If you want students (and faculty) to trust an AI assistant, every claim has to be checkable. That’s why mentorAI’s Document Retrieval feature makes answers citable by default: when a learner asks something—“Can you explain key epidemiological study designs?” was a recent example we discussed with Boston College—the reply names the exact source (e.g., “Lecture 11 — Slides 35–36”) and shows a live Source Panel with the documents used, ranked by relevance. One click opens the original file so learners can read the surrounding context immediately.
This isn’t a bolt-on; it’s how the assistant is meant to work. Instructors can load course PDFs, slide decks, or readings and keep tight control over what’s shown to students. Crucially, visibility is a toggle, not a retrain: admins flip an eye icon per file to decide whether it appears in the Source Panel while still letting the model use it to answer questions. That means you can keep some materials “behind the scenes” for assessments or proprietary content—and change your mind instantly without re-indexing.
How It Works For Learners
- Ask a question. The assistant retrieves the most relevant items from the mentor’s dataset and composes an answer with inline citations (e.g., lecture/slide/page).
- Scan the Source Panel. See which documents were used, ranked by relevance (often with a confidence indicator). The panel updates as the conversation evolves.
- Open any source. Click through to the original slide, PDF, or reading to verify claims and keep studying.
Why Faculty Like It
- Transparent, citable answers. Replies point back to the exact lecture, slide, or page—great for research habits and academic integrity.
- Guided reading & deeper study. Students jump straight from a summary to the exact place in the materials.
- Instructor QA & gap finding. It’s obvious when the assistant cites the wrong thing—or when a course needs an extra reading.
- Assessment support. Keep certain files hidden while still letting the assistant draw on them; reveal later as needed.
Why Admins Appreciate It
- No retraining to change what’s shown. Per-file Visible toggles control what appears in the Source Panel; the assistant can still use hidden files to answer. Changes apply instantly.
- Works at scale. Even with large training sets, sources are still ranked and cited for each response.
Fits Real Course Workflows
Mentors are typically scoped at the
course level to avoid cross-level leakage (e.g., Pre-Calc answers pulling Calc III content). Instructors can
drag-and-drop their own materials to build each mentor’s knowledge base; retrieval/citation then anchors every response to those faculty-approved files.
Conclusion
Citations shouldn’t be a nice-to-have. With mentorAI,
answers are verifiable by design—named sources inside the reply, a
ranked Source Panel alongside it, and
one-click opening of the original document—so students learn to check evidence, and faculty stay in control of what’s shown. If you’d like to explore AI mentors that can cite your materials, visit
https://ibl.ai/contact