Dicitque mihi mea pagina, fur es. [My page cries out to me, You are a thief.]

If that severe doom of Synesius be true, “it is a greater offence to steal dead men’s labours, than their clothes,” what shall become of most writers?

Synesius was a Greek-Libyan neoplatonist in the 4th c. AD.

Burton was a scholar, and a well-read one at that. He was clear on the fact that little can be said that was not said once before; his liberal verbatim quotes testify to that. And as he also writes, he is his own worst critic: he admits from the start that all he is doing is recycling those he has read, and that this is realistically all any writer can expect to do.

Synesius was not referring to learning from someone and putting their knowledge to work, though. He meant that we must not take credit for something that we did not achieve alone — a dead man has little use for shoes, but a contribution to the sum of human knowledge ought to remain his forever.

Writers must charge themselves with plagiarism before they are charged more grievously by others. Only by identifying one’s own thefts can one put things back in an orderly way.

I have only this of Macrobius to say for myself, Omne meum, nihil meum, ’tis all mine, and none mine.

As apothecaries we make new mixtures everyday, pour out of one vessel into another… we skim off the cream of other men’s wits, pick the choice flowers of their tilled gardens to set out our own sterile plots.

As a good housewife out of divers fleeces weaves one piece of cloth, a bee gathers wax and honey out of many flowers, and makes a new bundle of all.

Floriferis ut apes in saltibus omnia libant [As bees in flowery glades sip from each cup], I have laboriously collected this cento out of divers writers, and that sine injuria [without harming them], I have wronged no authors, but given every man his own.

We can say nothing but what hath been said, the composition and method is ours only, and shows a scholar.

(Incidentally, Burton’s flair for repetition is on full view here — he loves to offer multiple metaphors or examples, not necessarily the better to communicate it, but (I suspect) because he simply enjoyed the way each example differently expresses the same idea. It was the floriferis ut apes that stuck with me.)

Textual notes aside, here Burton’s criticism of sterile writing and thinking is highly apropos to today’s AI-assisted composition and inquiry.

It is one thing to draw from the sources around you — there is seldom any other option, as truly original ideas are rare to begin with. But the result of collecting all that pollen and nectar must be honey that is sweet in its own way.

What AI subtracts from this is first, the method of sampling. If current systems are even capable of reliably seeking individual sources and crediting them properly, it does not follow that many, if any, pursue this option. The pitch, after all, is speed and convenience, not comprehensiveness and careful delineation.

To receive a summary of a topic is useful, to be sure. But it abandons rigor. Few who use AI to do this type of work will delve beyond that summary, meaning, in the first place, they don’t know if it’s accurate; and in the second, fundamentally they won’t know who said what or why, because that type of context is superfluous to the surface-level understanding of the topic they seek.

In other words, they are outsourcing to a random bee the task of traversing the flowery glade. For actual bees this type of division of labor makes sense. For someone trying to understand something, it skips the crucial step of first understanding what it is that you are trying to understand. As we will examine elsewhere, one has not learned to learn.

This leads the next step into twin peril.

If the person attempts to write (or, as the case may be, do a podcast, or an interview) about the topic at issue, they are likely not going to be speaking from a position of authority, information, or arguably even genuine interest.

Not to suggest that someone dabbling in a topic via AI is faking in it, but I do think there is a difference between wanting to learn about something, and doing the minimum to appear knowledgeable about it. The latter is not always the worse, but it is performative; I often had to do it myself as a reporter in order to establish rapport with a real expert. (This predated AI — I would just skim a few obvious sources.)

Lacking familiarity or real interest in the source material — which is to say, the material that matters — means what they express will only reflect the most palatable, simplified interpretation of it. Distilling that is something AI is very good at. But by definition if you don’t combine that and create a new or refined thought with it, you’re just regurgitating. And I don’t know that anyone actually wants a human-regurgitated summary of an AI-generated summary of anything.

Yet there is a worse case still: when the person, having asked the AI for an overview of a topic, then asks it to write the piece, which even had they written it would have been nearly valueless. One can imagine how this already thin gruel may be further diluted, but it is not worth the time to do so.

Cum non sint re vera doctiores, sed loquaciores [they have not learned more, only spoken more].

There was no AI in Burton’s time, and its qualities are debatable in this one, but there were tools employed to the same purpose: to present a facade of knowledge without its supporting structures. Like any false front, it collapses as soon as anyone pushes on it.