Secondary Source Comparison

Purpose

This exercise helps students compare AI-generated legal explanation against an authoritative secondary source.

By this point in the course, students have already seen that AI can be useful for orientation, issue-spotting, and framing. They have also seen that AI outputs may converge on a general doctrinal answer while diverging significantly in the authorities they cite or emphasize. This exercise moves that observation into the secondary-source context.

Students ask: if AI can sound like a fast, synthetic secondary source, what is missing? What does a real secondary source do differently? Where does AI help with orientation, and where does it create verification work?

The goal is not to prove that AI is useless or that secondary sources are always perfect. The goal is to help students identify the respective functions of each: AI as a possible starting point for framing and vocabulary, and real secondary sources as structured, sourced, jurisdictionally disciplined research tools that point the researcher toward primary law.

Learning Goals

By the end of this exercise, students should be able to:

1. Explain the functional role of secondary sources in legal research.
2. Evaluate an AI-generated “secondary-style” explanation for coverage, nuance, terminology, and research-roadmap value.
3. Compare AI output to a real secondary source addressing the same legal concept.
4. Identify what the real secondary source clarifies that the AI output leaves vague, overgeneralized, unsupported, or untethered from authority.
5. Recognize when AI is useful for initial framing and when it creates more verification work than it saves.
6. Identify at least one primary-law verification step that should follow from the comparison.

Materials

• Access to a general-purpose AI tool, legal-specific AI tool, or other approved AI research platform.
• Access to legal research databases containing secondary sources, treatises, practice guides, legal encyclopedias, annotations, or other credible explanatory materials.
• A comparison prompt or worksheet asking students to evaluate coverage, precision and nuance, terminology, research-roadmap quality, confidence versus uncertainty, and what the real source clarifies that the AI output does not.
• A short post-class submission prompt asking for one strength of the AI output, one way the real secondary source was superior, and one primary-law verification step to take next.

Setup

This exercise works best after students have already worked with AI on primary-law research.

In the class version, the exercise followed a prior group activity in which students used different AI tools to research a primary-law problem. The class observed that groups could reach similar broad doctrinal conclusions while citing different cases or relying on different authority sets. That finding created the transition into secondary sources: if AI can help frame doctrine, students still need to know whether that framing is sourced, complete, current, and jurisdictionally disciplined.

Before the exercise, students should understand the basic role of secondary sources. Secondary sources help researchers understand unfamiliar doctrine, learn vocabulary, identify rules and exceptions, and build a research roadmap. They are most valuable early in the research process, but they do not replace primary law.

Students should also understand the parallel risk with AI. AI often performs a similar framing function, it can explain concepts, generate checklists, identify likely doctrines, and suggest next steps, but it often lacks source transparency, authority awareness, and jurisdictional discipline.

Run of Show

Total, approximately 30–45 minutes

1. Frame the comparison~5 min Explain that students will create their own AI-generated “secondary source” and compare it against a real one. The comparison focuses on function: what each source helps the researcher do, what each leaves out, and what must be verified next.
2. Choose a topic~3 min Students choose a legal concept they might encounter in practice, noncompete enforceability, data-breach notification duties, products-liability warning defects, contract termination, or housing habitability standards, and pick a jurisdiction.
3. Constrain the problem~3 min Before prompting, students identify the jurisdiction, client posture, procedural posture, and, if useful, two or three legally relevant facts. The problem should be broad enough to generate a checklist or overview, not so fact-specific that the tool produces a one-off answer.
4. Generate the AI “secondary source”~7 min Students ask an AI tool to explain the concept as if a partner were advising a new associate, asking for key elements, important exceptions, issue-spotting questions, special considerations, and next research steps.
5. Locate a real secondary source~7 min Students find one credible secondary source on the same concept, a treatise excerpt, practice-guide section, legal encyclopedia entry, or similar source with enough substance to evaluate structure and nuance.
6. Compare~10 min Students compare the AI output and the real secondary source on coverage, precision, terminology, organization, jurisdictional handling, confidence versus uncertainty, and research-roadmap quality.
7. Save for submission~5 min Students save or excerpt the AI output, record the real secondary source citation, and note what the AI did well, what the real source did better, and what they would verify next using primary law.
8. Post-class submissionafter class Students submit one short paragraph: one strength of the AI output, one way the real secondary source was superior, and one primary-law verification step they would take next.

Student Instructions

Instructor Notes

The central teaching point is that AI can resemble a secondary source without actually being one.

Students may find that the AI output is readable, quick, and useful for orientation, it may translate an unfamiliar topic into accessible vocabulary, generate an issue list, or suggest a sensible research sequence. Those are real strengths.

The comparison becomes useful when students see what the real secondary source adds: source transparency, doctrinal structure, jurisdictional specificity, citations, exceptions, caveats, and a clearer path toward primary law.

Common AI failure modes include:

• Overgeneralizing across jurisdictions.
• Flattening flexible standards into rigid rules.
• Blending majority and minority approaches.
• Omitting exceptions.
• Sounding confident where the law is uncertain.
• Producing citation-shaped or authority-sounding language without reliable provenance.
• Giving a useful overview but not enough structure to guide serious research.

The strongest debrief questions are:

• Where was AI genuinely useful as an on-ramp?
• Where did the real secondary source make the doctrine more precise?
• What did the real source help you see that the AI did not?
• Did the AI output make you faster, or did it create more verification work?
• What primary law would you check before relying on either source?

Reinforce that neither AI nor secondary sources are endpoints. Both can help frame the research, but primary law remains necessary for verification, authority, and final legal analysis.

Adaptation

Here are some suggestions on how to adapt this in other teaching contexts. The adaptations below have been limited to firm trainings and asynchronous suggestions, but there are other possible adaptations for workshops, seminars, and other contexts. Be creative!