Assignment Package: Staged Research or Writing Project

Use this pattern for a substantial paper, research memorandum, policy project, drafting portfolio, or other work that should develop through feedback. Students submit a small number of meaningful checkpoints. Each stage has its own AI rule, so “AI permitted” never becomes an ambiguous course-wide instruction.

Purpose

Staging serves learning first: it creates timely feedback, catches unworkable research paths, and gives students a reason to revise. It also distributes the evidence of learning across more than one polished final artifact. Do not collect process documents that neither student nor instructor will use.

Appropriate courses

Seminars with research papers or supervised writing requirements
Advanced legal writing, drafting, legislation, policy, and capstone courses
Clinics or externships only under controlling supervision and confidentiality rules
Doctrinal courses using a shorter three-stage analytical project

Learning outcomes

Students should be able to:

define a feasible legal question and audience;
build and adapt a research strategy;
develop a supportable thesis or professional objective;
organize authority and facts into a coherent analysis;
use feedback to make consequential revisions;
supervise and disclose permitted AI assistance; and
produce a final work product responsive to its purpose.

Assignment sequence

The full sequence below is modular. Select the few checkpoints that will change the quality of the work.

Stage	Core question	Suggested evidence	Example AI rule
1. Proposal	Is the question feasible and important?	1-page problem, audience, scope	Brainstorming permitted; student selects and frames the question
2. Source plan	Where will reliable authority come from?	Search plan and preliminary source map	Leads permitted; every source must be independently found and read
3. Thesis	What does the student presently claim?	Thesis and strongest counterargument	Critique permitted; AI may not supply unverified propositions
4. Outline	How will the reasoning unfold?	Analytic outline tied to sources	Structure feedback permitted after student outline
5. Conference	What needs to change next?	Questions and priority memo	Student must be able to explain all submitted work
6. Draft	Does the argument work for its audience?	Complete or substantial draft	State whether drafting or only critique is permitted
7. Revision	Which feedback matters and why?	Revision plan or memo	Feedback synthesis permitted; decisions remain the student’s
8. Final	Does the product meet the outcomes?	Final work and use statement	Assignment rule controls through submission

Student-facing instructions

This project develops through the checkpoints listed below. Each checkpoint is part of the work, not a miniature final paper. Use feedback from one stage to improve the next.
The AI rule appears on each checkpoint. A permitted use at one stage does not authorize that use at another. Any source, quotation, legal proposition, or factual claim surfaced by AI must be independently located, read, and verified before it appears in submitted work.
At the final stage, submit a revision memo explaining the two or three decisions that most changed the project. The memo should connect feedback, research, and your own judgment; it need not narrate every edit.

AI rule

Publish a stage table with one of these labels for every required submission:

Independent: no generative AI for the stage’s substantive work.
Leads only: AI may suggest search terms, issues, or possible sources; students must independently locate and evaluate them.
Critique after submission/draft: students first create the artifact, then may seek bounded feedback.
Integrated and disclosed: AI may assist specified work, with verification and a use statement.

Name covered activities—brainstorming, research, outlining, drafting, revising, citation checking, editing—rather than relying on labels alone. The AI Syllabus Guide provides copyable policy language.

Disclosure

Use a compact running record rather than requiring entire chat exports:

Stage	Tool	Purpose	Material effect on submission	Verification
[stage]	[tool/none]	[task]	[brief description]	[source or method]

The final statement should identify only consequential assistance. Disclosure documents compliance; it should not be graded as evidence of quality unless reflective judgment is an outcome.

Deliverables

Recommended seminar version:

Proposal and scope statement
Source plan with preliminary authorities
Thesis, counterargument, and analytic outline
Conference preparation note
Substantial draft
Final product
Revision memo and AI-use record

Recommended large-class version: proposal, source-backed outline, final product, and 300-word revision memo.

Evaluation rubric

Dimension	Weight	Successful evidence
Question, purpose, and audience	10%	Defines a feasible problem and a clear use for the work.
Research and authority	25%	Uses relevant, reliable, and sufficiently current sources; verifies and accurately characterizes them.
Analysis and judgment	30%	Develops a supportable position, engages material counterarguments, and recognizes uncertainty and consequences.
Organization and communication	20%	Structure serves the reader; prose, citations, and format meet the assignment’s professional expectations.
Revision	10%	Responds thoughtfully to feedback and makes consequential rather than merely cosmetic changes.
Process compliance	5%	Required checkpoints and disclosures are timely, candid, and usable.

Give formative checkpoints completion credit or low weight unless performance at that stage is itself an outcome. A rough draft should be allowed to function as a draft.

Class-size variants

Small class: individual conferences at proposal and draft stages; customized stage rules; peer workshop.

Medium class: group proposal clinics; rotating peer review using one rubric dimension; brief instructor conference.

Large class: four checkpoints; structured templates; representative whole-class feedback; sampled process audits. Do not solve workload by demanding more documents than can receive feedback.

Accessibility and equity

Publish all deadlines and AI rules together so students can plan access and accommodations.
Offer an equivalent non-AI route when use is optional; required tools must be accessible and institutionally appropriate.
Permit approved assistive technologies at every stage and distinguish them from prohibited generative assistance.
Use flexible conference formats where appropriate.
Keep process grading modest so caregiving time, paid-tool access, and familiarity with research systems do not become hidden constructs.

Evidence and limitations

Staging follows backward-design and formative-feedback principles and is common in legal writing instruction. It can improve the opportunities to observe developing judgment, but the optimal number and weighting of checkpoints are not established for every context. AI-era guidance also cautions against assuming that a realistic or “authentic” final prompt alone protects assessment validity.

See Dawson et al., “Validity Matters More Than Cheating”, and Kofinas, Tsay & Pike, “The Impact of Generative AI on Academic Integrity of Authentic Assessments”. The latter studied selected UK business-school assessments and detection, not law-student learning.