AI in Higher Education Newsletter

July 13, 2026 · Vol. 24

A weekly brief for the Management Department, Isenberg School of Management, UMass Amherst. By Matthew D. Langenkamp / 雷邁德, prepared in collaboration with Thea 🪻✨.

← Back to archive · ← Back to Dictionary

Executive Summary

This week’s higher-ed AI story is the return of controlled learning conditions. Brown’s widely discussed welfare economics case and the University of Chicago Law School’s new AI strategy point in the same direction: take-home, artifact-only assessment is increasingly fragile, and some parts of learning need protected conditions where students think, write, discuss, calculate, and explain without AI at hand.

That does not mean a blanket retreat from AI. UChicago Law’s strategy is especially useful because it combines no-device 1L core classes and oral discussions for substantial research papers with explicit instruction in responsible professional AI use. The design principle is conditional AI: students should learn to think without the tool, with the tool, and about the tool.

For Isenberg, the practical question is no longer “allow AI or ban AI?” It is: which learning moments should be AI-free, which should be AI-assisted, and which should require students to supervise AI outputs professionally? That is a better frame for fall course design than another round of generic syllabus language.

Overview

The public conversation still likes simple categories: pro-AI faculty versus anti-AI faculty, cheating versus innovation, detection versus permission. The better institutional category is learning conditions. A student drafting a strategy memo, practicing a concept, defending a recommendation, using AI as a tutor, and completing an in-class exam are not the same educational act. They should not all receive the same AI rule.

A workable course design now needs three zones: AI-free work for foundational fluency and independent reasoning; AI-assisted work where students may use tools but must disclose, verify, and remain accountable; and AI-required work where the objective is to practice delegating to, checking, and improving AI in a professional context.

1. Brown shows the fragility of take-home evidence

Inside Higher Ed reported on July 8 that Brown economics professor Roberto Serrano suspected widespread AI use on a take-home midterm after unusually high scores in his Welfare Economics and Social Choice Theory course. Serrano then moved the final exam in person. The reported result was stark: the midterm average was 96 percent, while the in-person final average was 48.6 percent; 18 students dropped the course, nine stayed enrolled but did not take the final, and 19 students failed.

The details still need careful institutional handling. Suspicion is not adjudication, and large academic integrity cases create due-process problems for students and workload problems for faculty. But the teaching implication is already clear: if the only evidence is a polished take-home artifact, the assessment may no longer tell us what we think it tells us.

Teaching implication — For AI-vulnerable work, the final artifact needs companion evidence: in-class reasoning, process notes, short oral defense, version history, source checking, or a targeted follow-up question. The point is not surveillance. The point is better evidence of learning.

2. Chicago Law offers a stronger design pattern

The University of Chicago Law School’s July 9 AI strategy is the best current example of conditional AI. Beginning in the fall quarter, the Law School will pilot a general prohibition on electronic devices in core 1L classes, with limited exceptions. It will also require students completing substantial research papers to engage in an oral discussion of their topic, either in class or one-on-one with the professor.

At the same time, Chicago is not rejecting AI. Its strategy explicitly includes teaching responsible, effective, and ethical AI use, integrating AI tools in clinics, and expanding access to professional AI tools where they support real learning and practice. The stated goal is AI-resilient pedagogy: steering students toward uses of AI that strengthen learning rather than inhibit it.

That is a mature position. It recognizes that AI can be a shortcut in one setting and a legitimate professional instrument in another. The distinction depends on the learning objective.

3. Students are ambivalent, and that matters

Inside Higher Ed’s June Student Voice survey complicates the assumption that students simply want unrestricted AI access. The survey included 1,038 students across 203 two- and four-year institutions. About four in 10 students said they are worried about dependence on AI tools, even while six in 10 see AI’s main college value as learning support. A majority, 55 percent, expect AI to negatively affect career prospects.

Only about one in 10 students said their institution is handling AI’s rise very well. Their top desired responses were not “let us use everything.” They were teaching effective and responsible AI use and preserving authentic learning through curriculum and assessment design.

That should change the faculty conversation. Students need clarity, but they also need protected learning. They know, often more sharply than we assume, that frictionless AI help can weaken the very capabilities they came to college to build.

4. Business schools need curriculum-level AI capability

AACSB’s July 7 article on closing the AI gap in business education argues that business schools need coordinated strategy, faculty development, external partnerships, and academic cultures that support responsible AI use. The useful point for Isenberg is that AI readiness is not a tools workshop. It is a curriculum and governance problem.

The department-level question is therefore not just whether a particular instructor permits ChatGPT on a particular assignment. It is what AI-era capability every graduate should demonstrate. Can students frame a problem before using AI? Can they identify bad assumptions in an AI-generated analysis? Can they verify claims and data? Can they explain the recommendation in their own words? Can they disclose tool use professionally?

Those are business skills. AI has made them more visible, not less important.

5. Transparency should become boring and normal

EDUCAUSE Review’s March framework on transparent GenAI use is useful because it separates internal documentation from external disclosure. Internal documentation preserves the record: tool, model, prompts, outputs, edits, validation. External disclosure tells students, colleagues, or reviewers what role AI played, in language calibrated to the level of influence.

That same distinction can help students. Instead of treating AI disclosure as confession, we should treat it as professional documentation. A simple student AI-use note can ask:

This is a small move with a large cultural effect. It tells students that accountable AI use is not hiding the tool; it is documenting the human judgment around the tool.

6. AI-free learning spaces may return under new names

Karen Spira’s July 2 Inside Higher Ed essay proposes “human intelligence labs”: staffed, welcoming, AI-free spaces where students can practice reading, writing, problem solving, and sustained cognitive attention without AI shortcuts. The full proposal raises practical questions about space, staffing, accessibility, and trust. But the concept is useful because it reframes AI-free work as learning infrastructure rather than suspicion.

For Isenberg, the course-level version is more immediately practical than a new physical lab:

The purpose is not nostalgia. It is preserving the cognitive work that AI can too easily bypass.

7. Practical implications for Isenberg

Before fall, the most useful move is to label major assignments by learning condition. A single course policy is not enough if different assignments measure different kinds of capability.

This frame also helps with fairness. Students should know when using AI would undermine the assignment, when it is permitted with documentation, and when it is part of the expected professional workflow. That clarity is better for learning, better for integrity, and closer to the world our graduates are entering.

What to Watch Next Week

  1. Whether more universities move from broad AI statements to assignment-level learning-condition labels.
  2. Whether academic integrity systems adapt to large-scale AI cases without leaning on unreliable detector evidence.
  3. Whether business schools define AI readiness as curriculum-level capability rather than optional tool exposure.
  4. Whether “AI-free” work is reframed as protected learning infrastructure rather than anti-technology policy.
  5. Whether students’ concerns about dependence and careers become central to institutional AI planning.

Questions or topics for next week? Reply to mlangenkamp@umass.edu.

Prepared in collaboration with Thea 🪻✨

Sources

Inside Higher Ed, “Brown Professor Suspects Most of His Class Used AI to Cheat,” July 8, 2026, link.  ·  University of Chicago Law School, “UChicago Law Unveils New AI Strategy,” July 9, 2026, link.  ·  University of Chicago Law School, “Rethinking Legal Education in the AI Era,” July 2026, link.  ·  Business Insider, “AI-enabled cheating is forcing some schools to go analog,” July 2026, link.  ·  Inside Higher Ed, “Why Students Aren’t All In on AI – What They Want From Colleges,” June 11, 2026, link.  ·  AACSB, “Closing the AI Gap in Business Education,” July 7, 2026, link.  ·  EDUCAUSE Review, “From Prompt to Practice: A Framework for Transparent GenAI Use in Higher Education,” March 2026, link.  ·  Inside Higher Ed, Karen Spira, “With AI, Colleges Need Human Intelligence Labs,” July 2, 2026, link.  ·  EDUCAUSE, “The Impact of AI on Work in Higher Education,” 2026, link.

Return to Dictionary All Entries (A–Z) For Students Other Writing