Nice way, Rolf. As someone who has worked in AI for the last eight years, I really appreciate how you think through the risks and rewards to come up with your approach. My fear is that I know what's coming. That will be multi-agentic approaches that can not only read the material, digest it, regurgitate it, and polish it up with minimal knowledge created, but AIs on the other side that will read what read written, suggest it, and critique it as a neutral observer. That AI can tell another AI (the first model again, perhaps) to rewrite with the critiques included. It is almost like having a peer review before the professor even sees the work. What then?
My suggestion is do what Frank Luntz did at Penn--grade on class participation, understanding of the subject used in practical application, and force debate. This was one of my favorite classes there, and an approach that seems like it would work on almost any liberal art and be very well positioned to isolate the student from the AI. Just a thought.
Astra Taylor in her book, 'The Age of Insecurity', comments on being home schooled (actually self-taught) that she spent equal amounts of time being bored, hanging out with the dogs, and furiously learning about things that interested her, with of course, much guidance from her devoted parents / teachers. In the end, the only thing missing was an understanding of social interaction, which was for-better as well as for-worse. So much for professional 'education' with forced learnings and examinations in a regimented system of norm-creation and associated punishments.
Re grades, I learned that all you need is a coin toss. I remember the lunch poster session well. Three of us had to review about 25 posters in an hour and choose the top three. It was very hard work. Well, we got together with 10 minutes to choose and decided to each list our top three. They were the same - not in exact order but taken as a whole we all recognized the same outstanding posters. Ditto the worst (mostly because no work had been done). The rest were in the middle.
So our choice didn't take long, And if we'd needed more - say five - we could have tossed a coin for two more 'middles'.
Of course, you could repeat the process for the middles if you wanted finer grade distinctions but I thought there was something 'right' about recognizing outstanding effort, lack of effort and workmanlike efforts in the middle.
I suppose AI makes this impossible. It's a shame that yet another ceremony of innocence will be drowned by passionate intensity as we slouch towards our dazzling new future.
Rolf - thanks for a wonderful reflection!
Nice way, Rolf. As someone who has worked in AI for the last eight years, I really appreciate how you think through the risks and rewards to come up with your approach. My fear is that I know what's coming. That will be multi-agentic approaches that can not only read the material, digest it, regurgitate it, and polish it up with minimal knowledge created, but AIs on the other side that will read what read written, suggest it, and critique it as a neutral observer. That AI can tell another AI (the first model again, perhaps) to rewrite with the critiques included. It is almost like having a peer review before the professor even sees the work. What then?
My suggestion is do what Frank Luntz did at Penn--grade on class participation, understanding of the subject used in practical application, and force debate. This was one of my favorite classes there, and an approach that seems like it would work on almost any liberal art and be very well positioned to isolate the student from the AI. Just a thought.
Astra Taylor in her book, 'The Age of Insecurity', comments on being home schooled (actually self-taught) that she spent equal amounts of time being bored, hanging out with the dogs, and furiously learning about things that interested her, with of course, much guidance from her devoted parents / teachers. In the end, the only thing missing was an understanding of social interaction, which was for-better as well as for-worse. So much for professional 'education' with forced learnings and examinations in a regimented system of norm-creation and associated punishments.
Re grades, I learned that all you need is a coin toss. I remember the lunch poster session well. Three of us had to review about 25 posters in an hour and choose the top three. It was very hard work. Well, we got together with 10 minutes to choose and decided to each list our top three. They were the same - not in exact order but taken as a whole we all recognized the same outstanding posters. Ditto the worst (mostly because no work had been done). The rest were in the middle.
So our choice didn't take long, And if we'd needed more - say five - we could have tossed a coin for two more 'middles'.
Of course, you could repeat the process for the middles if you wanted finer grade distinctions but I thought there was something 'right' about recognizing outstanding effort, lack of effort and workmanlike efforts in the middle.
I suppose AI makes this impossible. It's a shame that yet another ceremony of innocence will be drowned by passionate intensity as we slouch towards our dazzling new future.