Brown University professors across three disciplines are redesigning their courses to move beyond simple debates about banning AI. Their syllabi now treat generative AI as a tool students must learn to wield critically - and as a subject that demands scrutiny of its own risks to privacy, labor, and human rights.
The approaches come in response to a summer report from the University's Generative Artificial Intelligence in Teaching and Learning Committee. That report found wide variation in AI adoption across subject areas and recommended that courses emphasize the process of learning, not just final exams or papers.
Rethinking pedagogy, not just policing cheating
In EDUC 1485: "AI and Education: Critical and Applied Perspectives," lecturer TJ Kalaitzidis asks students to question the structure of education itself. Traditional teaching methods, he said, emphasize exam outcomes and final products, which "incentivizes students to outsource their faith" to AI tools.
"If we are to create educational spaces that serve the needs of students moving forward, we need to rethink what our pedagogy looks like," said Kalaitzidis, who also serves as assistant director of learning systems innovation at the Sheridan Center for Teaching and Learning. His course allows AI use but requires students to demonstrate understanding without it through presentations and one-on-one discussions.
One assignment has the class read an article, reach a shared interpretation, and then ask a large language model to analyze the same text. The AI's output, Kalaitzidis said, is a "generic and uninteresting" answer compared to what students produce together. The exercise makes the limits of automated reasoning visible rather than merely warning about them - a practical example of AI for Education strategies that focus on critical engagement over prohibition.
Governing AI means defining the problem first
In CSCI 2953B: "So You Think You Want to Govern AI?", Professor Suresh Venkatasubramanian teaches students to break broad anxieties into specific, addressable problems. The course covers deepfakes, copyright, privacy, national security, and automated decision-making.
"People say, 'Well, I'm very worried about this thing with AI.' I'll say, 'Okay, what exactly is the problem you're worried about?'" Venkatasubramanian said. When a student cites fear of job loss, he pushes them to identify the precise mechanism - which jobs, which sectors, over what timeline. "The more clear you can be about what exactly the concern is, the more clear you can be about what you're trying to intervene on."
Venkatasubramanian initially banned AI in his classroom but revised his policy to permit it for background research and understanding the literature. He used the University's AI Policy Explorer to generate a template. For him, governance does not inherently block innovation - it channels it. The question is not whether to govern rapidly accelerating technologies, but how and who gets to do it.
Human rights and the "slow violence" of AI
IAPA 1801I: "Human Rights and AI: Impacts, Risks and Opportunities," taught by human rights lawyer and visiting professor Malika Saada Saar, examines AI's potential to both undermine and advance fundamental rights. Saada Saar, who previously worked on technology policy at Google, structures the course around what she calls AI's "peril and promise" in the context of justice.
She draws attention to what advocates term the "slow violence" of AI - harms that accumulate gradually. "What is freedom of expression when we can't discern whether it's a human being's expression or a bot?" she said. "What does free speech mean when so much of our information ecosystem is supercharged by misinformation and disinformation?" The course also addresses discrimination, surveillance, and the disproportionate online shaming of women and girls.
Saada Saar does not want students to conclude AI is purely destructive. She points to applications that could expand access to health care, address educational inequalities, and improve accessibility. Her classroom policy reflects this dual stance: students can use AI as a "critical thought partner," but it cannot do their writing. The goal is comfort with the technology, not dependence on a single tool.
Why this matters for educators
These three courses share a practical throughline for teachers and instructional designers. Each professor treats AI literacy as a discipline-specific skill, not a generic workshop topic. Kalaitzidis redesigns assignments to make the learning process visible and unskippable. Venkatasubramanian trains students to define problems before proposing governance fixes. Saada Saar connects technical systems to long-term rights impacts. For educators building their own AI policies, the message is clear: effective approaches tie permitted uses to explicit pedagogical goals and require students to show their thinking - not just submit a polished artifact. Programs like AI for Teachers offer structured pathways for developing those classroom strategies without reducing the conversation to a binary of embrace or ban.
Your membership also unlocks:
