Stanford commits $10m to redesign undergraduate majors for the AI era

Stanford is committing $10 million to help departments redesign undergraduate majors around the abilities graduates will need as AI becomes part of study and work. The initiative moves the question beyond individual assignments: departments will examine how knowledge and skills develop across an entire degree, including what students should demonstrate independently and what they should do with automated assistance.[1]

Announced on November 3, the provost-funded programme offers grants of $250,000–$750,000 for one- to two-year projects. Departments can begin applying in winter quarter 2027, with applications considered on a rolling quarterly basis. Stanford expects to support about 15 departments over a programme running from the 2026–27 through 2028–29 academic years.[1]

Why a departmental redesign is different

One lecturer can introduce an AI-supported assignment without changing what students learn in the next course. A degree programme has more difficult dependencies. If first-year students use AI to solve problems once intended to build foundational skills, the second-year course must know which independent abilities students still possess. If advanced courses expect students to supervise automated systems, the programme may have to teach how to verify and challenge machine outputs explicitly.[1]

Stanford's new initiative invites departments to examine this progression from introductory courses to capstone work. It builds on earlier seed grants aimed at individual classes. The new scale matters: a programme can decide where AI use is appropriate, where it should be restricted and how to make those decisions consistent across courses, rather than leave each teacher to improvise unrelated policies.[1]

The announcement contains an important counterweight to the rush toward compulsory AI literacy. Departments may use assessment methods that deliberately exclude AI, including in-class activities and oral examinations. Those are not necessarily signs of hostility toward technology. They can test whether students understand a method, recognise an error or make an argument they can explain without automated assistance.[1]

Consider an engineering major as a hypothetical illustration. AI might help students explore design alternatives, write simulation scripts and organise measurements. But a critical assessment might ask each student to derive a safety constraint or explain why a numerical model fails under unfamiliar conditions. The design objective would be to retain independent judgement while allowing useful automation elsewhere. The university has not mandated this precise engineering curriculum; it demonstrates the kind of curriculum-level choice the initiative enables.[1]

Supporting changes across departments

Stanford says the Office of the Vice Provost for Undergraduate Education will manage the grants with support from the Stanford Accelerator for Learning. A multiyear programme can give departments time to align introductory courses, advanced requirements and capstone work, and share approaches with other disciplines. The published grant range and expected department count describe the intended scale; individual awards have yet to be made through the application process.[1]

That offers an opportunity to move beyond surveying student opinions about AI. Departments could define concrete outcome measures: success on independently completed core problems, performance on new tasks with AI assistance, ability to detect factual and computational errors, and progression into advanced work. Those are possible evaluation measures, not metrics that Stanford has already reported. The published announcement does not specify a common experimental design for every funded major.[1]

The programme also faces a moving target. An exercise that tests high-level reasoning today may be largely automated by a newer model in 2027. The educational goal therefore needs to be expressed in terms of intellectual responsibility rather than the transient difficulty of a particular homework task. Critical thinking is harder to assess than mere access to a new tool, but it is a more durable outcome.[1]

An oral examination, a supervised problem or a critique of an AI-generated solution could reveal different aspects of a student’s understanding. Departments will need to decide which evidence is appropriate for their discipline and make those expectations coherent across the degree. The grant programme gives that work an institutional timetable and resources.[1]

Report an error or suggest a correction