Right Tool, Right Job: How Rice's Library Gets the Upside of AI Without the Downsides
Hannah Edlund is a librarian at Rice University in Houston, Texas. She shares their experience adopting Consensus and leaning into the benefits that generative AI can have in scholarship.
When generative AI went mainstream, university libraries got hit first. Hannah Edlund, a librarian at Rice, watched early LLMs, or "statistical prediction machines," as she calls them, hand students convincing but "completely bogus citations." Rice's response wasn't to ban anything or panic. The library formed a working group to track AI's impact on research, a role eventually got created for Hannah to navigate it full-time, and the team landed on a motto that's stuck ever since: "right tool, right job."
That motto is the whole story. General-purpose AI is genuinely powerful. But in the classroom, that power points in the wrong direction. A general-purpose chatbot will happily write your entire literature review, taking the learning out of your hands, while also producing something the tool was simply not designed for. What students need isn't less capability, it's capability aimed at the right job: getting them to real, relevant, research faster, while leaving them the "stumbling blocks" where topic development actually happens.
So the question became: how do you get AI's speed without giving up academic integrity? For Hannah, Consensus is what it looks like when generative AI is put "through a process to become something genuinely useful in a specific context." Purpose-built, not general-purpose. Because "librarians are citation nerds," her team cared a lot that every answer comes with specific citations to where the information came from, presented cleanly.
The adoption numbers tell you it's landing at Rice. More than 700 Rice users have signed up and used Consensus this year alone. In on-campus UX sessions, business students used it to dig up niche resources that traditional databases buried. Undergrads put it more simply: "Oh, this is easier than your catalog." And the fast path from question to full paper matters more than it sounds — as Hannah puts it, "when people can't get to their papers, they'll just give up on things."
Faculty came around too. When one professor made the case for just using ChatGPT, Hannah's counter was simple: "With Consensus you don't need to build your prompt so specifically." One plain question, reliable summaries, real papers underneath.
None of this replaces critical thinking, and Hannah makes sure students know it. Her favorite example: a freshman asked Consensus, "Are vaccines safe?" One highlighted paper showed vaccines causing more adverse reactions than placebos. Read carelessly, that looks like the AI calling vaccines dangerous. Read carefully, it's a placebo comparison doing exactly what it should. "I love to point to it as a way for people to still need to do some of that extra analysis work," she says.
Her closing metaphor is the one we'd steal if we could: AI is a hammer. Great at driving nails. But the human still decides where the nail goes. Right tool, right job.
