Today is just a quick round-up of recent news and trends re: artificial intelligence in higher education – hopefully one which is a bit different from the everything-is-awesome/everything-is-terrible style of think pieces that you often see on this subject.
Artificial intelligence is having significant impacts in fields like astronomy and molecular biology, and large language models quite unexpectedly seem to be capable of making significant contributions to mathematics. In other fields, AI does not eliminate any steps in the research process: what it can do, rather, as Yale economist Paul Goldsmith-Pinkham noted here is to compress each one of them, making it possible to do more research in total. This assumes, though, that the datasets one is using are relatively clean, which is a big if in some cases. Not all fields are equally susceptible to this approach because it’s not equally good at all eight phases and not all fields use these various steps to the same degree (anything involving some variation of storytelling – cultural studies, history, and even political science to some degree do not fit this model very easily, nor do any fields involving artistic creation). So, it is not universally applicable and the effects of AI vary from one field to another.
Which is perfectly fine: it would be stranger, perhaps, if a new technology’s impact was perfectly equal across disciplines.
On the basis of these kinds of results, countries are going all-in on artificial intelligence. The United Arab Emirates, Kazakhstan, and Ethiopia, among others, have set up entire AI universities, while India has created a university which is “AI-powered” (no, I am not 100% clear on the distinction either). Korea is investing hundreds of billions of won to create specialized graduate schools in AI, as well as to pay dozens of universities to provide training to workers in AI fields. Pakistan is making it mandatory across all fields of study to complete at least three credits of training in artificial intelligence. India is setting up 500 new AI programs in its universities.
Notice that all those examples are on the other side of the globe. Asia – plus Ethiopia – seems to have quite a different approach to artificial intelligence. Maybe more than that – artificial intelligence is actually causing a global split on the relationship between higher education and disciplines. Where we see this most directly is in China, where the rise of artificial intelligence is causing the country not only to open many new majors in AI and other areas of technology, but also to start cutting disciplines where, shall we say, it looks like artificial intelligence might render careers obsolete (most notably in translation). This isn’t just one or two institutions doing it – this is government policy, and it will progressively be rolled out nationwide in a couple of years. Similarly, Indonesia has put a moratorium on new programs in the humanities. In Asia, the rise of AI and the need to reduce emphasis on humanities sometimes appear to be two sides of the same coin.
Compare that to North America and Europe, where a diametrically opposed argument tends to be made: that AI makes the humanities even more important (see also here), based on a rhetorical positioning of humanities as a bastion of “critical thinking” – which it surely is, though not obviously more so than any other set of disciplines. The Asian argument and the North American argument can’t both be right. Though to be fair, they could both be wrong: it might be that we need fewer humanities graduates and higher humanities content in degrees across the board. It’s interesting that no one seems to want to make this argument, though.
In any event, the humanities and social sciences also have a completely separate set of issues with artificial intelligence. Unlike in STEM fields, assessment in arts faculties tends to rely on assessment of writing. And writing is the one thing that LLMs do well in the sense that they can produce a lot of output which (with a modest amount of editing) is not reliably detectable. Use of LLMs is therefore seen (uncharitably) as cheating or (somewhat more charitably) as a serious challenge to the economics of assessment. Put in the context of considerable evidence, for instance, that use of AI interferes with students’ ability to actually master skills (a phenomenon sometimes attributed to “cognitive surrender”), this all leads to claims, such as those that have been present in Portugal and the Netherlands, to ban AI in classrooms, lest the technology create a generation of “digital idiots”. I am not sure how helpful any of this is: my own feeling is that we just kind of need to get on with finding alternatives to the essay as an assessment mechanism, but we need to be mindful of the politics of all this.
But here’s a point I think about a lot, especially as the issue of transnational education grows more important (to Canadian institutions at least). What if, over the next decade, Asia continues on a different pathway than North America and Europe, and its institutions become comparatively much more AI and STEM-heavy. And what if China’s lead in AI and other STEM fields turns into an overall technological and economic dominance, much as Japan had (ephemerally at least) in the 1980s? What does that do to regional students’ perceptions of what a university is and should be?
And then: if North American/European universities don’t adhere to a more AI-heavy model of higher education are they likely to be able to convince students in Asia to attend their institutions over, say, Chinese ones? Doesn’t the transnational model fall apart if the countries trying to sell their services aren’t automatically seen as being superior because of technological prowess, etc.? This, of course, is not a reason to abandon a particular model of higher education and disciplinary distribution; it is, however, to say, that we shouldn’t fool ourselves into thinking that our model of higher education will permanently have such cachet that we can charge a premium for it.
Meanwhile, over on the non-academic side of the house, I think it’s fair to say that rapid productivity gains from AI aren’t materializing very quickly. There have certainly been some interesting uses of AI chatbots to nudge students in ways that improve student retention (though this initiative pre-dates the LLM boom of late 2022) and to speed up credit transfer decisions. I hear some institutions are doing things like piloting the use of AI in setting up research agreements, but we haven’t seen much in terms of AI actually replacing jobs in higher education. What I havedefinitely heard of, though, is institutions using AI to replace HR consultants in downsizing at institutions in financial distress : instead of having NOUS or someone work out who needs to go, just throw your budget, org chart and salary data into Claude and see how it recommends you trim your staffing with the least disruption possible. Whether this is a step forward or not is in the eyes of the beholder, I guess.
Anyways, there you have it – my state of AI in higher education. I hope you enjoyed it and hope you can join us this November for our AI-cademy conference in Vancouver on November 9-10. There’s a lot to talk about, and a big future still to be shaped.
