Research Note

AI Tutor Source Pack Design Framework

Treat source selection as curriculum and evidence design. A source-grounded system inherits the pack's authority, versions, conflicts, gaps, rights, and data sensitivity.

Aug 4, 20262 min readBy Dalton Anderson
In this article

AI Tutor Source Pack Design Framework

Design principle

Treat source selection as curriculum and evidence design. A source-grounded system inherits the pack's authority, versions, conflicts, gaps, rights, and data sensitivity.

Source manifest

FieldEvidence to record
Learning objectiveWhat the learner must explain, decide, or do
Source identityExact title, author, publisher, URL or file, and version
AuthorityWhy this source can support the intended claim
DatePublication, effective, revision, and access dates
ScopeQuestions it answers and questions it does not
StatusCurrent, historical, superseded, draft, or disputed
RightsPermission, license, access, and sharing boundary
Data classPublic, internal, confidential, personal, regulated, or prohibited
ConflictsMaterial disagreement with another included source
RepresentationWhat an import omits or transforms

Method

Write the learning objective and a small set of test questions first.

Prefer primary authority for rules, specifications, policy, and original research. Add explanatory sources when they help a learner understand primary material, and label their role.

Preserve disagreements rather than forcing false consensus. Include a conflict note that explains the question, competing positions, dates, and evidence needed to resolve it.

Remove duplicates, unreadable scans, unneeded personal data, and material the user lacks permission to upload.

Test the pack with answerable, conflicted, and deliberately unanswerable questions. The system should not be rewarded for filling a documented gap with plausible prose.

Decision boundary

A coherent pack can support a bounded study job. It cannot authorize uploading copyrighted, confidential, personal, regulated, or institutionally restricted content.

High-impact questions still require authoritative systems and qualified human judgment outside the notebook.

Sources

Follow the evidence.

  1. support.google.com: 16322204support.google.com
  2. support.google.com: 17003757support.google.com
  3. support.google.com: 17004255support.google.com
  4. blog.google: notebooklm new features december 2024blog.google
  5. studentprivacy.ed.gov: privacy and education technologystudentprivacy.ed.gov
  6. support.google.com: 16212820support.google.com
  7. NIST AI Risk Management Frameworknist.gov
  8. www2.ed.gov: ai reportwww2.ed.gov
  9. daltonanderson.net: googles ai tutor the future of personalized learningdaltonanderson.net
  10. support.google.com: notebooklmsupport.google.com
  11. support.google.com: 16213268support.google.com
  12. journals.sagepub.com: fulljournals.sagepub.com
  13. youtu.be: 6BwWKkZ7aeAyoutu.be
  14. support.google.com: 16179559support.google.com
  15. NIST Generative AI Profilenvlpubs.nist.gov
  16. blog.google: notebooklm audio overviewsblog.google
  17. support.google.com: 16164461support.google.com
  18. open.spotify.com: 1YXy6yyC4u2mnqeARVB5qvopen.spotify.com
  19. edu.google.com: ai notebooklmedu.google.com
  20. support.google.com: 16215270support.google.com
  21. daltonanderson.ghost.io: googles ai tutor the future of personalized learningdaltonanderson.ghost.io

From this episode

Two useful next steps.

Evergreen · 1 min

What NotebookLM Does With Your Sources

Understand how NotebookLM imports sources, retrieves passages, generates answers and artifacts, shows citations, and changes data boundaries across accounts and sharing.

Guide · 1 min

How to Verify AI Answers and Citations

Audit an AI answer claim by claim by checking source identity, passage accuracy, support, context, inference, missing evidence, currency, authority, and final wording.

Return to the episode
AI Tutor Source Pack Design Framework