Why do schools and universities want to purchase AI?
Roaring language from OpenAI—they’re planning to invest a whopping 50 million under the banner NextGenAI in 15 American universities! To put that in context, they’re valued at 300 billion, so it’s only about 0.02 percent of their market value. What OpenAI gets in return is clear to me: they boost their image as a cutting-edge tech company while simultaneously undermining the credibility of higher education and scientific research.
Marc Watkins, a professor at the University of Mississippi (one of the 15 institutions involved), has often been overly positive about generative AI in education. Yet he recently penned a remarkably incisive piece asking: why do universities actually want to purchase and integrate AI solutions?
Students are already using AI all the time—and they manage just fine with free tools. Teachers looking to create lesson plans can often make do with existing solutions, for example, those integrated into Microsoft systems for which they’re already paying (and which offer some data protection guarantees). So why pay extra for AI? Watkins argues that “… institutions are doing so under the faulty assumption that buying access gives them greater control over how their users interact with AI.”
In a desperate attempt to take control of the situation—under the guise of “we have to do something, and this is something”—they go ahead and purchase systems and run pilots. The entire article is well worth a read, but that question is one we really need to ask ourselves—and our employers!
If you want a glimpse into how uninspired companies are when it comes to implementing AI in schools, just click on that ad that kept popping up on my LinkedIn all week. Follow the link and enjoy the pitch: “Enabling evidence-based decision-making in education through agentic AI.” It’s obvious the whole piece was machine-translated—you can tell from quirky details like “…the reading habits and performance of fourth-graders.” It even says “fourth-graders” (the AI must have gotten stuck on “4th graders” and kept the superscript)! Yes, a perfect way to convince us teachers that CapGemini is really the way to go!
Do students actually use AI for plagiarism?
And another assumption—that students are mass-using AI for homework assignments. It seems as obvious as day, right?! But research from Spain involving 500 students shows it’s not such a big issue. Only a small portion of students use AI to commit plagiarism, and those are typically unmotivated students who, in “the old days,” would have resorted to creative shortcuts (copying, using book summaries websites, etc.).
I often recall a colleague I once complained about—who noted that when students shared answers via WhatsApp, he said, “If many students are committing plagiarism, then you’re asking too much of them.” And now, research clearly backs that up. Students who want to learn, who see the value in an assignment and have the mental space for it, do the work themselves.
Peer review—Can it be done with AI?
Sometimes I really think I’m losing my mind, and this week that feeling hit again when I read a piece in Nature (!!) about using AI for peer review. The very first sentence reads:
“Do you ever feel that agreeing to review an academic paper guarantees a wasted workday? You’re not alone.”
Unbelievable, right? Of course, peer review takes time, and naturally I’m not thrilled when I have to do it. But that’s not a criticism of peer review itself. At its core, peer review is about contributing to the advancement of science—literally my job. The real issue is that, due to budget cuts, publication pressure, and a threefold increase in the number of students and PhD candidates, I have so many other things to do!
Really, Nature, cancel yourself. If you’re not going to stand up for the protection of the deep intellectual labor that peer review represents, then who will? (Yes, I know this a column; and Nature’s editors might argue that you should be allowed to express any opinion, but it’s Nature—it’s not just any publication, and they don’t explicitly state that it isn’t the editors’ view.)
And if it’s not about reviewing but about research—do scientists want to use AI for new discoveries? No. Google already has to answer for a host of claims regarding inventions.
And what about the people who create the algorithms themselves? Do they really want their employees using them? Also no.
And I almost forgot Women’s Day!
I was a guest at Windesheim last Thursday, giving a lecture on programming and feminism—and it was fantastic (a video will be coming in a future newsletter). As some of you might remember, last year in honor of Women’s Day I spearheaded a project about women on the radio: FemFM.
That’s why it’s especially cool that Buma/Stemra has released a new report on the state of things in the Netherlands. And, this piece is a few weeks old but still relevant! The Dutch for Human Rights has decided that Meta violates human rights by showing women different job ads than men.
How do you debate someone who does not want to listen?
A beautifully crafted, short video explains how to debate with people who say, “Yeah, but research shows that…” without actually reading the research—using it solely as an attack. The advice: shift the discussion towards values rather than cold facts. The video is very well produced!
Movie Tip!
I originally wanted to stick strictly to AI news, but, well, people contain multitudes—and besides, I wasn’t all business on Twitter back in the day, so here’s a little extra. You absolutely must not miss the film A Real Pain, for which Kieran Culkin (you know, Roman Roy—but also… Cousin Fuller in Home Alone!) has just won an Oscar. What a film—laughing out loud at a movie about the Holocaust is something only the truly great can pull off. Culkin’s counterpart, Jesse Eisenberg, who also wrote and directed the film (and, in my opinion, was undeservedly snubbed for an Oscar for screenplay), has truly delivered a unique piece.
This article was translated by Johanna Guacide.
