As the university’s Academic Integrity Working Group proctoring pilot enters its third year, the group has partnered with faculty and departments across campus to update guidance for instructors on generative AI use and test administration.

“Our work with campus partners is helping instructors preserve the importance of independent student work while also making sure our students are prepared for a future in which AI tools will be integral to many professions,” said AIWG Faculty Co-Chair Jennifer Schwartz Poehlmann.

The Academic Integrity Working Group (AIWG) was formed in winter 2024 to study Stanford’s academic landscape and identify the scope of academic dishonesty, including its root causes and its relationship to teaching practices. A multi-year pilot study of equitable proctoring practices launched in spring 2024 with seven courses. More than 50 courses across multiple schools are participating in the pilot this quarter.

“The AIWG’s work reinforces the university’s commitment to advancing academic integrity initiatives and leverages many campus partners,” Provost Jenny Martinez said. “As strategic campus partners in academic integrity, departments can lead these conversations, developing approaches that reflect the unique contexts of their disciplines in a changing educational landscape.”

As professors across the university prepare for midterms and exams this fall, the AIWG, with support from the provost’s office, has several recommendations to share with faculty. The group is also calling for departments to consider their role in initiating critical dialogue and discipline-specific policy around generative AI.

Proctoring and in-person assessments

For courses not participating in the proctoring pilot – which is currently the only place proctoring is permitted – the AIWG website outlines best practices for faculty conducting their own in-person exams, ranging from seating strategies to recommendations around cell phone use.

“The proctoring pilot remains an important tool for creating consistent and equitable testing conditions,” Poehlmann said. “However, we want instructors to also feel empowered to create their own strategies for managing in-person exams and know that there are tools available to help.”

For instructors who wish to limit students’ use of AI in specific circumstances, including for high-stakes assessments, the group recommends in-person formats such as oral exams and in-class writing assignments. In their study of generative AI in higher education, the group identified several issues with tools designed to detect AI use in student work, including bias and occurrences of false positives and negatives. They also found that these tools often fail to reliably assess the extent of AI involvement in texts that mix AI and human writing, making them unsuitable for high-stakes situations, especially as evidence in academic misconduct cases.

Departmental policies and preparing students for an AI-enabled future

The AIWG emphasizes that the most effective approaches to student AI use go beyond individual course and exam management.

“Departments and multi-section courses are encouraged to establish consistent and enforceable AI policies that can be applied fairly across sections. These policies should be reasonable, communicated clearly, and discussed with students regularly throughout the academic year,” Poehlmann said.

The AIMES website, launched by the Center for Teaching and Learning (CTL) and VPUE, provides guidance and examples of how AI can be integrated into coursework to support student learning and career preparation. In addition, the CTL, a key partner in AIWG’s work and a resource for all faculty, encourages departments to reach out for individual consultations and department-level workshops.

Bringing students into the conversation about academic integrity and why policies matter can help strengthen understanding and buy-in. The AIMES website also includes a guide specifically for students, AI and Your Learning: A Guide for Students, with practical advice on using AI responsibly.

“We recognize that restrictions on AI use are essential to achieving certain learning goals, and at the same time, faculty are exploring constructive ways of weaving AI into their teaching and student learning,” said Cassandra Volpe Horri, associate vice provost for education and director of the Center for Teaching and Learning.