It’s the start of the semester, so I had a look at the advice I give to undergraduates. It’s pretty good, if I do so say so myself.

During your undergraduate studies, I strongly suggest you develop a sustainable habit of reading, writing and searching the literature. Make an effort to become academically literate by familiarizing yourself with the resources that a university provides. Set aside time on a regular basis, preferably at least half an hour every day, to read some scholarly prose, to write a paragraph or two about something you know, and to use your library’s databases to find a relevant book or article.

Learn what paragraphs and references are and how they work. A paragraph takes about one minute to read, during which it supports, elaborates or defends a single, well-defined claim. It will normally consist of no less than six sentences and no more than two-hundred words. For everything you know, you do well to learn how to write a paragraph about it in under half an hour. This includes providing the proper references, which provide information about the sources you have used. You should learn the difference between the “in-text citation”, which you provide in parentheses within the paragraph you are writing, and the “reference list”, which  goes at the end of your paper and provides all the information your reader needs to locate your source. Paragraphs and references are governed by  convention, and part of being a good scholar (and therefore a good student) is knowing what those conventions are.

As a general rule, every paragraph has a “key sentence”. This means that every paper you write can be summarized simply by listing your key sentences. If you’ve written an eleven-paragraph paper, you should have made eleven key claims, each of which can be stated in a simple declarative sentence, and supported, elaborated, or defended by five or more further sentences. If you copy just your key sentences (one for each paragraph) into a separate document, and list them in order, they should make sense, separate from the paragraphs you have written. This “after-the-fact outline” is a good way to see whether you have produced a coherent line of argument in your essay. It’s also obviously a good way of keeping track of what you’ve learned.

One of the most important things to learn as an undergraduate is that “knowing” something in an academic setting is the ability to discuss it reasonably with other knowledgeable people. Your authority to claim that something is true stems from the reasons you are able to provide for believing it. In your studies, then, you are not just acquiring a new set of “truths”, you are learning how to support your beliefs with reasons, and how to discard them in the face of better reasons. All of this happens within a “conversation” among scholars we call discourse. It consists of everything that can be reasonably said (and written) about a subject, and since there are always arguments and disagreements among scholars, the discourse doesn’t just consist of true statements. That’s important to keep in mind. You will have to get used to being told that something you believe is wrong, and to telling others that they are wrong. One of the most essential functions of scholarship in an academic setting is to correct our errors in thinking.

In academia, we call people are who qualified to tell us we are wrong “peers”. You do well to keep a clear image of these people in your mind as you proceed. Think of them as the best and the brightest in your class, the most serious students in your cohort. Don’t always imagine yourself conversing with your teachers and examiners or their peers. One day, of course, you will begin to do so. In fact, your teachers are trying to make you into their peers — they are trying to qualify you to one day tell them that they’re wrong. It’s fine to experiment with this possibility, both in your thoughts and in your conversations, but don’t think that your teacher’s judgment is the only thing that matters. In your assignments, your teacher is testing your ability to converse with your intellectual equals, not with people who have read and thought much more about the subject than you have. Focus on the conversation you could reasonably have with students in your class, then, not the conversation you may one day have with a Nobel Prize winner in your area of expertise.

It’s not just about what you know but how you know it. Your discipline is defined by a set of theories and methods as well as a vast body of accepted facts. In every class you take, you will improve your understanding of them. But you will also improve your ability to discuss them, to participate in the discourse in your area of expertise. Your awareness of “the literature” will improve through your reading, and your literacy skills will improve through your writing. Keep track of your references — the ever-expanding store of sources you draw on to support your reasoning about the questions that you and your peers are trying to answer. Your theories and methods, too, have a history; and your sources — all those names and dates you invoke in your in-text citations — tell the story of the struggle of the people who came before you to make sense of the world in which we live. Using a theory and a method doesn’t give you any simple answers, nor any final authority. Rather, they frame your questions and ground your answers in a conversation with your peers, who know the same theories and methods on the basis of the same sources. Having that conversation is what a university education is all about.

By what route is the soul or history or society or the subconscious transformed into a series of black lines on a white page?

Italo Calvino

Every morning and evening I draw a picture of my hand. It only takes a few minutes. I look at my hand, try out a few different positions, settle on one, and then try my best to render it on the page. I’m not a great artist and it is probably too late to become an even halfway capable illustrator. There is something about my lines that just feels amateurish. I know people who draw with much greater confidence and whose drawings look like there is a real “causal” relationship with the things they depict. It’s like they are able to project their visual image onto the page and then simply trace it. It’s like the light from the object that hits their eye somehow directly impacts the page.

That’s not how it works, of course. Or, at least, I assume that’s not how it works. Artists learn to draw like the rest of us by looking at the world and training their hands to make lines that we all recognize as pictures of it — two dimensional shapes on a plane that indicate three dimensional objects in space.

In his lecture, “Cybernetics and Ghosts,” from 1967, which I had hoped to write much more about last week, Italo Calvino describes his struggle with writing in similar terms. “Literature as I knew it was a constant series of attempts to get one word to stay put after another,” he tells us. Notice that this removes a dimension from my description of the problem of drawing. In fact, writing is more difficult (if it is) than drawing because it is usually an attempt to render a four-dimensional object (a story unfolding in time) along a one-dimensional line (one word after another). The artist only has to flatten the object once; the writer has to do it three times, transforming the breadth, the depth, and the duration, of an experience into a sentence. Then again, perhaps both operations are infinite in their scope. As Calvino suggests, whatever lines we choose to draw or write, we’re putting our eternal soul on the ephemeral page.

I’m going to have to leave it there for a while. The semester is beginning and I have to turn my attention to teaching students how to write paragraphs about the things they learn. I have to teach them to appreciate the finitude of the problem, get them to see the limits of the page as a friend, help them to use it to get the words to stay put.

See also: , “One-Dimensional Prose”, “The Two-Dimensional Page”, “The Three-Dimensional Book”, “The Fourth Dimension”

Last summer, Mike Sharples drew my attention to a lecture that Italo Calvino toured Italy with in 1967, reprinted as “Cybernetics and Ghosts” in the Vintage collection The Literature Machine. It presents a puzzling and somewhat disturbing argument. But the other day I was rereading it and think I discovered a key to unlocking its meaning and dispelling some of my concerns.

First, let’s quickly summarize the concerns. Calvino takes his cue from the meeting of theoretical and technological developments that must have appeared quite exciting at the time, although the excitement, which I suppose I was literally born into only four years after Calvino’s lecture, had always, or at least until last summer, seemed to me a little overblown. Simplifying somewhat, he asks us to imagine the consequences of feeding our structuralist and formalist theories of language and literature into a sophisticated computer — indeed, a computer whose sophistication can be seen simply as an extrapolation from Raymond Queneau’s “rudimentary model of a machine for making sonnets,” which I have myself suggested is a good model for thinking about large language models.

Having laid down these procedures and entrusted a computer with the task of carrying out these operations, will we have a machine capable of replacing the poet and the author? Just as we already have machines that can read, machines that perform a linguistic analysis of literary texts, machines that make translations and summaries, will we also have machines capable of conceiving and composing poems and novels? (12)

His answer to this question seems at first to be yes. He even welcomes the prospect. Like Barthes, Calvino imagines a time when “the author vanishes — that spoiled child of ignorance — to give place to a more thoughtful person who will know that the author is a machine, and will know how this machine works” (16). Of course, he was able to say this at a time when this situation was still a long way off. In fact, I don’t think the machines he mentions — machines that can read, analyze, translate, and summarize literary texts — existed in 1967 as he claims. There is no question that they do today, except there is one thing: it is not clear that knowing how a large language model works tells us very much about how that “spoiled child of ignorance” works. Maybe authors aren’t machines after all?

I’m not doing any of these ideas justice because I’m in a rush to get to the passage that suddenly made me realize that Calvino wasn’t being entirely serious.

Literature is a combinatorial game that pursues the possibilities implicit in its own material, independent of the personality of the poet, but it is a game that at a certain point is invested with an unexpected meaning, a meaning that is not patent on the linguistic plane on which we were working but has slipped in from another level, activating something that on that second level is of great concern to the author or his society. The literature machine can perform all the permutations possible on a given material, but the poetic result will be the particular effect of one of the permutations on a man endowed with a consciousness and an unconscious, that is, an empirical and historical man. It will be the shock that occurs only if the writing machine is surrounded by the hidden ghosts of the individual and of his society. (22)

There’s a lot to work with here. And I’m going to try to pick up the threads on Wednesday and Friday again. As I’ve mentioned many times before, already before Calvino’s lecture (and I’m sure Calvino was aware of it) Borges had rejected the idea that literature is a “combinatorial game”: “Those who play that game forget that a book is more than a verbal structure, or a series of verbal structures; a book is the dialogue with the reader, and the peculiar accent he gives to its voice, and the changing and durable images it leaves in his memory.” In this passage, Calvino seems to be making the same point. Literature needs a soul. I’d argue that all writing does.

I’m one of those people who thinks that “artificial intelligence” is a misnomer. So is “machine learning” and, I hasten to admit, I sometimes extend this line of criticism all the way to such familiar things as “computer memory” and “programming language”. That is, I’m a bit of a kook. It is my deep conviction that computers can’t think, speak, or remember, and I don’t just mean that they can’t do it like we do it; I mean they can’t do it at all. There is no literal sense in which a machine can learn and, if we are to take it figuratively, the metaphor is stone dead. Much of the discourse about the impending rise of artificial general intelligence these days sounds to me like people earnestly claiming that, since they have “legs”, surely tables are soon going to walk. We have not, if you’ll pardon it, taken the first step towards machines that can think.

To fully appreciate my seriousness on this point, let me try to persuade you that pocket calculators can’t do math either. They have a long and very interesting history and, until I looked into it, I didn’t realize that it was actually Blaise Pascal who invented the mechanical adding machine – the Pascaline — in 1642. But, before this, there was always the abacus, and before this there was the root of the word “calculus”, namely, “chalk,” or the small limestone pebbles that were used as counters and eventually become the beads on the wires of counting frames. Interestingly, the word “abacus” originally denoted a writing surface of sand that was also used to do calculations. In any case, I hope you will immediately agree that a pile of pebbles or a string of beads can’t “do math”. A paper bag that you put two apples into and then another two is not capable of addition just because there are, in fact, four apples in the bag when you look.

Is it such a leap from this to the electronic calculator? Do we have to imbue it with “intelligence” of any kind to make sense of it? Some will point out that calculators use numbers and operators in their symbolic form as input and output. You type “2 +2 =” and you get “4” back. But suppose you have an ordinary kitchen scale and a bunch of labelled standard weights. Surely you can get it to “add up” the weights for you? Or you can imagine a plumbing system that moves quantities of water in and out of tanks, marked off with rulers so that the quantities are made explicit. Whether these systems are manual, hydraulic, mechanical, or electronic doesn’t change the fact that these are just physical processes that, properly labelled, give us an output that we humans can make sense of.

That’s all very well, you might say, but have I tried ChatGPT? How do I explain the perfectly intelligible output it produces daily? Here’s my take: it articulates words, like a calculator calculates numbers. To calculate is basically just to “keep count”. To articulate is just to “join up”. A machine that is capable of keeping count is no more intelligent than a paper bag, nor is a machine that is capable of joining words together. Imagine you have three bags, numbered 1, 2, 3. In bag number 1 there are slips of paper with names on them: “Thomas”, “Zahra”, “Sami”, “Linda”. In bag number 2 there are phrases like “is a”, “wants to be a”, “was once a”, and so forth. In bag number 3, there are the names of professions: “doctor”, “carpenter”, “lawyer”, “teacher”, etc. You can probably see where this is going. To operate this “machine”, you pull a slip of paper out of each bag and arrange them 1-2-3 in order. You’ll always get a string of words that “make sense”. Can this arrangement of three bags write? Of course not!

In a famous essay, Borges credits Ramon Llull with the invention of a “thinking machine” that works much like this. Or, rather, doesn’t work: it is not “capable of thinking a single thought, however rudimentary or fallacious.” Llull’s device (or at least the device that Borges imagined based on Llull’s writings) consisted of a series of discs that allowed the operator to combine (or “correlate”) agents, patients, acts. (I wonder if this is where the “robot rights” people got the idea of distinguishing between the moral “agency” and “patiency” of subjects.) “Measured against its objective, judged by its inventor’s illustrious goal,” Borges tells us, “the thinking machine does not work,” but, like all metaphysical systems, he also points out, its “public and well-known futility does not diminish [its] interest.”

The machinery has become a lot more complicated since Pascal and Llull. But I want to insist that it has not become more mysterious and, by the same token, not more intelligent. That is, we don’t need a “ghost in the machine” to explain how the prompt we give to a language model spurs it to produce an output that, on the face of it, “makes sense”. Just as the calculator has no “understanding” of numbers or quantities or the mathematical operations it carries out, so, too, does ChatGPT have no conception of what the prose it generates means. All it does is to convert a text into a string of numbers, those numbers into vectors in a hundred-dimensional space, in which it then locates the nearest points and, from these, chooses one of them at somewhat random (depending on the “temperature”). It converts the result (which is just a number) back into a word or part of a word (by looking it up in a table) and adds this to the string it is building. It repeats the process to find the next word. If “artificial intelligence” is a misnomer, what would I call these machines? They are electronic articulators.

I doubt this will catch on. In 1971, the year I was born, Sharp marketed its EL-8 calculator as “a really fast thinker” and I remember, ten years later, my elementary school principal announced that a computer (an Apple II+) would “moving in”. It was all very exciting.