One professor's reflections on the end of an era, as AI tools such as ChatGPT have murdered the student essay (RIP). Here's why that threatens the future of human cognition—and how to save ourselves.
When I was studying history at the University of Vienna many years ago, our final exam was in the form of an hour-long Q&A discussion group of 3-4 students led by the professor. The quality of the answers and insights determined a large part of the ultimate grade for the course, with levels such as unsatisfactory - satisfactory - completely satisfactory - excellent - outstanding.
I’m reminded of a certain professor I had in grad school (engineering) in the late ‘80s. Homework wasn’t graded, but the exams were all oral — Mano a mano. He was much feared. Most students avoided his classes. Professor S ran into one of my classmates on campus a full month after the oral final of one class and insisted he take the exam again, on the spot, because he’d done so poorly the first time. That classmate dropped out by the end of that day. True story. I took everything Professor S offered and had him on my thesis committee. He insisted on a mano a mano session with me before the oral prelim before the entire committee. Arrows were flung and arrows were deflected. Back. Forth. For one hour. Then two. And then…he seized up at one of my answers and smirked that he’d gotten me. Insisted I was wrong. I stood my ground. He started yelling. WRONG! WRONG! Told me to pick up a book and find the answer. I said no. Not wrong. Fifteen minute standoff and I saw a change come over his face as he realized HE was wrong. He shouted, “GAMMA! Just say GAMMA!” So I said gamma and we moved on. I’d passed. Take that, ChatGPT.
I'm gleefully guilty of 2 things: 1.Not using my brain's limited memory space for remembering phone numbers.
2. Instantly stopping whatever I am doing to enrich my brain and soul with the writings/thoughts of Brian Klaas.
I am always grateful for your deep dives into the causations and consequences of humanity's ugly tendencies and shallow trends... Followed by inspiring reasons for us to resist them, and persist on our higher roads.
I wholeheartedly agree that writing/thinking/note-taking/research/reading and seminars are crucial to the development of intellect and mind. I have seen medicine going down the tubes for similar reasons (excuse the awkward metaphor).
This is a heartrending eulogy. I remember sweating it out during each of my semesters in college and grad school, how much I benefitted from those old timey blue books and term papers. As I've matured, I've come to appreciate how much this process helped mature my thinking. But one thing that might be considered is the focus on grading essays and papers, which distorts the purity of this process. Wouldn't a pass fail, with maybe some 'gold stars' for merit, help take this unnecessarily daunting pressure off? It might also be useful to do some meta-thinking, so as to fully recognize what's being lost: the ability to contemplate what you've read, a subtle relational process between writer and reader. Inherently creative, it brings forth new understandings. The value of contemplation was more clearly understood in earlier eras, I think, as a religious-philosophical process called lectio devina that contributed to insight. There is something about effort which opens the mind, and the subsequent combining of received and self-generated ideas. Sad that this 'alchemy' will be lost because adolescents will fail to grasp this value while rebelling against bending the knee to the judgment of those who have gone before them.
I blame multiple-choice testing. Guessing. I’m good at it, but essays brought out studying and over-studying. I knew lists I never needed, just in case it was on the test. I condensed, compared/contrasted, even studied the professor and what He/She seemed to think was important (I called it ‘professorology’).
When l was teaching Art Appreciation (would meet the Gen.Ed. arts requirement), the first assignment given had 3 parts:
1. Students had to generate a description of Marc Chagal's I and the Village using AI. It was obvious the prompts they used was deficient.
2. Next, students had to use the description and analysis process of Feldman's method of art criticism to do the same task with the same painting (his method involves very deep LOOKING in the description and analysis).
3. Write an essay comparing the 2.
I got some interesting results.
I also took great pains to give writing/looking prompts that couldn't be done with Ai, like " Your budget is unlimited and you are buying a piece of art for your best friend, that matches their personality. Describe them and why the piece you chose fits that description visually." I found they had a very hard time writing (never mind the spelling and grammatical errors). The biggest complaint in my course evaluations were l made them write too much for a 100 level course (they only had 1 writing assignment a week).
I think you are correct that so few young people read ANYTHING; all is for immediate gratification (why read when you can see the movie?!?), and it absolutely interferes with any development of creative and critical thinking, and as we're seeing with this administration, seeing and understanding consequences for actions beyond the immediate result (i.e.entire departments gone, now there'sno ine to predict hurricanesaccurately, etc.). I'm not sure how to change that.
Yes! Once again, delighted to engage with your thoughtful, insightful reflections, Brian. Ah, shall we of the Blue Book exam generation not also spare a moment to appreciate the value of the handwritten essay challenge to the student's learning process, (in spite of the professor's challenge to decipher the handwriting!).
I remember taking the first AI course my university offered over 50 years ago. Programming in LISP - attempting to craft something that was “artificially intelligent.” My Greek CS/philosophy professor waxing on about what was to come. After class a small group would retire to the ‘Knollwood Tavern’ to discuss the class and speculate on what the future held. The consensus was “not in our lifetime” - but today - reality. The tavern is gone as are some of my classmates but we are in a new reality and as I look at what is to come I find it frightening. Brian thanks for your thoughts.
While there are those who disagree, I find it implausible that AI is truly sentient - that it thinks and perhaps, more importantly, that it “feels”. As a “stochastic parrot” it rearranges and regurgitates language that has previously been vetted by sentient humans, so it superficially reads well. But can it really ever generate any novel insights? Does it ever have a “Eureka!” moment when something new and interesting comes together and resonates? The joy of learning at any stage in life seems to me to depend on such moments and those moments, I think, are dependent on the work and experience/sensations of wet, metabolizing, biological brains/minds. By outsourcing that work to electronic circuitry we merely playback the linguistic musings of previous sentient humans and close ourselves off to the wonder of the process itself and the promise of true discovery.
I found Graham Burnett’s piece in the New Yorker, “Will the humanities survive artificial intelligence?” (can’t provide a link, sorry) to be very helpful and provocative in thinking about ways that writing can still be a way of helping students learn about to think.
This semester I plan to create assignments where students turn in dialogues with ChatGPT, where they need to persuade it of something or defend some pre-existing assumption of their own. I find in my own work getting into in-depth exchanges with AI models about all kinds of things sharpens my thinking and frees me to play with ideas in a way that is liberating.
When I am writing alone I am in a constant exchange of ideas with myself and that is truly a mind-expanding cognitive workout. I won’t want to give that up but I’m also a fan of collaboration and Chat GPT can be a stimulating co-author.
I’m going to try to introduce students to the collaborative assignment where they turn in the exchange as conducted, rather than an essay, as the first draft. Where I will take it from there I haven’t quite decided. Maybe an in-class exercise where they write a review critiquing the exchange or their conclusions. I’m also going to make them responsible for fact-checking the bot. Any hallucinations will count against their grade.
Brilliant piece. Thank you.
Your writings always make me think.
When I was studying history at the University of Vienna many years ago, our final exam was in the form of an hour-long Q&A discussion group of 3-4 students led by the professor. The quality of the answers and insights determined a large part of the ultimate grade for the course, with levels such as unsatisfactory - satisfactory - completely satisfactory - excellent - outstanding.
I’m reminded of a certain professor I had in grad school (engineering) in the late ‘80s. Homework wasn’t graded, but the exams were all oral — Mano a mano. He was much feared. Most students avoided his classes. Professor S ran into one of my classmates on campus a full month after the oral final of one class and insisted he take the exam again, on the spot, because he’d done so poorly the first time. That classmate dropped out by the end of that day. True story. I took everything Professor S offered and had him on my thesis committee. He insisted on a mano a mano session with me before the oral prelim before the entire committee. Arrows were flung and arrows were deflected. Back. Forth. For one hour. Then two. And then…he seized up at one of my answers and smirked that he’d gotten me. Insisted I was wrong. I stood my ground. He started yelling. WRONG! WRONG! Told me to pick up a book and find the answer. I said no. Not wrong. Fifteen minute standoff and I saw a change come over his face as he realized HE was wrong. He shouted, “GAMMA! Just say GAMMA!” So I said gamma and we moved on. I’d passed. Take that, ChatGPT.
I'm gleefully guilty of 2 things: 1.Not using my brain's limited memory space for remembering phone numbers.
2. Instantly stopping whatever I am doing to enrich my brain and soul with the writings/thoughts of Brian Klaas.
I am always grateful for your deep dives into the causations and consequences of humanity's ugly tendencies and shallow trends... Followed by inspiring reasons for us to resist them, and persist on our higher roads.
I am also sad that I don’t remember any phone numbers - not even my kids! Love your Bankman-fried reference and comment.
Cogent arguments, Benjamin ;)
I wholeheartedly agree that writing/thinking/note-taking/research/reading and seminars are crucial to the development of intellect and mind. I have seen medicine going down the tubes for similar reasons (excuse the awkward metaphor).
Yeah that "Benjamin" is awesome, right? LoL!
This is a heartrending eulogy. I remember sweating it out during each of my semesters in college and grad school, how much I benefitted from those old timey blue books and term papers. As I've matured, I've come to appreciate how much this process helped mature my thinking. But one thing that might be considered is the focus on grading essays and papers, which distorts the purity of this process. Wouldn't a pass fail, with maybe some 'gold stars' for merit, help take this unnecessarily daunting pressure off? It might also be useful to do some meta-thinking, so as to fully recognize what's being lost: the ability to contemplate what you've read, a subtle relational process between writer and reader. Inherently creative, it brings forth new understandings. The value of contemplation was more clearly understood in earlier eras, I think, as a religious-philosophical process called lectio devina that contributed to insight. There is something about effort which opens the mind, and the subsequent combining of received and self-generated ideas. Sad that this 'alchemy' will be lost because adolescents will fail to grasp this value while rebelling against bending the knee to the judgment of those who have gone before them.
I blame multiple-choice testing. Guessing. I’m good at it, but essays brought out studying and over-studying. I knew lists I never needed, just in case it was on the test. I condensed, compared/contrasted, even studied the professor and what He/She seemed to think was important (I called it ‘professorology’).
When l was teaching Art Appreciation (would meet the Gen.Ed. arts requirement), the first assignment given had 3 parts:
1. Students had to generate a description of Marc Chagal's I and the Village using AI. It was obvious the prompts they used was deficient.
2. Next, students had to use the description and analysis process of Feldman's method of art criticism to do the same task with the same painting (his method involves very deep LOOKING in the description and analysis).
3. Write an essay comparing the 2.
I got some interesting results.
I also took great pains to give writing/looking prompts that couldn't be done with Ai, like " Your budget is unlimited and you are buying a piece of art for your best friend, that matches their personality. Describe them and why the piece you chose fits that description visually." I found they had a very hard time writing (never mind the spelling and grammatical errors). The biggest complaint in my course evaluations were l made them write too much for a 100 level course (they only had 1 writing assignment a week).
I think you are correct that so few young people read ANYTHING; all is for immediate gratification (why read when you can see the movie?!?), and it absolutely interferes with any development of creative and critical thinking, and as we're seeing with this administration, seeing and understanding consequences for actions beyond the immediate result (i.e.entire departments gone, now there'sno ine to predict hurricanesaccurately, etc.). I'm not sure how to change that.
I was once told by a repairman that he had given up studying to be a mechanical engineers because all the answers were on line.
I replied that in my career as an electrical engineer that I had determined that the key in engineering was not the answer, it was the question.
Yes! Once again, delighted to engage with your thoughtful, insightful reflections, Brian. Ah, shall we of the Blue Book exam generation not also spare a moment to appreciate the value of the handwritten essay challenge to the student's learning process, (in spite of the professor's challenge to decipher the handwriting!).
I remember taking the first AI course my university offered over 50 years ago. Programming in LISP - attempting to craft something that was “artificially intelligent.” My Greek CS/philosophy professor waxing on about what was to come. After class a small group would retire to the ‘Knollwood Tavern’ to discuss the class and speculate on what the future held. The consensus was “not in our lifetime” - but today - reality. The tavern is gone as are some of my classmates but we are in a new reality and as I look at what is to come I find it frightening. Brian thanks for your thoughts.
While there are those who disagree, I find it implausible that AI is truly sentient - that it thinks and perhaps, more importantly, that it “feels”. As a “stochastic parrot” it rearranges and regurgitates language that has previously been vetted by sentient humans, so it superficially reads well. But can it really ever generate any novel insights? Does it ever have a “Eureka!” moment when something new and interesting comes together and resonates? The joy of learning at any stage in life seems to me to depend on such moments and those moments, I think, are dependent on the work and experience/sensations of wet, metabolizing, biological brains/minds. By outsourcing that work to electronic circuitry we merely playback the linguistic musings of previous sentient humans and close ourselves off to the wonder of the process itself and the promise of true discovery.
I especially liked the other two articles you linked. Quite the hat trick.
I found Graham Burnett’s piece in the New Yorker, “Will the humanities survive artificial intelligence?” (can’t provide a link, sorry) to be very helpful and provocative in thinking about ways that writing can still be a way of helping students learn about to think.
This semester I plan to create assignments where students turn in dialogues with ChatGPT, where they need to persuade it of something or defend some pre-existing assumption of their own. I find in my own work getting into in-depth exchanges with AI models about all kinds of things sharpens my thinking and frees me to play with ideas in a way that is liberating.
When I am writing alone I am in a constant exchange of ideas with myself and that is truly a mind-expanding cognitive workout. I won’t want to give that up but I’m also a fan of collaboration and Chat GPT can be a stimulating co-author.
I’m going to try to introduce students to the collaborative assignment where they turn in the exchange as conducted, rather than an essay, as the first draft. Where I will take it from there I haven’t quite decided. Maybe an in-class exercise where they write a review critiquing the exchange or their conclusions. I’m also going to make them responsible for fact-checking the bot. Any hallucinations will count against their grade.
Edited to add the link: https://www.newyorker.com/culture/the-weekend-essay/will-the-humanities-survive-artificial-intelligence