S6.4 - University Teaching and Learning in an AI World

Well, the first thing that I should say is that AI is an umbrella term for various kinds of technologies that we associate both with artificial intelligence in the sense that it comes from programming in some basic sense, and also that has been represented as intelligence.

We don't have to go through the whole... history of how this term arose, but in my view, it's originally a marketing term for a set of, a whole range really of specific technologies that aren't intelligent.

You know, you don't think your thermostat in your house that you turn up and down or that to regulate the heat or that turns it on and off for you when you're not there to do that is intelligent.

but it is reactive to its environment on the basis of code that's been written to respond to certain, do certain things under certain conditions.

Okay, so the kind of AI, as it's come to be called, that we're going to be talking about in the context of the university, is, for lack of a better word, a large language model.

And I assume that you've had other episodes that have laid this out.

So I'm not going to say much more about how those work.

But I think the important thing to realize at the outset is that AI in the university is a large language model that is inserted into an institution that is constituted by or operates through the production of language.

So that's not just the sense of management in which email constitutes people's relationships.

It's also in terms of teaching and research.

So the core activities of universities operate through language in a way that is not the same case, say, in the construction industry or the design industry, architecture, although lots of code is involved in that also.

So it's an institution that is vulnerable to, as well as can take advantage of, large language models that essentially produce automated writing.

That's kind of the core offering.

And so in terms of the question of where is AI in the university, the answer to that is everywhere.

So it's in course management systems that for 20 or 25 years have become kind of the backbone of course instruction, where students can go online and get the readings for the course.

They can get supplemental material.

They're often this site where exams or quizzes take place where teaching assistants interact with students.

So those are now platforms for machine learning of various kinds that comes out of these models as LLMs.

It's also gotten inserted into Student advising, so many of us picture students as going to a university where they go to the office hours of their professor, they make an appointment, and they can talk to their professor about their coursework, how their different courses fit together into an overall pattern.

how they're succeeding or not succeeding in particular courses, how they could do better.

In other words, that there's a person that they can talk to about their intellectual lives and how education is contributing to that or not, and then do course corrections or whatever needs to be done.

And much of that has now been automated at a lot of different universities.

You can go online and look at your transcripts.

But it's gone way beyond that.

So in places like Arizona State University, essentially your curricular interaction is mediated by technology that tracks what you've done, makes suggestions about where you should go, tells you not only what required courses you haven't done yet, but also makes suggestions about...