- About
- Consultation
- Programs
- Resources
- News & Events
Back to Top Nav
Back to Top Nav
Back to Top Nav
On two days in July, faculty from across Dartmouth's schools gathered virtually to engage with one of the most pressing questions in higher education today: How do we teach effectively in an age of generative AI? The July 2025 Teaching with Gen AI Institute, hosted by DCAL and LDI, brought together 15 instructors and staff-educators for an intensive exploration of both the technical realities and pedagogical possibilities of AI in education, and specifically in courses and assignments.
"I came in as a skeptic. I'm still terrified that AI will replace a lot of learning…," shared one participant. "BUT it's here whether I like it or not, and I feel a lot better prepared to face it and take advantage of it already."
The Teaching with Gen AI Institute is designed around two complementary goals: building Gen AI literacy and translating that understanding into practical assignment design. This structure reflected DCAL and LDI's recognition that effective AI integration requires both technical understanding and intentional pedagogical reflection.
The first day of the institute focuses on demystifying generative AI through hands-on exploration and values-centered discussion. Facilitators guide participants through common misconceptions about how GAI works. One key concept is that "LLMs," or large language models on which GenAI tools depend, don't "know things" in the traditional sense, but rather generate text based on statistical likelihood in a given context. Another core idea is that ChatGPT isn't an LLM itself, but rather a software layer with prompts, guardrails, and memory built on top of underlying models.
One participant described the first day as a "...very good explanation of software/model relationship, noting that it was the, "First time I have really gotten that info and I find it quite valuable.". For facilitators, this feedback highlighted the importance of building foundational understanding before diving into teaching applications.
The second day shifts focus to practical implementation within the classroom, beginning with an introduction to Retrieval-Augmented Generation (RAG) led by Research Computing and Library Research Data Services partners.
Participants then engage in a crucial exercise: running their own assignments through multiple AI models to see what the tools did well and poorly, then reflecting on alignment within the context of their course learning objectives.
The day's centerpiece is dedicated time for assignment revision and creation, where faculty, instructors, post-docs, and staff educators brainstorm ways that students might effectively utilize Gen AI for editing, generating samples, brainstorming, and engaging with course content. Using a metacognitive activity - called Journey Mapping with Janus led by Rafe Steinhauer, Assistant Professor at Thayer School of Engineering - participants designed student experiences that intentionally leveraged GAI while maintaining focus on learning outcomes.
"The afternoon 'time to play' gave me the tools and confidence to explore on my own," reflected one participant, emphasizing the value of structured, hands-on exploration.
Throughout both days, the institute emphasizes that effective GAI integration must align with instructor values around deep learning, creativity, authorship, and critical thinking. Case study discussions highlight how different faculty approached similar challenges based on their pedagogical priorities—some focusing on preserving "productive struggle" while others prioritized building student confidence or fostering critical evaluation of GAI outputs.
"I really appreciated the 'forced' time to play with the tools ourselves," shared one participant. "The organizers hit the nail on the head when they said people learn a lot by playing with it but don't take (or give themselves permission to take) the time to do so."
The institute reinforces three critical insights that are shaping Dartmouth's approach to GAI in education.
First, access to multiple AI models often proves more valuable than relying on a single vendor. Through Dartmouth's chat.dartmouth.edu interface, faculty, staff, and students can experiment with different models' strengths; some excel at creative tasks, others at analytical work, and still others in specific domains. This diversity allows for more nuanced and effective GAI integration into the course, especially when the needs require more disciplinary specificity. An added bonus to using Dartmouth Chat lies in its privacy protections, including options for complete chats within Dartmouth's computing environment (read about data privacy).
Second, AI detection software is not a useful solution for addressing academic integrity concerns. Detection tools are unreliable and can create more problems than they solve, often flagging legitimate work while missing actual GAI use. This also erodes trust between teachers and students, which can be problematic in the context of Dartmouth's honor policy.
The third insight offers a more promising path: thoughtful assignment design. Participants are learning to integrate GAI purposefully by scaffolding activities, creating multiple versions of assignments, incorporating in-class work, and designing tasks that make productive use of GAI while maintaining academic rigor. As one participant noted about the hands-on approach: "You can hear so much about these tools but getting hands-on with them (in a structured way) really helps you understand things."
The institute's impact extends beyond individual faculty development to broader institutional capacity-building. Participants left with concrete plans for GAI integration in their courses, enhanced technical literacy, and—perhaps most importantly—a framework for making values-based decisions about when and how to use these tools.
The collaborative workshop format proves especially valuable, with faculty appreciating both the structured learning and peer dialogue. "The structure of the session was perfect. You all achieved an excellent balance between interactive and solo work, as well as having us use the models," noted one participant.
The July institute represents the continued evolution of DCAL and LDI's Teaching with Gen AI Initiative, which includes grant programs, ongoing institutes, and resource development. Faculty interested in deeper engagement can apply for a Teaching with Gen AI course grant, which provides project support, community connection, and stipend funding for courses integrating GAI. The upcoming call for proposals for Winter and Spring courses will be announced in October.
Recently, DCAL and LDI hosted two conversations on the use of Gen AI in the academic setting as our community prepared for the start of fall term. Gen AI & Academic Honor: Your Fall Term Playbook shares key takeaways and strategies for teaching this term.
As Dartmouth continues to navigate the GAI landscape in education, the institute model offers a promising approach: combining technical literacy with pedagogical reflection, emphasizing learning-based decision-making, and providing structured opportunities for hands-on experimentation. As a result, participants are not just better prepared to use GAI tools, but equipped to use them thoughtfully in the service of student learning.
If you are looking to deepen your knowledge of Gen AI in the context of teaching and learning, learn more about and apply for the next Teaching with Gen AI Institute, held December 9 and 11 in the LINK (180 Berry).