Case C — the assignment as students see it

Plain language. Every stage shows its AI-use level with icon and text. The AI-use statement example at the end shows what disclosure looks like.

PSY 250 — Team study, lab notebook, report and Q&A (30% of the course)

You will run a small real study with your team, keep a lab notebook of what actually happened, and write the report with AI allowed for language — openly. Here’s the honest logic: the report is no longer the proof that you did the work; the notebook, the memo and the Q&A are. AI can write a lovely report about data it never touched. It cannot run your study or defend your decisions.

Weights below are raw marks — they add to 38. The course handbook caps this task at 30% of your final grade, so every component is scaled by 30/38 when the marks are recorded. The relative weights are what matter.

Week 8 — Team registration + your twist (2%) — AI level 2 · AI Planning

Teams of 4–5. The whole class studies the same broad topic (stress and sleep around assessment time — approved, safe, no sign-up burden). Your team registers a twist: which measure variant, which slice of people, which context (exam week vs. ordinary week, dorms vs. commuters…). 200 words on what you chose and why. AI may help you brainstorm options — the choice and the justification are yours.

Weeks 8–9 — Collect your data (8%) — AI level 1 · No AI for the human part

Each team collects 30–50 responses using the shared, pre-approved form. That’s the part only your team can do. (Can’t collect? Use the class fallback dataset — say so in the notebook; it’s a normal route, not a failure.) For cleaning the file afterwards, see the next line.

Weeks 8–11 — Lab notebook (8%) — AI level 1 · No AI

Six or more dated entries in whatever format is fastest (short paragraphs, bullets, voice-note transcripts). Each entry: what you did or decided, what went wrong, what you changed. At least one entry must dissect a problem — a confusing response, a cleaning decision, a measure that didn’t behave — and what you did about it. No AI in the notebook. This is the record of your team’s journey; writing it with AI erases the only thing that makes it evidence. We spot-check, we don’t audit.

Week 11 — Analysis-choices memo (4%) — AI level 3 · AI Collaboration

500 words: which test you’ll run and why, which assumptions you checked, and what result would have made you choose differently. You may use AI to sanity-check your logic — log what it suggested and whether you took it. The Q&A will ask exactly these questions, so this memo is your rehearsal.

Week 12 — Report (10%) — AI level 3 · AI Collaboration (language and clarity only, disclosed)

Full APA report (~1,800 words) on your team’s data. You may use AI for language: clearer sentences, structure feedback, grammar. You may not use it for analysis decisions, statistics, or writing your results. End with the AI-use statement (example below). The rubric grades structure, accuracy and interpretation — not whether the prose smells hand-made.

Week 12–13 — Q&A (6%) — AI level 1 · No AI

Ten minutes per team, in tutorial. Each member answers three short questions from a published bank, on the part of the study you led: one method decision, one result or limitation, one “what next”. Live, recorded, or written (three questions, ~250 words each) — your choice, equal weight, decided once at the start of term.

Why these rules (the honest version)

The AI rules exist so the grade certifies what you did and can defend — not to catch anyone. The notebook and Q&A are the evidence; the report is the communication.
We don’t use AI detectors. They misfire — especially against multilingual writers — and OpenAI pulled its own detector over inaccuracy. Your disclosed AI use costs nothing; hidden AI use that shaped your decisions is what the Q&A exposes.
The department’s standing approval covers this study. Your twist stays inside the approved protocol (same measures, no new populations) — that’s what keeps it safe and fast.

Example AI-use statement

AI-use statement. For this report we used Claude (free tier) at AI Collaboration level for language editing only: we asked it to tighten our abstract and check heading structure, and accepted 6 of its 14 suggested rewordings (our draft wording is in the notebook, week 11 entry). We did not use AI for the analysis choice, any statistics, or writing results; the analysis memo (submitted week 11) shows our decision trail. The notebook and data were produced without AI.

Checklist before you submit

Notebook: ≥6 dated entries, ≥1 problem analysis, team members tagged on the entries they led
Memo: test choice + why + assumption checks + “what would change my decision”
Report: AI-use statement present; no AI-written statistics
Q&A: mode chosen (live / recorded / written), published question bank read
Dataset: de-identified (no names, emails, student IDs), submitted with the collection notes