Every exercise page includes purpose, learning goals, time, materials, setup, student instructions, instructor notes, and adaptation options for different formats and audiences. Filter by pillar to find an exercise that matches the skill you want students to practice, or start with the capstone if you want students to compare a full traditional and AI-assisted workflow.
The capstone asks students to compare a traditional legal research workflow with an AI-assisted workflow, focusing on what changed, what improved, what became riskier, and what would need to be documented. The goal is judgment, not a verdict that AI is better.
Students receive a realistic client email containing sensitive employment-related facts and prepare it for possible use in a public-facing AI tool. The goal is not a perfect redaction, it is to experience how hard it is to remove identifying and sensitive information while preserving enough legal substance for the tool to be useful.
Students create an AI-generated “secondary source” on a legal topic, then compare it to a real secondary source covering the same concept. The exercise asks students to treat AI as a possible framing tool, not as authority, and to decide what the real secondary source clarifies, structures, or verifies that the AI output does not.
Students receive an AI-generated research memo seeded with realistic errors, fabricated citations, misstated holdings, and confidently-wrong synthesis, and must red-line it to publication standard.
Students diagnose a weak legal prompt before improving it. They decide what kind of task the prompt is really asking the AI to perform, what tool is appropriate, what context is missing, what constraints are needed, and what verification step should follow.
Compare AI-generated statutory interpretation against an annotated code, focusing on current-law verification, enacted-versus-proposed changes, effective dates, jurisdictional variation, and cases interpreting the statute.
Students draft or receive a short legal memo, then prompt an AI tool to critique it. They evaluate the critique itself, separating useful structural feedback from overconfident, generic, inaccurate, or strategically unhelpful suggestions.
Students map a legal research workflow on paper or a shared digital board. They identify which steps benefit from AI, which steps should remain traditional, which steps require lawyer judgment, and where verification gates must be built into the process. Think of this as a mini version of the capstone.
Students use AI to enter a complex regulatory scheme, then follow the trail themselves. At each cross-reference, definition, exception, or agency-specific term, they decide whether AI can be used for orientation, whether the claim must be verified in the CFR or agency source, or whether the work should pivot entirely to the official regulatory text.
Students match legal research tasks to the tools or sources best suited for the work. They decide whether a task belongs with AI, a legal research database, an official source, an annotated source, a human lawyer, or some combination, and explain where the boundary is.
Students act as a firm AI review committee. They evaluate a proposed AI tool, identify confidentiality, verification, supervision, and workflow risks, and recommend whether the organization should adopt, pilot, or pass on the tool.
Participants take a weak legal research prompt and revise it into a stronger prompt by adding jurisdiction, timeframe, source type, task, constraints, and a follow-up verification plan.
Participants review an AI-generated legal draft through a supervision lens, identifying what must be checked, revised, verified, reframed, or rejected before the work could be shared with a supervisor, client, or court.