Five faculty projects that involve innovative approaches to improving student learning will be honored next month with the Provost’s Teaching Innovation Prizes.

The winning projects were chosen from 36 nominations from students, faculty and staff. Innovations were encouraged from five focus areas:

• Enhancing student success through the use of alternative forms of assessment.
• Innovations to create student-centered learning environments that empower students to succeed by narrowing gaps in outcomes.
• Innovative approaches to handling the challenges and opportunities posed by Generative AI tools, including helping students develop skills for a future that includes GenAI.
• Exploring the creation of equitable, ethical and effective tech-free learning environments that promote deep thought and focus.
• Creative and meaningful linkages among coursework, curriculum and career preparation, with an emphasis on transferable skills.

The honorees will talk about their projects during a virtual discussion at 2 p.m. May 6 as part of U-M’s annual Enriching Scholarship conference. Angela Dillard, vice provost for undergraduate education, will moderate the session. 

The Provost’s Teaching Innovation Prize is sponsored by the Office of the Provost, the Center for Research on Learning and Teaching and the University Library. The winners receive $5,000.

These summaries of the 2026 TIP honorees were submitted by CRLT.

Community-Engaged Understandings of Homelessness: Connecting Past Policy to Present Conditions 

Ayesha Ghazi Edwin, clinical assistant professor of social work, School of Social Work

Ghazi Edwin redesigned a required social work course around a central reframing: homelessness is not an individual condition, but the result of historical and ongoing housing policy decisions.

Rather than beginning with homelessness, the course begins with housing policies rooted in anti-Blackness, including racially restrictive covenants, exclusionary zoning, and the creation and retraction of housing assistance and social welfare programs that shaped housing access and continue to drive present-day disparities.

In this course, Ann Arbor becomes the classroom. Students use the city as a case study, analyzing how single family zoning has contributed to rising housing costs, a primary driver of homelessness, and come to understand how these policies have constrained housing supply and driven up costs for everyone, including themselves. They engage with Ann Arbor’s Comprehensive Land Use Plan to evaluate how zoning and land use decisions shape affordability and housing instability, while identifying narratives of inclusion and exclusion embedded within policy debates.

Students confront assumptions and explore how housing instability shapes outcomes across the life course while being immersed in Washtenaw County’s homelessness response system. Through visits to historic Black neighborhoods, conversations at the Elks Lodge, and engagement with housing providers, health organizations, and community advocates, students encounter housing injustice as a living reality in their local community. 

They leave understanding stable housing as a foundational intervention that shapes health, recidivism, education, employment, wealth and family stability, and one that must be integrated into every type of social work practice — from clinical therapy to community organizing to policy advocacy. 

How Do You Make a Large Class Feel Small?

Jeffrey Koller, lecturer III in mechanical engineering; and Alex Shorter, associate professor of mechanical engineering, both in the College of Engineering

The redesign of the 100-student “Introduction to Dynamics and Vibrations” (ME 240) recreates the benefits of a small classroom, where high student-instructor and peer-to-peer contact fosters community and continuous assessment, within a large-enrollment environment. Motivation for this work stemmed from observing that in the traditional course format (heavily lecture based with two exams and a final), struggling students were not identified until the first exam, nearly a third of the way through the semester. 

Given the course’s cumulative nature, this late identification made recovery difficult and exacerbated equity gaps observed in foundational STEM courses. The course redesign addresses these issues through two key interventions: a mechanism for early, conversational feedback and building a robust peer community in the large classroom environment. 

Koller and Shorter shifted course content to five learning modules with frequent, low-stakes quizzes and moved technical derivations to pre-recorded videos. This freed up class time for student-centered active learning in consistent teams of five to six students. 

These structural changes enabled their core innovation: the “Checkout.” Before each quiz, teams verbally present their solution process for a representative problem to the instructional team, including undergraduate IAs. Unlike passive problem-solving, the “Checkout” creates a low-stakes environment where feedback happens through dialogue rather than a red pen on an assessment. 

This process teaches students to identify effective problem-solving strategies, develops technical communication, and fosters community. By reimagining the large lecture as a collection of small, supportive communities, Koller and Shorter provide students with the technical mastery and peer networks necessary to thrive throughout their engineering careers.

Opt-In Active Learning in a Large Gateway Course: A Two-Pathway Model for STATS 250

John Keane and Alicia Romero, both lecturer III in statistics; and Mark Rulkowski, lecturer II in statistics, all in LSA 

The adage “What I hear, I forget. What I see, I remember. What I do, I understand,” underscores the importance of active-learning strategies in education. In STEM, these methods can foster greater student engagement, retention, and comprehension but are often underutilized in large-enrollment courses due to scalability challenges. 

To address this issue in STATS 250, a course that serves about 3,500 students from more than 166 different majors across campus, Keane, Romero and Rulkowski developed a new lecture curriculum that emphasizes interactivity, hands-on practice, and socialized learning. Their new model allows students to select either an active or traditional learning approach. 

In the active learning approach, which approximately 80% of students selected this semester, students complete an in-class group activity every class meeting. Submissions are scored on accuracy and exams are not as heavily weighted in the overall grade. In the traditional approach, students are not required to submit group work activities and exams are more heavily weighted. 

Revamped 80-minute class sessions dedicate the last 30 minutes to small-group activities where students engage in data analyses using R. These activities provide low-pressure opportunities for students to assess their understanding of content introduced each day before tackling weekly assignments. GSIs and IAs join class during these sessions to offer guidance and probe students’ thinking. 

After implementing this change, the team saw daily in-person attendance jump from roughly 25% to 75% and up to 40% reductions in achievement gaps between students historically marginalized in STEM fields and their counterparts. 

Problem First, AI Second: Teaching Agency in an Age of Agents

Branko Kerkez, Arthur F. Thurnau Professor; and professor of civil and environmental engineering, College of Engineering

Most conversations about AI in the classroom start with the role of technology. Kerkez flips the paradigm by ignoring AI and starting with the problem. Students spend weeks developing domain expertise and identifying questions worth answering. By the time AI enters the picture, every student has a problem they want to solve and the domain knowledge to judge whether a generated solution is any good.

The AI students encounter is not a chatbot. They build agents: student-guided systems that follow instructions, watch for new data, reason about it, and take actions. Students leverage an open-source, no-code platform using visual diagrams and plain English. 

Each student provides the domain knowledge, and the agent provides the automation. Because the innovation is in the sequencing, not the subject matter, an instructor in any field with open-ended problems can adopt the same structure: build domain expertise, then hand students control of the tools.

In ENGR100 (Smart Cities), first-year teams research how cities work, analyze tradeoffs and risks, and write a proposal to city managers and residents on the role of tech in solving community challenges. Only then do they design and deploy AI agents that address problems real communities face (from pothole detection to trash pickup). Successful adaptation in the graduate course, CEE575 (Sensors and Data), demonstrates transferability. 

Across both courses, AI-assisted reflections revealed students’ perceptions and hesitations around AI. When surveyed, students rated goal setting, knowing what they want the technology to accomplish, as the most important skill in an era of learning with AI.

Students Learning From Students: A Change That Makes Exams Engaging and Possibly Even Fun

Regina Baucom, professor of ecology and evolutionary biology, and Hilary Archbold, lecturer IV in molecular, cellular and developmental biology, both in LSA

From years of teaching the large, approximately 500-student course Introduction to Genetics (Biology 305), Baucom noticed that some of the most constructive and engaging learning happened during office hours, where students learned not only from her explanations but also from one another as they worked through challenging problems together. 

However, only a small fraction of students regularly attended office hours. She sought to recreate this collaborative learning dynamic at a much larger scale. Along with Archbold, longtime course coordinator for BIO 305, the two developed a two-stage exam process that transforms exams from primarily individual evaluative tools into a second opportunity for students to engage with course material through structured collaboration.

Students first individually complete a traditional, multiple-choice exam during lecture. Soon thereafter, they take a second “discussion exam” in their discussion sections, where they work in groups to solve more complex problems requiring deeper reasoning and synthesis. Students submit individual answers following group discussion, encouraging peer-to-peer learning while preserving individual reasoning and choice.

The discussion exams extend to all students in this gateway course the kinds of collaborative problem solving typically limited to especially interactive discussion sections or office hours. Student feedback indicates that the format improves conceptual understanding, increases confidence, and reduces the sense of working through difficult material alone.

Because the collaborative component takes place within existing discussion sections and uses the current instructional team, the model requires no additional contact hours or financial resources, thereby increasing the innovation’s potential to scale and transfer across disciplines.