Time: about 12 minutes. Read the unit page first.
Part 1 — sort six questions (4 minutes)
Most of the questions below are about rights, attribution or disclosure. Label each, or say that it is none of these, and in one sentence say who could answer it.
| The question | |
|---|---|
| 1 | May I put this generated diagram in a poster I am selling? |
| 2 | Should the paper whose argument I am summarising appear in my bibliography? |
| 3 | Does my reader need to know an AI system drafted this section? |
| 4 | The output looks a lot like a figure I have seen before. What now? |
| 5 | I found this source through a generated summary and then read it. Do I cite it? |
| 6 | Am I allowed to use an AI system for this coursework at all? |
Some of these have answers this course can give you, some have answers your institution, a publisher or a contract can give you, and some do not have settled answers at all. Some sit partly in two places. Say which is which.
Part 2 — the two failures (3 minutes)
Here are two disclosure-and-credit statements for the same piece of work.
A. "This report was written with AI assistance."
B. A full bibliography of eleven sources, correctly formatted, and no other statement.
For each, write one line saying what a reader still does not know.
Then write a third version that does the job of both, in no more than three sentences.
Part 3 — your own worked case (5 minutes)
Take a real or realistic piece of work of your own where you used an AI system. Write down four things:
| Cite: whose ideas the work rests on, and where each citation goes | |
| Do not cite: what you will not cite, and why | |
| Disclose: how the work was made, in one or two sentences | |
| Unsettled: any rights question you cannot answer, and who could |
If you have no AI access, take the literature-review case from the reading, or a piece of your own work with an imagined outline from a system, and fill the four rows for that.
The second row is the one that does the teaching. A source you did not read, a paper you read but did not use, and a system that produced an outline are not sources of ideas you can cite — for three different reasons. Say which reason applies to each of yours.
Keep Part 3. The disclosure line is the same artefact Unit 17 asks for, and this is the version of it that also answers "whose ideas".
Now take the checkpoints
cp14-three-obligations is about Part 1.
cp14-settled-or-not is about knowing which questions have answers.
python3 selfcheck.py run cp14-three-obligations # or in the browser
python3 selfcheck.py run cp14-settled-or-not # or in the browser
Run the Python route from The verification lab and self-check, or use the browser self-check.
Optional extensions
Optional. These sit outside the core study time, are not required for the checkpoints, and nothing later in the course depends on them.
Read one set of terms of service (about 6 minutes). Find what the service you use says about ownership of output and about commercial use. Record the clause and the date you read it. This is a Unit 9 grounding exercise on a document that actually governs you, and most people have never opened it.
Find where your discipline has already decided (about 5 minutes). Many journals, conferences and professional bodies have published a position on declaring AI assistance. Find one in your field, note what it requires, and compare it with the three obligations in the reading. The interesting question is which of the three it addresses and which it leaves silent.
Marking guidance — open this once you have done the activity, to check your own work
Read this after attempting the activity.
Part 1 — the six questions
| About | Who can answer it | |
|---|---|---|
| 1 | Rights | Partly the terms of service, which exist and are readable; partly genuinely unsettled, since whether the output attracts copyright and whose it is varies by jurisdiction. |
| 2 | Attribution | This course. Yes — you are resting on their argument, and how you found it is irrelevant. |
| 3 | Disclosure | Yes, on this course's argument: saying how the work was made is part of being honest about it (Unit 17 is about how). Whether it is required, and in what form, is your institution's or publisher's rule. |
| 4 | Rights | Not settled in general. If it reproduces a substantial part of a protected work, ordinary rules apply; resemblance without copying is the genuinely open case. If being wrong would matter, as when you mean to publish or sell it, escalate, as Unit 12 does when the consequence is large and you cannot assess it. |
| 5 | Attribution | This course. Yes, cite it — you read it. Finding a source in an unusual way does not change what you owe its author. |
| 6 | Neither, and not this course's | Your institution. It is a rule about assessed work, not about rights or credit. |
So the split is: this course answers 2 and 5, and argues for the "yes" in 3; your institution or publisher answers 6 and decides whether 3 is required and in what form; the terms of service answer part of 1; and the rest of 1, and 4, have no settled general answer.
The two the course answers outright are 2 and 5, and they are the same question twice -- which is deliberate, because the second one is where people hesitate. Reading a paper you found through a generated summary makes it a source you read. Nothing about the route to it changes the obligation.
Question 6 is the one that most often gets attached to the wrong authority. It is not a copyright question and not an attribution question, and no amount of careful citation answers it.
Part 2 — what each version leaves out
Version A tells the reader how the text was produced and nothing about whose ideas it rests on. A reader cannot tell whether the claims are the author's, drawn from sources, or invented — and there is no way to follow anything up.
Version B tells the reader whose ideas and nothing about how the text was produced. A reader may reasonably assume the prose is the author's own work, and that assumption has not been either confirmed or corrected.
A third version that does both:
The argument in sections 2 and 3 follows [Author A, year] and [Author B, year], cited at the relevant claims. An AI system was used to produce an initial outline and to redraft two paragraphs for length; all sources were retrieved and read by me, and I am responsible for the final text.
Two sentences. The citations can be checked; the sentence on how the work was made cannot be confirmed from the text, so it rests on your honesty and on any records you kept. Note what it does not do: it does not claim the work is correct. That is a defence, and Unit 17 handles it separately.
Part 3 — the "do not cite" row
Three different reasons, and separating them is the point:
- A source you did not read. Not cited because you have not checked it, and a
citation asserts that you have. Unit 9's failures
fabricated-locator(the cited page does not exist) andmisattributed(a finding credited to the wrong work) are what happens when this rule is broken. - A source you read but did not use. Not cited because your work does not rest on it. A bibliography is not a reading list, and padding one implies support that is not there.
- The system that produced an outline. Not cited as a source of ideas, because a system is not someone whose ideas can be credited or questioned. If the outline shaped your work -- the awkward case in the reading -- that is a fact about how the work was made, so it goes in the disclosure, where the question is "how was this made". If your referencing style also asks for an entry for the tool or a retained chat, follow it: that records a tool, not an author, and it does not replace citing the underlying sources.
If your answer named the system as an author, that can come from a good instinct — wanting to be transparent. The instinct is right and the place is wrong: naming a system as an author states that something can answer for the claims, and nothing in the system can.
The checkpoints
cp14-three-obligations tests the separation. The trap is treating disclosure as
covering credit, which is the failure Version A shows.
cp14-settled-or-not tests something less common in a checkpoint: knowing which
questions have answers. The tempting wrong answers are the confident ones, in
both directions — and a course that trains you to catch confident-unsupported
claims should not make one here.
Your private activity record
Browser storage is not a permanent copy
Progress is kept only in this browser, profile and device. Private browsing, clearing site data, removing the profile, a browser reset, storage eviction or device loss can erase it. Keep important answers and contributions separately.
These notes stay in this browser unless you download a backup or activity log.