We’ve launched a semi-regular feature to keep you informed about how artificial intelligence (AI)—especially generative AI (GenAI)—is shaping teaching and learning at Carleton and beyond. This month we’re turning our attention to a new advancement in the technology:
Agentic AI – From Assistance to Agency
Agentic AI refers to systems that go beyond simple prompt-and-response to act on behalf of the user—reasoning across sessions, managing tasks and adapting to context. These tools are designed to anticipate needs and make decisions rather than merely offer suggestions.
In teaching and learning, this shift invites us to rethink how agency is distributed in the learning process. When AI can search, summarize and synthesize with minimal human direction, what does it mean for students to remain active participants in their own learning?
Rather than viewing these tools as replacements for cognitive effort, we can use them to spark metacognitive reflection. Try asking:
- How can students make their thinking visible when AI assists in their process?
- What aspects of learning should remain human-led, even when AI can “do” the task?
- How might we guide students to use AI tools for exploration rather than expediency?
Tool Spotlight: Perplexity’s Comet Browser
Perplexity’s Comet browser represents a new generation of agentic AI. It integrates a generative AI assistant directly into the browser, allowing it to interpret web pages, reason about user intent and take actions such as navigating sites, summarizing content, comparing information across tabs, filling out forms and even completing transactions.
While these capabilities may feel futuristic, they’re already reshaping how students and educators interact with digital environments. Agentic browsers like Comet can autonomously navigate learning platforms, identify tasks and in some cases even complete online assessments. This activity is currently undetectable within Brightspace or other learning management systems and cannot be technically blocked.
That doesn’t mean panic is the answer. It means design matters more than ever.
The emergence of agentic AI highlights the need to move beyond reactive measures (like detection or lockdown tools) and instead focus on fostering authentic engagement and integrity through assessment design. Long-term strategies include:
- Transparency and reflection: Encouraging students to disclose how AI supports their work and reflect on what learning remains uniquely human.
- Process-driven assessment: Incorporating drafts, planning notes or portfolios that make thinking visible across time.
- Authentic contexts: Designing assignments that emphasize reasoning, collaboration and application over simple information retrieval.
At the same time, Comet and similar tools present real opportunities. For example, they could enhance accessibility for learners with disabilities or language barriers, reduce digital friction for neurodivergent students, and model how AI can be used ethically to streamline research and exploration.
In short: agentic AI will redefine how students learn—but educators still define why.
Have something to share or a question about AI in your course?
We want to hear from you! Reach out to our team with your ideas, challenges, or success stories: TLS Support Portal.