Faculty Examples: Build AI Literacy

Help students understand how AI tools work, including their limitations and their social and economic impacts. That understanding lets them use AI without giving up their own critical thinking.

As the AI and Learning page shows, students who understand how AI works experience meaningfully smaller declines in critical thinking than students who don't, regardless of how often they use AI tools. Literacy isn't the same as competency: knowing how to operate a tool is different from understanding it holistically. See the Course & Assignment Design page for more on how this move fits into the fuller design picture. 

Below are examples from UO faculty.

In the Neural Networth assignment, Professor Cordes gives students beads and wire and asks them to make representations of neural nets in groups. The purpose is to find new ways to visualize neural nets using art-based research methods. Students then show their visualizations in groups of five and freewrite on two questions: whether the hands-on process of creating a beaded neural network changes their understanding of how information, memory, and connection operate within AI systems, and what the tactile, labor-intensive act of beading might reveal that stays hidden when neural networks are treated only as abstract digital technologies.

Design Moves

• Students can use GenAI for inspiration, but in their explication of the project they must describe how the visualizations produced by GenAI compared and contrasted with their own designs.
• When students use generative AI in a formal assignment, she requires a Resource Use Acknowledgement: a statement acknowledging that the use and training of AI tools contributes to global carbon emissions and ecological degradation, or naming the types of energy drawn on. Her sample statement starts from the University of Oregon's hydroelectric power, notes that carbon-free does not mean problem-free, and points to the changed salmon habitats of the Columbia and Snake Rivers.

In her 96-student ANTH 255 course, Professor Carter is shifting toward mixed AI use anchored in in-person, group, and experiential work. To build AI literacy specifically, she runs a group activity where students test the same site or topic against a progression of increasingly specific Copilot prompts and compare how the responses change. Prompts used in this activity:

1. What can you tell me about [your site/topic]?
2. What can you tell me about what Graham Hancock says about [your topic/site]?
3. What do archaeologists say about [your site/topic]?
4. Is the information presented in the Netflix Ancient Apocalypse episode true? What do archaeologists say about the claims in this episode?

Design Moves

• Groups work through four prompts of increasing specificity: a broad topic query, a query naming a specific non-expert, a query for disciplinary consensus, and a direct fact-check against a specific media claim.
• A shared table has students record Copilot's initial response and evaluate the quality of its follow-up questions, judging the output, not just collecting it.
• Surfaces where AI defers to popular but non-expert sources versus actual scholarly consensus, giving literacy work a concrete misinformation case rather than an abstract one.
• Small-group format makes testing a shared, discussable experience instead of a solo exercise.

In the same policy document referenced on the Faculty Examples: Communicate with Your Students Transparently page, Professor Gopalan explains why uncritical AI use specifically damages learning to code, naming the exact failure mode rather than gesturing at AI use as generally bad.

She says, "It often gives you code that is wrong or that contains many inefficiencies... This seriously precludes you from learning the efficient ways to code, and because you don't know enough of the language you're working with, you won't understand why or what's going on."

Design Moves

• Names a specific, discipline-grounded mechanism for how copy-paste AI use undermines learning, tied to what goes wrong when debugging code you don't understand.
• Names the downstream risk explicitly: continued dependency on AI tools "when they might not be available to you."
• Posts ongoing resources on Canvas about the AI tools themselves, treating literacy as something built over the term, not covered once.
• Explicitly invites students to ask questions and meet one-on-one about how AI helps or hinders their specific learning.