Colleagues — this week's theme is the question universities keep avoiding: what are we for? A system president in Wisconsin lost his job partly over an AI strategy his board found hollow; the Chronicle's new print issue asks whether the research university's postwar bundle survives AI at all; two more pieces show the gap between the critical thinking we advertise and what we actually teach, and how students privately feel about the tools they can't stop using. Meanwhile OpenAI spent the week recruiting on campus at both ends — 100,000 researchers above, paid undergraduate organizers below. The middle of the institution is where we live.
Newly released public records, reported by the Chronicle's Nell Gluckman, show how far disagreement over AI drove the University of Wisconsin regents to oust system president Jay Rothman. During a February board presentation on AI, frustrated regents texted each other — “Pulling teeth to get him to get proactive with strategy and now this is what we get” — unimpressed that Rothman had handed them a partly AI-generated analysis of AI. By March they asked him to resign; he refused and was fired in early April. One regent told state senators: “There is no regents guidelines or guardrails on AI,” demanding guidelines “not one crafted by Silicon Valley.” Rothman countered that he “was of the belief that a systemwide policy would not be helpful since each university was on top of the evolving situation.” Elon's Lee Rainie sums up the moment: “This is what chaos looks like. It's hard to be a leader in chaos.”
Why it matters for us: the Wisconsin lesson is not “have an AI policy” — it's that boards, faculty, and students now read the absence of a visible process as a leadership failure. NYU's Clay Shirky offers the workable alternative in the same piece: convene “a broad, complicated conversation that has no easy answers.” A campus our size can actually do that — one room, every division at the table — faster than any system office can draft guidelines that are stale by the afternoon.
Read: the Chronicle's report (also in the July 31 print issue)
The big-think essay of the Chronicle's new issue. Gilman (Berggruen Institute, formerly UC Berkeley) argues that Clark Kerr's “multiversity” — research engine, credentialing mechanism, coming-of-age experience, all bundled — is finished, with AI acting as a “catalytic solvent” on contradictions the bundle always contained. His bluntest line: “The term paper, as an assessment instrument, is dead.” What replaces it is live assessment — oral examination, structured adversarial debate, documented decision logs — and a professoriate reconceived as interlocutors rather than lecturers, on the model of the Oxbridge tutorial and “the seminars of many small liberal-arts colleges.” The counterintuitive conclusion: because these formats are labor-intensive by design, “the coming of AI is going to mean there will be demand for more professors, rather than fewer.”
Why it matters for us: Gilman is describing, as the research university's future, roughly what we already do: small seminars, close contact, a liberal-arts core. That's a genuine structural advantage — but only if we lean into the assessment half of his argument. If our seminars still end in take-home essays graded as proxies for thinking, we have the labor-intensive form without the AI-proof function. Worth circulating to anyone redesigning a signature-work or capstone rubric this fall.
Read: the essay in the Chronicle Review
Scott Carlson's Chronicle feature starts from the consensus answer to “what's left for colleges in the AI age?” — critical thinking — and then asks the uncomfortable follow-up: do we actually teach it? José Antonio Bowen, co-author of Teaching with AI, concedes: “We don't teach it well, we don't teach it specifically, we don't follow through, we don't measure it — all of those things are true.” Cognitive psychologist Daniel Willingham adds that critical thinking is largely domain-specific — thinking critically in mathematics is not the same as in English literature — so teaching it as a free-floating general skill has had limited success. Former Carleton professor Louis Newman notes that nearly every liberal-arts mission statement invokes critical thinking, then asks: “Where are the classes that are teaching it?” His fix is disarmingly cheap: name the moves explicitly, in every class, when students make them.
Why it matters for us: our mission language leans on exactly the claims this piece stress-tests. The practical takeaway isn't a new course — it's Newman's move: when a student asks a good question, say why it's good, out loud, in disciplinary terms. That costs nothing, works in a seminar of twelve better than a lecture of two hundred, and doubles as the “AI literacy” employers keep asking about — evaluating an answer instead of accepting it.
Read: “Can Colleges AI-Proof Their Students?”
A Dartmouth creative-writing professor used the confidentiality of the workshop to ask students what's really going on. First answer, hand raised: “Everybody uses it.” Asked what “uses it” means: “Cheats.” The anonymous written responses he later collected “ranged from resignation to despair,” some reading “like substance-abuse testimonies” — students who swore off AI and felt it creeping back, students who feel the terms of an elite education are now “submit to the machine or fail.” Sharlet's sharpest analytical move is the comparison to BP's popularization of the “carbon footprint”: institutions capitulate or dodge, and students internalize systemic failure as personal guilt. His conclusion: this isn't really an essay about cheating, which is “only a symptom,” but about students who feel abandoned by the institutions and the tools alike.
Why it matters for us: every integrity policy we write assumes students are calculating advantage-seekers; Sharlet's students describe something closer to ambient compulsion plus shame. On a small residential campus we can do what he did — ask, in settings where honesty is safe and ungraded — before we legislate. A first-year advising conversation or writing-seminar check-in that surfaces this privately will tell us more than any detection dashboard.
Read: the essay in the Chronicle Review
On July 29 OpenAI launched “ChatGPT for Academic Researchers”: free access to its frontier models — GPT‑5.6 Sol Pro at launch — for researchers at selected institutions, starting with 10,000 this summer (already live at the Institute for Advanced Study and École normale supérieure) and expanding to 100,000 through 2027. Approved researchers can invite up to four collaborators; workspaces carry business-grade privacy, and data is not used for training by default. OpenAI frames it with adoption numbers: each week “roughly 1.3 million people use ChatGPT for advanced science and mathematics, generating about 8.4 million messages.” The program sits inside a commitment of more than $250 million through 2027 for external scientific research. Eligibility is limited to “recognized, degree-granting colleges or universities with a high level of research activity.”
The announcement includes charts on scientific usage by workflow category and by task length — see the figures in the original post.
Why it matters for us: the “high level of research activity” criterion is built for R1s, and it's unclear whether a small liberal-arts campus qualifies on its own — worth someone checking the eligible-institutions list, and whether affiliations through partner universities count. The larger signal: research-computing access is becoming a per-lab perk granted by an AI company, not a library subscription negotiated by an institution. Faculty who apply individually should note what runs through a vendor workspace versus institutional infrastructure.
Read: OpenAI's announcement
The same week, OpenAI opened applications (through August 10, 2026) for its Student Collective: undergraduate “Campus Leads” who work in pairs to run four workshops per semester, weekly “Studio Hours” coworking sessions, and a semesterly showcase — for students “from every field,” explicitly not just technical ones. Leads commit 4–6 hours per week from August 2026 through June 2027 and receive a cash stipend each semester, a ChatGPT subscription, Codex credits, event funding, and training. Eligibility: enrolled undergraduates, 18 or older, graduating after December 2027, studying in the United States, Canada, the United Kingdom, France, Germany, India, Japan, or South Korea — and not simultaneously serving as an ambassador for any other AI company or program.
Why it matters for us: China is not on the list, so our students can't apply — but the design is worth copying. Peer-led workshops, weekly coworking, an end-of-term showcase: that's an orientation-week and co-curricular structure we could run ourselves, with our own norms (disclosure, reflection, critique of the tools) rather than a vendor's. Note also the exclusivity clause: AI companies are now competing for student organizers the way consulting firms compete for campus ambassadors. Better that structure exists on our terms before it arrives on someone else's.
Read: the Student Collective program page
If you try a live-assessment format Gilman would approve of — an oral defense, an adversarial debate, a decision log — or simply ask your students the Sharlet question and hear something we should know, tell me. We are collecting examples for AI experimentation at DKU.
— Yisu