← Back to Contents
Note: The first drafts of these discussion questions were generated using Claude (Anthropic's AI assistant) and then reviewed and edited for the in-class use of this course.
Week 6 • Discussion

💬 Week 6 — Discussion Questions

To think about before class

These are example discussion points for you to think about before class. You are not expected to engage with all of them — pick the ones that speak most directly to your own research, and bring two or three rough answers to the in-class session. The full description of how to use these pages, including what the question tags mean, is on the Week 1 Discussion page.

Sub-lessons

Writing as Thinking

  1. Calibrate “Writing IS thinking” is the lesson's core claim. Pick a recent piece of your own writing where the act of drafting genuinely changed what you thought. What would have been different if AI had drafted the first version?
  2. Apply Draw your own spectrum of AI writing assistance from “clearly useful for me” to “clearly inappropriate for me.” Place three concrete writing tasks from your current project at specific points along the spectrum. Where is your hardest case?
  3. Critical The “cognitive dissonance” framing implies a tension between the writer-as-thinker tradition and the productivity case for AI writing. Is that tension genuine, or has the productivity case been framed in a way that makes the conflict look sharper than it is?
  4. Connect Week 4 asked you to draw your own line between AI-as-instrument and AI-as-collaborator; Week 5 then made you live with that line in literature review. Now apply the same line to writing. Where does “writing IS thinking” push the line one way that literature-review work did not, and is the line you draw for writing the same line you drew in Weeks 4 and 5?

Research Ideation with AI

  1. Calibrate Generate ten research ideas with AI on a topic you know well. Read them through the “idea monoculture” lens. How much is the ten genuinely diverse, and how much is it the same handful of ideas in different phrasing?
  2. Apply Use a prompting strategy from the lesson to brainstorm extensions for your own current project. Pick the most surprising idea AI gave you. Would you actually pursue it — and what would have stopped you finding it without the tool?
  3. Critical The lesson warns about anchoring and premature convergence. Honestly: in your last AI-assisted brainstorm, were you anchored or convergent? What would you do differently next time?
  4. Connect Week 5 introduced literature-review tools that rank and surface what the field is “saying” about your topic; this sub-lesson asks whether AI-assisted ideation flattens the diversity of what could be said next. Put the two together. Are AI-mediated literature search and AI-mediated ideation pushing your field toward the same kind of homogenisation, or toward different ones?

AI Writing Tools Landscape and Honest Assessment

  1. Calibrate The lesson sorts tools into four categories. Pick the tool you use most and place it. What does the category framing surface, and what does it hide?
  2. Apply For a non-native-English-speaker doing research in English, which category of tool is most defensibly useful and which most concerning? Be precise about the use-case that flips the answer.
  3. Critical The homogenisation problem and the multilingual-equity angle pull in opposite directions: more AI assistance helps marginalised English writers, but also flattens written style. Which side of the tension do you actually take, and where does that show up in your practice?
  4. Connect Week 5's honest assessment of paid versus free literature tools warned against marketing language standing in for capability. Apply the same scepticism to the writing-tools category here. Where would you trust the published effectiveness evidence on AI writing tools, and where would you assume the evidence is doing more PR work than research work?

Scientific Integrity and the Writing Pipeline

  1. Calibrate Place a recent piece of your own AI-assisted writing on the integrity spectrum the lesson lays out. Did the writing fall where you expected it to, or further toward “harmful” than you would have thought before reading the lesson?
  2. Apply Pick a journal you intend to submit to and write a one-paragraph integrity statement that would be honest and that the journal would accept. Where would you most have to negotiate with yourself?
  3. Critical The lesson describes AI detection as an “unreliable arms race.” Should universities and journals stop relying on detection entirely, double down, or do something else? Pick one and defend it.
  4. Connect Week 4 asked you to commit to a personal disclosure practice; Week 5 raised the issue of hallucinated citations as a writing-pipeline failure. This lesson sharpens both. Where in your current draft (literally, pick a section) would your Week 4 disclosure practice and your Week 5 citation-verification practice need to land if you applied them honestly today?

Using AI to Review Your Own Work

  1. Calibrate Run an AI pre-review on the most recent draft you have. Pick the single most useful thing the AI flagged and the single most useless thing. What does the comparison say about where AI peer review is currently strong and weak?
  2. Apply Design a personal multi-agent review protocol for your next paper (e.g. one model as a sceptical reviewer, one as a methods-checker, one as a writing editor). What instructions does each agent get, and how do you reconcile contradictory feedback?
  3. Critical The lesson distinguishes what AI can and cannot reliably judge. Is the “cannot judge” list stable, or do you expect it to shrink fast? What would force you to expand it again?
  4. Connect Week 5's hallucinated-citations lesson and Week 4's integrity framework both pointed at the same thing: AI fluency is not AI reliability. Where in your AI pre-review workflow does the fluency-vs-reliability gap show up most? What test would you build into the pre-review specifically to catch the fluent-but-wrong kind of failure?

Building Your AI Writing Workflow

  1. Calibrate Apply the reverse-outline technique to a section of your own current draft. Did the AI's reconstruction match what you thought the section was doing? Where it diverges, who is right — you or the AI?
  2. Apply Sketch the version-control practice you would actually keep up with for AI-assisted writing. (Be honest about what you will and won't maintain six months in.)
  3. Critical The “principled workflow” assumes you stay in charge of the writing. What concrete signal tells you that, in a given session, you have stopped being in charge?
  4. Connect Week 5's “Building Your Research Workflow with Claude” sub-lesson laid out a similar three-tier ladder (better prompting → Projects → Skills) for literature work. Compare the writing-workflow ladder here with the literature-workflow ladder there. Are they the same ladder applied to different tasks, or do they call for different rungs in different orders?

Hands-On Activities and Assessment

  1. Calibrate In the Writing Process Experiment, you write the same paragraph by hand and with AI. Read the two side-by-side. Which is more honest about what you actually know? Which would you submit?
  2. Apply Run the Audit Exercise on a paragraph an AI drafted for you. List the three changes you made that the AI couldn't have. Are those three changes the heart of your contribution as a researcher?
  3. Critical The Prompt Engineering for Ideation activity assumes prompting can systematically improve idea quality. Is there a ceiling on what prompting alone can do, regardless of skill?
  4. Connect Week 5's Hands-On activities introduced a verification protocol for AI-assisted literature work. The audit exercise here is the writing-side equivalent. Compare the two: which one was harder for you to do honestly, and what does that say about where your blind spots actually sit?