Teaching

Fudan University

Since 2024 Fudan has been rebuilding teaching around a “future teacher / future classroom / future learning centre.” In the room: a voice-woken AI lecture companion (live Q&A, screenshot tutoring, student “I don’t understand” flags) while the human is teaching.

Alongside that, an AI-BEST course ladder: more than 100 reconstructed courses, 13,000+ enrolments, 41 “X+AI” dual-bachelor programmes. President Jin Li: AI is not MSG in the dish; it is the main ingredient. Example: “AI Archaeology” does not exist if you remove the AI.

Owner: academic affairs plus the teacher-development centre.

Source: Jin Li, April 2026

Ohio State University

Every undergraduate, beginning with the Class of 2029, must graduate fluent in the major and in using AI inside that major. The required General Education Launch Seminar now carries three AI-infused modules. Colleges then map where, in existing or new major courses, students use AI on disciplinary tasks.

The university says this adds no extra credits. Roadmaps are meant to reach about 53,000 undergraduates. Instructors are not required to put AI in every course; colleges choose which courses carry the outcomes. The Drake Institute owns the pedagogy (course-design institutes, an endorsement, a CarmenCanvas course assistant).

Source: Ohio State AI Fluency launch

Arizona State University

ASU built CreateAI so faculty put custom models into courses (FERPA-aware, Canvas-integrated, many models, RAG). Verified in-course uses include simulated-patient avatars in medical education, biochemistry students building on course data, and a fashion studio that spends the semester building AI experiences.

Scale as of March 2026: 20,000+ users, 7,000+ AI experiences, about 1,000 active Syllabots. ChatGPT Edu is the secondary, research-oriented track; teaching data is meant to stay on CreateAI.

Source: CreateAI, 18 March 2026

Shanghai Jiao Tong University

The institutional teaching plan is the 2024–2026 “AI + education and teaching” action scheme. The piece that is actually in the room is the Teaching Development Center’s “human tutor + AI teacher” classroom programme. First wave, autumn 2024: ten benchmark courses (language, design, physics, information, mechanical engineering).

The stated division of labour: the AI teacher takes knowledge transmission and testing; the human teacher takes higher-order thinking and affective support. Digital humans and AI classrooms are in the support stack. This is still a funded cohort.

Source: 2025–26 classroom plan

University of Surrey

From September 2026, every programme (foundation year through postgraduate) is redesigned so AI is used inside the subject. Documented moves: politics students interrogate a chatbot against election data and theory; civil engineers use a model for a design stage, then verify by hand calculation; business students take an AI first draft and rebuild it as managers.

Applies to students already on course. Brand-new: as of late August 2026 this is days from first teaching, so the evidence is the redesign.

Source: Surrey, 27 April 2026

National University of Singapore

ScholAIstic (NUS AI Centre for Educational Technologies) lets instructors design course-specific role-play and Socratic agents. Social-work client simulations run as the skills lab. In a dentistry course, students must complete five X-ray cases with the agent before the tutorial; the live hour is then used for what the bot cannot do. That is a contact-hour flip around AI.

Scale: about 100 users at launch (May 2024); 6,000+ users across faculties and other institutions by May 2026.

Source: NUS News, 28 May 2026

Carnegie Mellon University

Eberly Center’s GAITAR programme funds teaching-as-research in live courses (fellowships; one published cycle of 27 fellows). Example: in Mechanics II, students tutor a deliberately wrong AI; they outperformed the prior semester on three of four concept areas.

The School of Computer Science is moving cognitive work back into the room because homework is now AI-solvable (in-class tests, TA code-review interviews). Dietrich’s DARE platform: 12+ classes since fall 2024, 2,000+ users. Scale is dozens of courses. Strongest evidence-based faculty programme in this set.

Source: CMU News, 27 August 2026