It comes as no surprise, but some pretty robust research by Michael Gerlich at the Centre for Strategic Corporate Foresight and Sustainability (who comes up with these names?) in Zurich has confirmed what most people already knew: there is a "significant negative correlation" between the use of generative AI (think ChatGPT) and critical thinking abilities.
So, higher dependence on AI is associated with lower critical thinking scores, particularly among younger people between 17 and 25 years of age. You'll note I say "associated with" rather than "results in", because technically the research does not show causation, merely association, but I think we all know what is really going on here.
This is not the only such study suggesting the same. A KPMG study last year found that 59% of post-secondary students are using AI in their school work, and two-thirds of those students admit that they don't think they are learning as much or retaining as much knowledge. A Chinese/Australian study found that using AI gave students a "short -term boost" but "long-term skill stagnation" (and many participants used the AI to cheat in the study despite being specifically directed not to!) Most teachers and lecturers will tell you the same, anecdotally.
Part of the problem is that many students have a cynical attitude towards university and college education in the first place: they are not there to learn for learning's sake; they just want a qualification to get them into whatever job they have set their minds on, and they will do that in any way they can, preferably with the least effort possible. Imagine teaching in such an environment! "It's terrible to teach in times of AI", remarked one weary professor. But an aversion to putting in unnecessary effort is a general human failing too.
Of course, this debate goes on whenever any new technology comes in, whether it be calculators, computers, the internet, GPS, etc. Hell, it goes all the way back to Socrates, who warned that the newfangled fashion of writing things down would surely erode out memories and our debating skills (although, as the article points out, we only know about Socrates' views because Plato wrote them down!)
But that doesn't mean that the debate is moot. It has been proven that people who regularly use GPS to navigate their way through life have atrophied hippocampus regions in their brains (the hippocampus is the seat of our spatial memories, but also of a bunch of other learning and memory functions). And yes, the generations brought up using calculators are indeed worse at mental arithmetic. (I know, I am one.) It makes sense.
There have been attempts to play down the negativity. Teachers back in the 1970s argued that allowing students to use calculators for rote arithmetic freed up brain-power and time for them to focus on more complex and challenging mathematical concepts, which may have been true, at least in some general sense.
In the same way, researchers and educators have looked for a silver lining in AI. About the best they have come up with is that students could use AI as a kind of intellectual sparring partner, to push for evidence, alternative views and logical gaps before writing up their answers. They can bounce ideas off computers just as they can with their fellow students or professors. But this is not a natural habit, and students would need to be taught this new way of learning. And anyway, what's to stop them from going the whole hog and getting the computer to write the essay entirely?
It's a thorny problem, and not one that's going to go away. AI is being incorporated into everything, whether we want it or not. OpenAI recently gave post-secondary students in Canada and the US access to a premium version of ChatGPT for a limited time. These are commercial companies that want to spread their products as widely as possible. The horse has well and truly left the barn, and the genie is out of the bottle; there is no going back.
The first step is to recognize that there is a potential problem. But how we deal with it now is anyone's guess - I haven't seen a convincing solution yet.