A long time ago, when I was a technology education professor, on the first day of class, I told students that they were to put a document on the internet. This was something that very few of them had ever considered, much less knew how to do. The assignment was simple: turn in a URL that pointed to a page on the internet, and have that page describe how it was they figured out how to create that document so that it could be retrieved with a web browser. This was a long time ago, so it required a bit of creativity, but for those who knew already, there was a requirement that you had to solve this problem a way that you’d never done it before—so a web designer who was in the class would have to use some new tool or method to solve the problem. No novel-to-you solution was off-limits. Only once in over a dozen iterations of the class did someone just get someone else to do it for them, which I thought was very clever.
The point of the assignment, and the whole class, really, was that you had at your fingertips (at least when you were somewhere that you had access to a computer) a new powerful tool that could help you solve problems. It didn’t matter to your students how you figured out how to put their assignments (or what have you) online, but only that you did it.
People were very concerned that students were using the controversial new site “wikipedia,” because “It was just made up by people.” I was pretty sure that even in its early days, Wikipedia would do OK against a 15 year old Worldbook Encyclopedia, and someone pitted Wikipedia against Britannica and published the findings in Nature, which proved me right.
Now everyone is all atwitter with AI LLMs and how they are going to allow people to “cheat”. Just like they did with the calculator and a whole bunch of other dangerous innovations at least as far way back as the invention of text, when Plato worried that text would “will create forgetfulness in the learners’ souls” or, from another translation “… this invention will produce forgetfulness in the minds of those who learn to use it, because they will not practice their memory.”
So here we are again.
If I were still teaching that class, I’d likely include some more “impossible” tasks (for non-technical people), like creating a web site that implements some game, or writing a Python program to do something. Those are ideas that I got from reading Ethan Mollick’s Co-Intelligence: Living and Working with AI. I do recommend it.
What am I doing with AI?
A month or so ago I finally turned on CoPilot in VSCode (I find myself using it for most of my programming work, but I still have an Emacs window open pretty much all of the time). I continue to be surprised at how good it is, and especially when I’m moving around from Ruby (which I know pretty well) to Javascript (which I still have some struggles with) and CSS (which still remains largely mysterious).
I’ve also been very successful asking it to help out with other programming chores like constructing a SQL subselect (I am not entirely sure they were covered in my database course back in 1986) or pasting in some random error from somewhere and had it do a pretty good job of not only explaining the problem (which I might have understood, or quickly understood from a Google search), but also producing a solution with my exact context like variable names.
The other thing I’ve experimented with is using the Discourse AI Plugin to create some Personas that provide students feedback on some assignments for a class someone I know is teaching. The idea was to have the AI provide a first round of feedback under the guidance of a prompt that we designed. The course was mostly about one part of a paper, but students can’t write just the one section, so having the AI help out with some of the more rote pieces seemed like a good idea. We never got far enough to actually give it a spin, but maybe next semester. One of the things that I found most appealing and compelling for an academic is not merely getting students to turn in better first drafts, but also having a record of what the AI told them and what they did as a result of it, all recorded and ready for analysis. If building some prompts and making your students use them is something that’s appealing to you, let me know and I’ll see how I can help.