I predict that asking students to hand-write assignments is not going to go well. Unfortunately, universities built on the consumer model (author teaches at Arizona State) are incentivized to listen to student feedback over the professor’s good intentions.
So don't accredit universities that want to turn into degree mills.
Beat this game of prisoner's dilemma with a club at the accreditation level. Students can complain all they want, but if they want a diploma which certifies that they are able to perform the skills they learned, they will have to actually perform those skills.
> So don't accredit universities that want to turn into degree mills.
This is way outside the scope of something that a faculty member who is, as the article says, trying to teach has any hope of implementing within a reasonable time frame. Of course the ideal is that faculty, as major stakeholders in the educational institution, should ideally be active in all levels of university governance, but I think it is important to realize how much of a prerequisite there is for an individual professor even to get their voice heard by an accrediting body, let alone to change its accrediting procedures.
That's setting aside the fact that, even if faculty really mobilized to make such changes, in the absolute best case the changes would be slow to implement, and the effects would be slow to manifest, as universities are on multi-year accreditation cycles and there would need to be at least a few reputable universities that were disaccredited before others started taking the guidance seriously. Even if I were willing to throw everything into the politics of university governance, which would make my teaching suffer immensely, I'm not willing to say that we'll just have to wait a decade to see the effects.
The consumer model isn't all bad. But it can lead to wildly different outcomes based on self-selection and incentives.
Take gyms, for example. You have your cheap commodity convenience gyms like Planet Fitness, where a lot of people sign up (especially at the beginning of the year) but few actually stick to it to get any real gains. But you also have pricy fitness clubs with mandatory beginner classes, where the member base tends to be smaller but very committed to it.
I feel like students that are OK with just phoning it in with AI fall into the Planet Fitness mindset. If you're serious about gains (physically or intellectually), you'll listen to the instructors and persist through uncomfortable challenges.
I think a better approach might be to get students to use AI as a writing coach. Get them to commit to a short handwritten essay during class, then use AI give them feedback on the essay. Their interaction with the AI and how they respond to the feedback becomes the assessment material. That's not compatible with the authors "Butlerian Jihad" ideology, though.
This is insane to me. Why not title the class "how to use AI?" Why not make this the title of every class?
I see no future in education other than making homework completely ungraded, and putting 100% of the grade into airgapped exams. Sure, the pen and paper CS exam isn't reflective of a real world situation, but the universities need some way to objectively measure understanding once the pupil has been disconnected from his oracle.
“Butlerian Jihad” ideology is definitely overselling it.
The bigger problem is that kids can just hand write an essay that an AI gave them.
I teach at a university and I just scale my homework assignments until they reach or exceed sightly the amount of work I expect a student to be able to do with AI. Before I would give them a problem set. Next semester homeworks will be more like entire projects.
Punishing honest students by ensuring that they will fail unless they cheat is an absurd solution. In school I went to great lengths to do my work well and on my own. It was disheartening to see other students openly cheat and do well, but at least I knew that I was performing well on my own merits.
Under your system, I would have been actively punished for not cheating. What's the point of developing a cure that's worse than the disease?
It isn't a punishment if the AI is doing the work. The goal is to make students utilize the skills they will need now that AI can supplement their abilities.
But are AIs now _required_? If you're tailoring the class to allow growth potential for students using AI, what happens to the students who cannot use it (for whatever reason)?
It's a bit similar to making a class harder because some students are getting extra help via private tutoring.
> I just scale my homework assignments until they reach or exceed sightly the amount of work I expect a student to be able to do with AI.
1. Absurd. The measurement should be learning not “work”. My students move rocks with a forklift… so I give them more rocks to move?
2. From the university I’m looking for intellectual leadership. Professors thinking critically about what learning means and how to discuss it with students. The potential is there, but let’s not walk like zombies unthinking into a future where the disappearance of ChatGPT 8.5 renders thousands of people unable to meet the basic requirements of their jobs. Or its appearance renders them unemployed.
The goal isn't work - its simply that I acknowledge that AI can perform a lot of labor that students previously had to spend a lot of time on. Because they have more time, my ambitions for what they can accomplish are higher.
I teach data science, which involves a lot of relatively unimportant glueing together of libraries. Yes, I want the students to know how to program, but the key skills are actually coming to grips with data, applying methods correctly, etc. The AI can make writing out the actual code substantially more efficient for them and I expect them to use that saved time to understand higher level skills.
Sounds like an optimization for the students not interested in learning at the expense of the students who are.
I understand your intentions but I'm skeptical even this solves the problem.
Realistically I think we're just moving away from knowledge-work and efforts to resuscitate it are just varying levels of triage for a bleeding limb.
In the actual workplace with people making hundreds of thousands a year (the top echelon of what your class is trying to prepare students for) I'm not seeing output increase with AI tools so clearly effort is just decreasing for the same amount of output.
Perhaps your class is just supposed to be easier now and that's okay.
What do you view as coming after knowledge work? Do you think we'll see a resurgence of physical, in-person work?
Not rhetorical, I'm genuinely curious and could see that being a real scenario.
With an ageing population, I don't think we are going to run out of the need for basic healthcare workers.
Serfdom, possibly WALL-E but that's an optimistic scenario. WALL-E people actually live a somewhat dignified existence compared to other possibilities.
Digging people out of the codebases they're going to shit out with this technology.
then you're discriminating against students not using AI. I for sure know I really would be depressed to be asked for a huge pile of work I'll do myself when other will just cheat and have free time to do something else work on interesting projects or see friends whatever.
Would the solution not be to pivot to more in-person demonstrations of skill and knowledge? Say the tests and exams become hand written, or taken in a controlled lab or whatever, so you need to eventually pick up the skill. But how you pick up the skill is irrelevant.
Maybe the issue is, somewhat, the concept of graded homework in the first place. It's meant to be practice material, but is only actually useful as practice material if students put in that work. A lot of students come to resent the mountains of at-home work as the busywork that it feels like in the moment, and I feel like this whole set of emotions underpins the argument but isn't really called out for what it is all that often. Teachers understand the value of actually doing that practice, but the grading system rewards, instead, rushing through the busywork as quickly as possible. Are we not testing for the right things?
> Would the solution not be to pivot to more in-person demonstrations of skill and knowledge?
Yeah, this seems like the obvious conclusion.
It probably is. But it's probably also more expensive and doesn't necessarily apply across all domains--certainly not all the time.
I don't consider the use of AI cheating. I think of it as a reasonable skill to be applied to work.
To be frank, a lot of programming is busywork, boilerplate, looking up information. Now that an AI can do that for my students I expect them to spend the time made up on developing higher level skills.
You sound like everything that is wrong with the US education system concentrated on one single person.
What made you get into teaching?
This claim is absurd and the comment is unserious.
Would the teacher then grade the massive workload with AI also? There isn't really a limit to how much output an AI can generate and the more someone demands, the less likely it is that the final result will be looked at in any depth by a human.
The point is that I will expect more high level thinking and less rote coding.
As long as you are upfront that you expect AI will be used, this seems like a solid and practical approach to me.
Do you expect someone to do the same for you? I mean to increase your workload until you cannot do it even with the AI help?