Mat Messerschmidt
AI, Individualized Learning, and the Future of Education
One of the great impediments to any breakthrough in the current educational crisis, the one being called the AI crisis, is that we are convinced that it started with AI. I was born in 1986, and the roots of this crisis came into the world before I did. Yet the contours of the educational malaise we have been mired in since my childhood did become much clearer to me during the AI years, in talking to college students about the emergence of AI.
When I first became aware of the existence of ChatGPT in the early months of 2023, I immediately discussed its existence with my students. They all acknowledged an awareness of the rules: ChatGPT cannot be used to write papers.
I came to have a far better understanding of students’ attitudes toward the use of AI, however, when I started to talk to students who were not my students. I spoke to students I sat next to in a campus café, who were not in my class. I talked to students at institutions at which I used to teach. I asked younger, college-aged cousins about their attitudes, and their peers’ attitudes, toward AI.
What surprised me, when I started having these conversations, was not that everyone was convinced that cheating was widespread. Rather, what surprised me was the fact that relatively few students invoked the concept of cheating at all. Students who were taking my class for a grade, and who were listening to me explicitly tell them that the use of ChatGPT was cheating, were obligated to echo my message that any ChatGPT use was unacceptable. But these students in cafes or on other campuses across the country or in Europe – students who didn’t need to worry in the same way about what I would think about their response – tended to tell me, with apparent honesty and clear consciences, that they were using ChatGPT, but not in the way that would be cheating. They were using AI to find passages, or to summarize chapters that they’d read, or to brainstorm paper arguments, or to find out what had already been said about an author, or to translate into English what they had written in their native language, or to find weaknesses in authors’ claims. None of these AI usages, they were convinced, counted as cheating. This sentiment repeated itself across different kinds of learning institutions, different ethnic groups, different continents.
It would be easy to decide that students who say this are being disingenuous, that they do in fact recognize that such actions at least border on the illicit. Such a conclusion would allow anyone over 30 to lazily label the current educational crisis as a generational problem: Gen Z’s don’t want to do the work that the rest of us did.
This response would be wrong, because the entire course of educational discourse over the last several decades has prepared today’s students to respond to the availability of AI exactly the way they have. Students (and parents) have been told that, in accordance with what was once called the theory of multiple intelligences, everyone learns differently. This implies that the greatest sin of the schooling model that has been with us over the lifetime of everyone currently alive is that its basic mode of operation is to have 25 students, facing the same direction in standardized desks, perform the same task, the same way, over roughly the same amount of time. The default sales pitch of elite (i.e. private) K-12 schools has become that of individualization: the best schools boast student-teacher ratios that allow the educational needs of your student to be professionally addressed apart from the rest of the crowd.
It is not wrong to see benefits in educational individualization. In fact, in more traditional educational contexts, those benefits sometimes need to be highlighted more than they are. When I taught K-12 in Romania between 2010 and 2014, I sought to emphasize the relevance of individualized learning plans for certain struggling or excelling students. It seemed like an American educational idea that could carry great benefits in Romania.
But it is also easy to see how such messaging can lead naturally to the deterioration of standards. Every schoolteacher will be familiar with the experience of being told, by the parents of a child with behavioral or motivational problems, “My child learns differently.” In other words, “You expect my child to learn your material your way, but my child will only learn the material if it is approached in some way that caters to the disposition of my child.” The problem is that this mantra, which many teachers in the Western world have likely bowed to at some time in some way, must ultimately change not only the means to learning but also, in the end, the substance of what is learned. We can argue that visual learners ought to be given the chance to act out scenes from a 350-page novel in order to help them visualize the dramatic action, but preparing and completing a skit takes time, and a variety of such activities may ultimately force the 350-page novel to be replaced by a 150-page graphic novel. In addition, to engage seriously with texts with no visual element could be said to be a core educational skill, one for which visual learners should work harder than a “reading/writing” learner might have to. But such a claim has been made retrograde by the current educational paradigm.
In the United States, the scope of the conviction that different students require different learning settings has dramatically expanded in recent years to cover large swaths of new territory. Many students are given extra time on “standardized” tests after having tested for a learning disability such as ADHD (of course, such tests are no longer “standardized” if students receive differing amounts of time to complete them). Other students have university-approved notes from a medical professional informing professors that they must be exempt from classroom rules such as a prohibition on laptops – or, again, that they must receive extra time on exams. Anyone who has been through a liberal arts education at a competitive 4-year college will recall that getting all the hours and hours of weekly reading completed was a challenge for everyone, including comparatively fast readers. If students need extra time on a standardized test in order to get into an elite college with heavy reading loads, what choice do instructors then have at such institutions when those students are filling their classes, other than to decrease the reading load – other than to lower standards, in other words?
The point I have been leading to is this: the many current college students who feel justified in using ChatGPT for the spot-treatment of their own individual “needs” are simply repeating the logic of the messaging that the adult world has been delivering to them since they first started school. Their parents told their teachers that their student needed individualized learning approaches. Other adults arranged for them to have individualized time constraints on tests. Now that they are responsible for themselves, they are taking the same individualized approach to AI. I am not saying that this does not represent a critical collapse of standards, because it does – what I am saying is that this collapse of standards is the result of this generation of students doing precisely what earlier generations relentlessly trained them to do, and not the result of some capricious decision by Gen Z to flaunt the rules.
We have gone too far in training students to believe that their failure to meet learning objectives is the result of an incompatibility of their learning style with the methods and rules of the classroom. This approach has resulted in students whose instincts have been trained to hear alarm bells whenever something becomes difficult or uncomfortable – whenever the task demands that they adjust, as opposed to the task being immediately adjusted to accommodate the students themselves. Rather than holding themselves responsible for becoming equal to the task, they have been told, in effect, that the task must come to them, as the individuals with the capacities and inclinations they currently have. In the context of this cultivated frailty, AI is less the dawn of a new educational dystopia than it is an accelerator of the train we chose to board a long time ago.
I said earlier that when I arrived in Romania to teach school, I sought to bring an American flexibility into my classroom. But, at this point, American education – and perhaps, I think, much of Western education – needs to emphasize inflexibility in certain ways: there are certain tasks that students must be able to perform in fairly conventional ways in order to say that an educational program has succeeded. Reading and reflecting on a book ought to mean reading and reflecting on a book – without necessary recourse to skits, movies, visuals, or SparkNotes – and without recourse to AI. It might have been easier to forbid the final element in this list of educational crutches – of tools, rather, that have become crutches – if we had not been treating all the other ones as basic components of the classroom for decades.
INDICAȚII DE CITARE
Mat Messerschmidt ,,AI, Individualized Learning, and the Future of Education’’ în Anthropos. Revista de filosofie, arte și umanioare nr. 1 / 2026
Acest articol este protejat de legea drepturilor de autor; orice reproducere / preluare integrală sau parțială, fără indicarea sursei, este strict interzisă.


