear, but now I question it. Pirsig says in chapter 7,

All the time we are aware of millions of things around us – these changing shapes, these burning hills, the sound of the engine, the feel of the throttle, each rock and weed and fence post and piece of debris beside the road – aware of these things but not really conscious of them unless there is something unusual or unless they reflect something we are predisposed to see. We could not possibly be conscious of these things and remember all of them because our mind would be so full of useless details we would be unable to think.

Why could we not be conscious of everything? Is not God so conscious? Some people try to be so conscious; at least, there is a thing called “Cancel Culture OCD,”

whereby people fixate with terror on the prospect of their own cancellation.

This is according to Andrew Kay in “Shadow of a Doubt” (Harper’s, July 2025), reporting on the “twenty-eighth Annual OCD Conference in July 2023,” in San Francisco. A psychologist tells him of

an OCD patient who recorded every moment of their waking life with their phone, then watched to see if they’d done anything objectionable.

I don’t think that can be literally true, unless it means the person spends the first half of each day reviewing what they did in the second half of the previous day (or perhaps spends the first third of each day watching, at double speed, what they did in the last two thirds of the previous day). The person may also review in private only what is done in public.

I should like to be conscious of everything that I have written in this blog: to be omniscient about it, so to speak. Failing that, I want to be able to bring everything to memory without too much trouble.

Possibly this is what fellow Pirsig reader Ian Glendinning had in mind, when he wrote recently on his own blog, on August 24, 2025, about how AI

could maybe help me transform my blog content into a proper “graph” version of a Zettelkasten – like eg Obsidian.

I’m not sure exactly what he was hoping for. Apparently it was a diagram showing links between blog posts; however, would the links be those made with the <a> tag in the underlying html files, or would they be links in meaning, somehow inferred by AI?

It would be remarkable if AI could detect what I suggested in “A New Kind of Science.” I thought that Collingwood had been thinking about the concept of a criteriological science, even in his first book, Religion and Philosophy (1916), although he did not come up with a term for it until The Principles of Art (1937). Could AI have detected this common thread in Collingwood’s œuvre, without a common name for it? It seems unlikely.

Perhaps I use my own blog as a Zettelkasten, for what is apparently called “personal knowledge management.” I suggested this briefly in “Anthropology of Mathematics,” though without using the term “Zettelkasten,” which I did not know then.

Pirsig put his ideas on index cards and arranged them in stacks, kept horizontally in trays, in the old fashion of a card file, which apparently is what Zettelkasten means.

A post of mine is a stack of cards. These all belong together somehow, and I try to order them in a reasonable way. I hope to be readable by you, if you have patience at least. However, any topic can come up in any post, if I somehow see a connection. I may later forget what is in which post – for example, Collingwood’s notion of civilization, taken up in the post on Book VI of the Iliad in Chapman’s translation. If I have planned ahead, I may be able to recover such information by means of the “tags” and “categories” that WordPress lets one assign. Searching for specific words or phrases may also do the job. AI might help too, but at the cost of letting my memory become even worse than those other tools have already let it be.

John Warner writes about this problem in “The Labor of Labor” (August 31, 2025). When doctors let AI do cancer-detection for them, then their own skill at this atrophies. Warner cites the New York Times for this finding, then points out,

None of this is unexpected. I have a section on the risks of deskilling in More Than Words. When you outsource a skill to automation, that skill will diminish. The biggest danger for students using this technology as they’re in key stages of development is that they will never actually learn the skills we’re trying to teach in the first place …

The reality, though, is that deskilling isn’t just about the loss of discrete skills. Skills are just one part of what I call a “practice,” which also includes knowledge, attitudes, and habits-of-mind …

Becoming someone whose job it is to evaluate issues flagged by an AI when previously you were someone who did the entire task will alter your attitudes, narrow your knowledge, and reshape your habits-of-mind. You will become worse at your job.

Right now, I’m not sure what to do with the fourfold distinction, within a “practice,” of “Skills … knowledge, attitudes, and habits-of-mind.” I was talking about memory; where does that fit in?

Here I must bring in an essay that I saved last November, though it feels like ages ago. I recovered it by searching for “memory” in the titles of the files in the most likely directory. If that hadn’t worked, I would have searched my email correspondence with the friend to whom, I thought (correctly), I had mentioned the essay – which is “The Lost Art of Memory,” by Adam Robbert. It is no longer among the articles that make up Robbert’s Substack, “The Base Camp.” Neither did the Internet Archive save this article. Maybe Robbert thought he could put the same ideas better in another post, but I don’t know which one this might be. I see nothing to be ashamed of in what I saved. I guess you have to take my word that Robbert opened “The Lost Art of Memory” by saying,

I’ve been reading Mary Carruthers’s great work The Book of Memory, where she opens with a surprisingly basic insight, from which I think much else flows:

It’s a simple distinction in emphasis about the importance of memory or recollection, for the medieval and antique mind, versus the importance of imagination and creativity, in the modern one. Carruthers explores this difference by way of a comparison between St. Thomas Aquinas and Albert Einstein … she’s interested in some key differences between them. She writes: “… in the one case this process and product are ascribed to intuition and imagination unfettered by ‘definite’ tracks [Einstein], in the other to a ‘rich and retentive memory’ [Thomas], which never forgot anything and in which knowledge increased ‘as page is added to page in the writing of a book.’ ”

I think as a people who have almost unending access to notepads, note apps, and voice and video recording, it seems reasonable for us to be a people who have let memory slide in favor of creativity. As the need to keep things in mind – lucidly, in active memory, rather than statically, in media – decreases substantially, it makes sense that we would make this kind of shift. Think, for example, of how many scholars read from prepared papers, rather than speak from memory. Yet this convenience likely comes at a cost – we may be losing access to a fundamentally different way of thinking and perceiving, one that emerges only through a deeper form of engagement with memory.

This certainly deserves further contemplation, but I shall not engage in it specifically here. I do raise the question now of what AI does to imagination and creativity.

I once performed one small experiment with AI. I sought a natural example in English of the construction that I thought I had found in the Greek of Aristotle’s Poetics (§ X.2, 1452^a14): a relative clause that is itself complex, because it includes an absolute clause, and the subject of that clause is the relative pronoun that makes the whole complex clause relative. The translation here is mine:

λέγω δὲ ἁπλῆν μὲν πρᾶξιν	|	I call simple the action
ἧς γινομένης	|	which coming about
ὥσπερ ὥρισται	|	as has been defined
συνεχοῦς καὶ μιᾶς	|	continuous and one
ἄνευ περιπετείας	|	without reversal
ἢ ἀναγνωρισμοῦ	|	or recognition
ἡ μετάβασις γίνεται.	|	the change comes about.

I came upon this while editing “Biological History” recently. See Appendix I there. I was investigating Collingwood’s explanation for why one could not reduce history to a natural science, such as biology, meteorology, or geology: the reduction would be μετάβασις εἰς ἄλλο γένος, “change into another genus.” Tracking down some of Aristotle’s uses of metabasis led me to, among other places, the Poetics and my post “Purity Obscurity.” I could not find an existing English example of the grammatical construction that I wanted. The AI available through DuckDuckGo did not help.

I compose my own example:

Jabberwock was the beast, which being dead, he returned with its head.

This is inspired by two verses of Lewis Carroll’s poem:

He left it dead, and with its head
He went galumphing back.

I used also a French translation, which actually features an absolute construction:

La bête défaite, avec sa tête,
Il rentre gallomphant.

The translation was included by Douglas R. Hofstadter in Gödel, Escher, Bach (1979), which I read while in high school. Hofstadter has come up a few times on this blog, as for example in “Writing, Typography, and Nature,” which is connected to this one through its contemplation of what it means to write one of these posts in the first place. In sixth grade, “Jabberwocky” was one of the poems that Walter Green had us memorize regularly, for recitation in front of the class. I don’t think I made a point of memorizing the French translation when I saw it, but by then I was studying French anyway, and I must have impressed myself by being able to read French nonsense verse.