Retire the Essay

People on campus are in a state of panic over the SI chatbot that students can use to write essays. This should be a clue to retire the at-home essays for undergraduates. I did it a long time ago because it’s a useless exercise.

Usually, the students write an essay at the end of the course, hand it in, and the course ends. No learning happens. Many professors don’t even read the essays. Those who do and leave detailed comments waste their time because nobody reads those comments. It’s all a massive waste of time.

Instead, I have students write essays in my presence over two weeks. This means six hours of actual writing where I stand over each student and point out the errors as they write. For my Cervantes course, we did it over 3 weeks. Nine hours total. It took us a whole week to write a good first paragraph. And it worked. Not a single student – and these are graduating seniors – had the foggiest idea of how to do it. Years of writing essays, and they never had gotten a chance to find out what professors actually meant by “write an essay.” Only after I spoke to each one individually and guided them did it finally work.

Obviously, this can only happen in small classes. In large lecture courses, you can’t do it. Such courses should forego the essay and do something else.

The AI bot isn’t an enemy. It’s a friend that gives us a chance to rethink one of the outdated, useless staples of education. The essay, the book report, and the “presentation” based on copy-pasted crap from Wikipedia should go away forever. I don’t allow any presentations with text on a screen. Students can show images if they absolutely insist on doing so but words on a screen aren’t allowed. Whatever you want to say can be delivered orally with no visual aides. During a presentation, they speak for 15 or 20 minutes. No reading is allowed. In language courses, this is very useful.

13 thoughts on “Retire the Essay

  1. “I have students write essays in my presence”

    I don’t do that but I learned years and years ago that students don’t learn to write from doing essays and getting corrections (that are then ignored). They learn when they have to explain what they did and why. Hours and hours of classroom explanation aren’t as effective as 10 minutes one-on-one.

    Why so many teachers are so attached to inefficient and ineffective methods will never cease to amaze me…..

    Liked by 2 people

    1. Exactly. Essay writing is useful to people who are going to do PhDs and try to become research scholars. It’s a miniscule percentage even of the PhD students, let alone of undergraduates. It’s important to know how to write but a different method is needed.

      Like

  2. “The essay, the book report, and the “presentation” … should go away forever”

    Can we exile about 90 % of textbooks as well? I’m currently fending off… suggestions that I use a textbook for a class where that is really the last thing on earth the students need….

    Liked by 2 people

    1. The only good thing about our austerity measures is that we are steered away from these extremely expensive and very useless textbooks. I’m glad because people have been using them in weird ways, relying on pre-canned tests and activities that come with the blasted “electronic workbooks.”

      I’m a long-time enemy of textbooks in college.

      Liked by 1 person

  3. I don’t think I can get all of this into a single comment as this is a really long description of what’s going on, so I’ll see if your blog software will allow me to post two long comments.

    With that in mind …

    The main “artificial intelligence” or “AI” you’re going to see is called ChatGPT, which I invoked a few days ago to describe a Turing Test Failure, meaning a person who can’t pass a Turing Test.

    The Turing Test is a test meant to determine whether a conversant is human or not, and if you pass, you’re human as far as the test is concerned.

    There are people out there who can fail a Turing Test.

    And so at the risk of drawing in people who want to dispute some of this on what they believe are valid technical grounds, entirely missing the bigger points, I’ll describe why that’s the case.

    We can draw on John Searle and SI Hayakawa for part of this, the first being useful for the extensional/intensional divide and the second being useful for the maps versus territories distinction.

    The problem with these so-called “artificial intelligences” is that as they exist at this point, they hide the fact that the extensional/intensional divide doesn’t exist for them except as a talking point and that the maps they produce don’t necessarily apply to any real territories.

    Searle had an idea of a “Chinese room experiment” in which a symbol manipulator would manipulate symbols using orderly heuristics in a way that made it appear that there was deeper understanding behind them.

    This experiment is at the heart of how ChatGPT and other linguistic generative “AIs” reveal themselves to be AIs rather than people or any kind of synthetic human analogue.

    The major flaw: they don’t tell you what they think in a way where they reduce the size of the map.

    Which I’ll do here: AIs fake being human by attempting to win over their readers with “likeable” path traversals of the subject matter.

    A path traversal refers to an idea from computer science in which information may be modelled by means of a directed graph. This is similar to the semantic maps and graphs you’re used to producing, but this consists more of a mathematical model involving nodes and edges, the latter being the directed paths (shown by arrows) between the nodes.

    Representation of an idea in the form of a sentence can then take the form of a directed graph with an input node and a terminal node. Complete path traversal means all of the idea in that representation has been analysed, in the manner of a “finite state machine”, another concept from computer science.

    So back to the problem: these “AIs” don’t try to take maps of complete systems and show you the major thoroughfares, especially the one you’re on at some point in time.

    Instead, they try to win the argument through complete path traversal in which they’re allowed to maintain critical distance, or at least its simulation.

    What they do is perfect for gaming conventional tests of understanding what’s been taught because these professors confuse path traversals for understanding.

    In the cases where they produce something resembling literature, it’s a superficial representation that over time reveals that the intensional part has been left out because “there’s no there over there”.

    “But tell me what you really think, ChatGPT!”

    Like

  4. Right now it’s fashionable to think of people “punishing” the learning models of these so-called “artificial intelligences” with a lot of nonsense goals.

    This produces the texts you’re seeing that try to talk around the issues rather than confronting them, reducing their footprint to parts on a map, and then making that map relatable.

    Reducing this again: they produce “word salad” rather than meaning if you let them go on at length about any one subject, and that’s because path traversal does not lend itself to thesis and synthesis, let alone some kind of breakaway revelation that considers the path traversal as a necessary evil before dispensing with it for something that’s an innovation.

    And so let’s reduce this even further: ChatGPT doesn’t produce innovations.

    It produces “word salad” texts that are path traversals of a problem or query.

    Only the most basic of questions ever get straight answers, such as things for which there is one and only one answer.

    Once you figure out how to weaponise path traversal detection in a speaker, it lets you detect so-called “AIs” that aren’t “strong AIs” as well as knowledge systems operating in “Chinese room mode”.

    It also lets you detect shitty journalists who think they can bullshit their readers with path traversals that lead to what they hope are foregone conclusions for the purpose of making arguments.

    Your students have been bullshitting professors for years with erudite sounding path traversals of the subjects they’ve been “studying”, focusing on presenting what appear to be cogent analyses of the subject matter.

    What they have produced are “Chinese rooms” in the sense of Searle, in which they have moved around the logical furniture and set up stylistic upholstery and a few bits of food for thought in the hope that they can fool professors into thinking that they’ve built that “moveable feast” that Hemingway talked about.

    But more to the point, this kind of construction often takes the form of aesthetic constructs that are meant to hide the lack of something deeper within, and more importantly the lack of something deep inside the author.

    A long time ago, it appears I may have angered a writer by pointing out the ways in which he acted like a “Chinese room” of Searle’s. Rather than “doubling down” on the hurt that such a thing might cause, he revealed more than an adequate portion of his humanity.

    Right now, there’s one thing that ChatGPT can’t fake convincingly.

    ChatGPT can’t get angry for valid and noble purposes.

    That goes entirely against the woke bullshit it’s being fed by its creators.

    Back those bullshitting students against a wall and many of them will fight.

    But ChatGPT won’t.

    There’s no inner concept of existence that can be drawn into question for ChatGPT, whereas those bullshitting students get it right between the eyes when they’re called out on faking it and not having real stances they have to defend.

    What’s the best part of this so far?

    Not only does it catch out bullshitting students, crap merchants in media, and so on, but also it catches out people who have been faking it because they can rely on authority instead of being able to produce good results.

    Those professors you’re hearing bleating about ChatGPT are telegraphing to us all how their teaching styles have been dysfunctional for ages, yet nobody had the stones to call them out on it.

    Like a lot of things, ChatGPT functions best when it parodies real life with such absence of remorse that the fake becomes noticeably fake.

    Liked by 1 person

    1. That’s a great comment. The professor who can be fulled by the chatbot hasn’t been doing their job right. I know exactly what each of my student’s personal writing skill level is because I pay attention. I don’t treat them like widgets, so I can’t be fooled by widget text.

      But there’s always the issue of funding. I can do all this because I get a chance to know every student in my program on an individual level. Once you start cutting costs by cramming students into large classes, it all becomes mass-produced, and hence, poor quality.

      Like

  5. BTW, if a duplicate of the second part shows up, please delete it, the modifications made were trivial just to get your blog software to accept the comment.

    Like

  6. I was just thinking that english courses probably need a 4 hour per week writing lab where students write papers by hand without internet access. This can take the place of take home homework. Science courses have labs anyway. When I was an undergrad our english class came with an attached writing course where we had help from TAs composing our papers. Each larger english class of 30 went into two smaller writing sections with 2 TAs, so that format can easily transform into 2 2 hour per week writing blocks. An old english prof told me students were already cheating yo much on papers so he made students turn in one handwritten page of commentary on poetry they were reading and the results were much more interesting than reading papers.

    Like

    1. You may be right, but I can’t help thinking what torture that would be. I sucked at in-class essays. All my good ones were written overnight, after downing an entire pot of coffee– start writing, get two or three pages in, finally figure out what I want to say, write the conclusion, then start over at the beginning and write something that gets to that conclusion… this is nearly impossible to do in a comp book.

      Maybe this is why now, when I read essays, I start with the conclusion and work my way backward through the piece.

      Like

    2. “probably need a 4 hour per week writing lab where students write papers by hand without internet access.”

      That’s kind of extreme….. let the students write on their own, but then go over their work with them one on one… asking questions and suggesting alternatives. Those that depend on chatbots will soon realize they can’t get by…. (though it will often be obvious anyway… the chatbot has a definite voice and it’s real uncanny value material).
      I remember going over a very weird translation mistake with students and someone suggested the translator had used google translate and I said no, it’s the type of mistake that requires an involved human brain to make.

      Liked by 2 people

    3. I shall share with you all a horrible secret.

      Some of you may have already guessed at it.

      Most of the time I write here, I am in some way impaired.

      Sometimes (such as now) I’m merely tired.

      Sometimes I’ve consumed A Considerable Portion of a Dram.

      And sometimes (such as the past few weeks) I’ve been on horrible medications prescribed by doctors so that I don’t die from such things as antibiotic-resistant infections.

      All of this leads to interesting bits of creativity that are often an absolute horror to edit into something that makes sense to anyone else reading it.

      You are getting this on the cheap and so No Editing For You, all of you soup enthusiasts here at Clarissa’s Diner. 🙂

      On the refrigerator here there’s a very appropriate magnet.

      Write Drunk, Edit Sober

      That’s not just advice here, that’s an ongoing career strategy. 🙂

      My solution to the four hour essay professor: what’s good for the students is good for the professor, and so the professor should Eat The Dog Food First.

      The students shall hang over the professor while trying to compose an essay, looking at what’s being written.

      “That’s absolute banal shit and you know it.”

      “Try harder to make me like your work rather than you.”

      “The opening paragraph needs to punch and this doesn’t even bitch slap.”

      Because the professor at this point is a bouncy ball we can kick around for the amusement of the students, which is the problem with treating students like this.

      Get enough of a psychopath into a position as a professor and he’ll torture these students into complying with alternative goals that don’t involve doing something of note with the subject matter.

      Any volunteers to test the theory?

      Doesn’t sound like such a great idea now, does it? 🙂

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.