The Hidden Obligations of Teaching

In my earlier essay, I argued that teaching and learning are related but distinct, and that instructors quietly embed assumptions about student preparation, motivation, and resilience into course design. This essay extends that argument. If learning belongs to the student, teaching must involve more than possession of knowledge or performance of expertise. In contemporary education, the professor’s work increasingly includes the design of structured variance, where multiple representations of knowledge, multiple forms of practice, and multiple pathways of engagement can allow different kinds of learners to reach the same core standards. What appears as accommodation from the outside is often simply the growing hidden obligation of serious teaching.

Pedagogical traditions have long moved in this direction, even if they use different language. Adaptive learning emphasizes the adjustment of instruction to learner needs. Scaffolding literature treats the teacher as someone who provides contingent supports and then fades them. UDL (universal design for learning) starts from the premise that no single mode of representation works best for every learner. The point is not that every student needs a separate course, but that a serious course cannot assume an idealized learner mold and call the result rigor.

Knowledge: From Scarcity to Abundance

In the past, access to advanced knowledge was more limited. I still remember wandering around the library trying to look up research articles from physical journals or asking to make copies of some rare textbooks. The pox upon those who borrowed class-related books and did not return right away! In this setting, professors’ institutional role naturally tilted toward gatekeeping. I think this is not due to maliciousness, but rather it is due to natural incentives. If knowledge is institutionalized and not broadly public, then professors are rewarded with the possession, custody, and transmission rather than usability of learning materials. It is the responsibility of students to transform these materials into their own knowledge.

Now that knowledge is abundant, that older arrangement is harder to defend. Students can access lectures, notes, tutorials, papers, demonstrations, and increasingly competent AI explanations with little effort. If so, the value of institutions and professors can no longer rest mainly on being where the knowledge lives. Their value must lie more in how knowledge is prepared, represented, practiced, tested, and made usable for real learners. This does not remove responsibility from students. It changes what professors can reasonably claim as the substance of teaching. After all, what is the point of going to a regional university when one can take online courses from MIT or Harvard for free? Students do not come to a regional university merely because information is unavailable elsewhere. They come for guidance, structure, accountability, feedback, community, and credentials grounded in informed human judgment and support from their professors.

MOOCs emerged as a serious alternative to traditional knowledge gatekeeping by treating access and modular design as central conditions of learnability. Platforms such as Coursera and edX greatly expanded access, but access alone is not the same thing as preparation. Much of online learning still treated personalization too narrowly. Breaking content into smaller pieces, or offering institutional variations of roughly the same material, is not the same as preparing multiple legitimate representations for different kinds of learners within the same course. MOOCs solved distribution better than they solved situated adaptation. That remaining work often falls back onto instructors. It is rarely spelled out in university contracts or departmental bylaws, but it has become one of the hidden obligations of modern teaching.

Teaching as Interface Design

As teaching and learning are distinct activities, the transfer of knowledge is done through a set of contact points between learner and subject. These include things like syllabus, explanations, examples, pacing, assignment structure, office hours, grading logic, feedback loops, and chances for revision. Metaphorically, as part of the teaching process, the professor is designing the interface through which the students encounter the subject. As Ben Shneiderman noted in his seminal discussion of universal usability, designing for a broad audience of unskilled users is a far greater challenge. In other words, interface design becomes most demanding precisely when the user base is heterogeneous rather than idealized.

The analogy here is less about graphical interfaces than about API (Application Programming Interface) design. Several familiar principles of API design map surprisingly well onto teaching. A good API preserves a stable contract while allowing flexible invocation. A good course preserves core standards while allowing multiple modes of access and practice. APIs often support overloaded signatures, optional parameters, wrappers, and adapters so that heterogeneous callers can still reach the same underlying functionality. Teaching does something similar when it offers multiple representations of a concept, scaffolded practice, prerequisite refreshers, revision opportunities, and alternative formats for engagement. The point is not to make the interface infinitely permissive. It is to make legitimate use possible for more kinds of users without changing what the system fundamentally does. This is why some traditional models of teaching feel increasingly inadequate. They are built as though every student were the same function caller: equally prepared, equally fluent in the local conventions, equally able to infer missing documentation, and equally tolerant of opaque failure. When such courses produce confusion or attrition, the result is often treated as evidence of rigor. But from the standpoint of interface design, it may simply reflect a brittle interface that was never designed for real user heterogeneity in the first place.

The flexibility and availability of varied learning interface do not imply that every student will meet the standard simply because the interface is well designed. Even in a course with clear syllabus, detailed lectures materials with many examples, scaffolded practice, and responsive feedback, some students will still not be ready, able, or willing to meet the demands of the subject within the available time. Good teaching design should not be understood as a promise of universal success. Rather, it is a commitment that failure, when it occurs, should come after the standards were made legible and the learning pathways were made reasonable. In that sense, another, perhaps also hidden, obligation of teaching is not to eliminate failure, but to ensure that failure is intellectually meaningful rather than merely the byproduct of opaque design.

It is clear then that good teaching is not easy. In fact, the analogy to API design clarifies just how demanding it is. The difficulty, of course, is that good teaching-learning interface design is labor-intensive. It asks professors to do more than know their subject and sort the students by performance. It asks them to anticipate behavioral variation, build alternative learning routes, diagnose student failures, and revise the interface over time. That is serious work, and it competes with research, service, and the many small bureaucracies that already dominate much of one’s academic life. Once teaching is understood as interface design under heterogeneous conditions, the next question is not whether such work matters, but how professors are supposed to sustain it at scale.

The Role of AI

This is where AI becomes relevant, not as a substitute for teaching, but as a way to reduce the added labor cost. Some of that is about supporting routine tasks such as first-pass checks for minor requirements, OCR of handwritten work, or preliminary sorting of obvious errors so that faculty attention can be spent where judgment matters more. Some of it is generative by nature such as drafting alternate quizzes, producing rubric skeletons, suggesting lecture variations, or red-teaming course materials from the standpoint of a confused student. It could help with analytical tasks such as surfacing patterns in student errors or identifying changes in performance over time. None of this removes the need for faculty judgment. The point of AI here is not to eliminate the professor, but to make the design and maintenance of pedagogical variance more feasible under real conditions of academic labor.

Why should professors do this? Not because every tradition is obsolete, and not because AI is about to make faculty disappear. The better answer is that as knowledge becomes more accessible and pedagogical support more automatable, the distinctive value of professors shifts toward the things that are hardest to commodify such as informed judgment, responsive support, intellectual sequencing, and the ability to decide which forms of variation genuinely serve learning. Faculty who refuse that shift do not protect teaching. Instead, they risk reducing their own role to the parts of teaching that are easiest for institutions to standardize, outsource, or automate. AI matters here because it may change not the purpose of teaching, but the cost of carrying out its previously hidden obligations. If AI-facilitated pedagogy is becoming part of the professional baseline for teaching, then any doctoral programs that prepare future faculty cannot treat it as an elective curiosity. UNESCO’s framework already treats AI pedagogy, ethics, and professional learning as teacher competencies rather than optional extras. The question is no longer whether future educators will work alongside AI-mediated systems, but whether they will be prepared to govern those systems with enough judgment to keep pedagogy human-centered.

Conclusion

The concluding thought of this essay is that the deeper issue is not about whether professors should remain central to teaching in the age of AI as they definitely should. However, the basis of that centrality is changing. It lies less in controlling access to knowledge and more in exercising informed judgment over how knowledge is disseminated for learners who do not all arrive through the same door. If teaching and learning are distinct, then teaching can no longer be defended as mere presentation. It is design work. And increasingly, it is design work under conditions of visible heterogeneity.