Assessment
Graded on judgment, not raw output.
Every student is required to use AI in this course — and required to disclose it and verify it. The marks follow that discipline: verification to source, sound prompting, awareness of bias and sycophancy, ethical handling, and a reflective account of process. The model's confidence is never the evidence.
How the course closes
The course closes with a dedicated 60-minute evaluation session, taken live. Graded work rewards judgment, verification and process over raw model output.
The scheme
How the grade is built
Five components. Participation and the hands-on work run throughout; the portfolio and the audit are assembled from what you produce in class; the evaluation session is taken live; and the capstone carries the most, because it asks you to do the whole job at once.
The capstone · after session 8
20%
Capstone — a verified AI-assisted work product
A work product in the student's own discipline — a research memo or contract for law students; a research brief or policy note for others — produced with AI and shipped only after every citation and proposition is verified, with the verification trail and a short reflective account of process and ethics.
| Component | Weight |
|---|---|
Participation & the hands-on segments Throughout Engagement across the live experiments, the commit-then-reveal polling, and the deliverable produced in each session's 30-minute hands-on half. | 10% |
Legal Prompt Portfolio Sessions 3–4 onward Documented prompts, outputs and critique across a set of tasks — an issue list, a clause, a judgment summary, a counter-argument — demonstrating the four computational-thinking pillars and iteration discipline. | 25% |
Hallucination Audit Session 8 The signature assignment: verify every citation and proposition in a seeded AI-drafted memo to source, catching all three classes of authority failure, and submit the audit report with its verification trail. | 25% |
Evaluation session Closing session The in-class 60-minute paper: an applied audit against supplied sources, plus short answers on the course's frameworks. | 20% |
Rubrics across every component reward correctness of verification, soundness of prompt strategy, awareness of bias and sycophancy, ethical handling (confidentiality, candour, disclosure), and reflective insight. Each graded component carries an AI-use disclosure declaration — you are required to use AI, and required to disclose and verify it.
What the rubric rewards
The same five things, every time
Across the portfolio, the audit, and the capstone, the rubric criteria converge on one profile of a competent AI-using lawyer. These themes recur in every component — so students always know what good looks like.
Verification to source
Every citation and proposition checked against a grounded tool or citator. The duty to check — the line between using AI well and a cost order — is what the marks reward first.
Sound prompt strategy
Decomposition, few-shot patterning, abstraction, and chain-of-thought, shown through iteration — explore prompts, then exploit the winner — rather than a single lucky shot.
Bias & sycophancy awareness
Steel-manning the opposing side, resisting a model that merely confirms you, and naming the limits of an output instead of presenting unverified authority as fact.
Ethical handling
Confidentiality, candour to the tribunal, and clear disclosure of AI use — kept in the work and in the write-up, not bolted on at the end.
Reflective insight
An honest account of process: which tools were used, what was checked, what was rejected, and how the student sits as the human in the loop who owns the result.
Integrity & AI use
Use AI. Disclose it. Verify it.
The course models the norms it teaches. Far from banning AI, it requires it — and then holds students to the professional standard that turns a tool into competent practice.
Required, not forbidden
AI use is central to every graded assignment. Students must state which models and tools they used and submit an AI-use disclosure with each submission.
Verify before you rely
Any legal claim in an output must be checked to source before it is relied on. No unverified or fabricated authority may appear in submitted work; outputs that are merely plausible are flagged, not presented as fact.
Protect the client
No privileged or confidential material may be entered into public LLMs — a confidentiality line carried through every lab and the capstone.
How it is marked
Grading explicitly rewards judgment, verification, and process over raw model output. Shipping an unverified or invented authority is not a minor slip — it is treated as the very failure the course certifies against, the lesson of Mata v. Avianca made consequential. The disclosure and the verification trail are part of the deliverable, not an afterthought.
Every component is practised before it is graded.
The portfolio entries, the audit and the capstone are all built from deliverables you produce in the thirty-minute hands-on half of a session — so nothing is assessed that the course has not already had you do.