Reclaiming Judgment: AI and the Fate of Higher Education

Could the revitalization of judgment—as a mode of discernment, evaluation, and taste—offer a path forward for the humanities in the age of generative artificial intelligence?

Bloom’s Taxonomy

While giving a recent session of the Course Design Institute at The P3 Collaboratory for Pedagogy, Professional Development, and Publicly-Engaged Scholarship at Rutgers University-Newark, the higher ed teaching and learning consultant Catherine Clepper shared two figures that prompted this question for me. The first will be immediately recognizable to anyone who has received even a passing orientation in the field of pedagogy:

Source: Vanderbilt Center for Teaching

Bloom’s Taxonomy reflects one of the most dominant and pervasive theories that underlie professional teaching practice. It provides a hierarchical schema for organizing intellectual skills of the cognitive domain. Its core premise is that lower-order intellectual tasks such as remembering facts and understanding ideas are the foundations of higher-order cognitive skills such as applying information in distinct contexts and performing analysis. For example, to compare the relative value of two stocks based on the price to earnings ratio of the companies (an analysis), a person would need to understand the concept of shares, value, and ratios, not to mention remember the formula for determining the ratio. In this schema, creation is the peak of the cognitive summit. To continue with the example taken from our hypothetical finance classroom, an assignment that asked students to develop an value-based investment strategy in a particular sector would allow them to demonstrate the highest-order cognitive skill.

For a few years now, I have thought of Bloom’s taxonomy as a justification for focusing on cultivating creativity in the classroom, that if you focus on incubating the act of creation, students would necessarily have to develop those other skills. Or at the very least, in an environment of time and resource scarcity, a teaching strategy that focused on student creativity would disproportionately yield outlier-type results. Experimenting with some project-based learning approaches has moderated this notion for me. Now, I better understand that even as I want to privilege a final creative act, the course should be organized in such a way that the supposedly lower-level skills are intentionally practiced and secured in the process.

Claude.ai and the Inversion of Bloom’s Pyramid

But then, Clepper showed us another figure:

Source: Anthropic Education Report: How university students use Claude (4/8/25)

This second graph, taken from AI company Anthropic’s self-study on student uses of Claude, shows an inversion of Bloom’s taxonomy. Anthropic analyzed one million interactions between students and claude.ai. The study’s authors found that almost 40 percent of the tasks students prompted could be classified as creation. It makes sense that generative AI would be used for this purpose. In contrast, under two percent of the functions were used for remembering. The now famous MIT Media Lab study, which brain activity in ChatGPT users versus non-ChatGPT, seems to corroborate this point as it found that the those who used a large language model struggled to remember the words of their own work.

What I am most interested in about this graph is the anomaly in the pattern of inversion compared to Bloom’s taxonomy. While evaluation is a higher level cognitive skill second only to creation, only 5.5% of the interactions between students and Claude required the examination of information and an accompanying judgment. Only “remembering” was performed less frequently than “evaluating.”

Why Aren’t We Teaching Judgment?

There is an easy explanation for this anomaly that we must consider. Of all the cognitive skills that teachers ask students to perform, judgement is probably among the least prevalent. It would probably be a mistake to read this graph as an indication of that the AI is good at doing. Rather, it reflects the distribution of the kinds of assignments that professors ask of their students. That judgement features so little in this distribution suggests that today’s college students are rarely asked to practice or perform this core human faculty. This problem, tied as it is to dominance of an instrumental and practical orientation to knowledge that invests in STEM and neglects the humanities, preceded the OpenAi’s rollout of ChatGPT at the end of 2022. While genAI may have no notion of truth, beauty, or justice, we should admit that neither do many of is professors. Or at the very least, we should realize that our understanding of the good has been bracketed out from our professional practice.

The Economic Case for Judgment (and Its Limits)

The exercise of judgement would seem to be the capacity, from a strictly economic point of view, worth investing in to enhance humanity’s comparative advantage in its competition with the machine. Taste, that hard to describe and impossible to quantify quality, is “eating Silicon Valley.” Consider the dialogue between the hosts Kevin Roose and Casey Newton of the New York Times’ Hard Fork podcast that serves as an introduction to the recent “AI Issue”:

“ I used to think that the most important question for people to ask themselves was, “Is A.I. smarter than me?” And now I’m starting to think that a better question is, “Is A.I. more interesting than me?” In the same way that a chatbot is very good at writing a very generic English-class essay about James Joyce, it is also going to be able to replace the median, middle-of-the-road performer at many jobs. But I think there are other qualities that are actually more important and harder to automate. The one everyone out here in Silicon Valley is talking about these days is taste, which they claim is the ultimate antidote to A.I. replacement.”

A Muslim Humanist Tradition of Judgment

To be sure, financial benefit is not the only value that serves as a warrant for the cultivation of judgement. Tradition probably does a better job at that. Let us remember that the traditional role of the Muslim scholar, for example, is the exercise of judgment. That is after all what the very sophisticated system of Islamic legal reasoning is all about: which actions, behaviors, and social arrangements are obligatory, and which are forbidden. And in between these absolutes of the imperative and the prohibited are degrees of permission, recommendation, and disapproval. Of course, the Muslim scholar based these judgments on a scripturally-based morality that defined the very meaning of knowledge but also required cultural understanding and called on the scholar to explain and justify to his or her peer.

In Shaykh Musa Kamara’s writings, we find a classical model of judgment anchored in a moral order, cultural understanding, and scholarly accountability. Whether or not we share his religious commitments, his intellectual method reminds us that judgment is not only possible but necessary if education is to remain truly higher. Reclaiming judgment may be our best chance—not just to resist automation, but to renew the ethical and intellectual ambition of the university itself.