Group work, and shared rules

The submission is joint and the consequences are not

When four people submit one document, all four are answerable for all of it. If one section contains a fabricated citation or generated prose that the group’s declaration did not cover, the investigation involves everybody, and I did not know is a weak position that also damages a friendship.

This is the most common way careful students end up in integrity meetings, and it is almost entirely preventable by a fifteen-minute conversation nobody wants to have at the start.

The conversation, in one message

Send it before the first working session:

Before we start: how are we using AI on this? My suggestion: fine for generating questions, critique of our own drafts, translation, and checking understanding. Not for prose that goes in the document. Everyone opens every source they cite. Everyone writes their own section. If anyone wants to work differently, say now rather than at the end.

Awkward for about ninety seconds. Then it is settled, in writing, with a timestamp, and everybody knows the standard.

What matters is not that everybody adopts the strictest possible rule. It is that the rule is explicit and shared. A group where two people think generated drafting is fine and two think it is cheating will have a problem, and it will arrive at the worst moment.

Check the actual policy first

Course policies on group work sometimes differ from individual-work policies, and some require a group declaration listing what each member did and what tools were used. Read yours before agreeing anything, so the group rule is at least as strict as the institutional one. If your policy is silent on group work, ask the module leader in writing: the reply is useful evidence later.

Working in a shared document

Practical arrangements that prevent most problems:

• Google Docs or Word online, with version history on. Every contribution is attributed and timestamped automatically. This is the single most useful thing you can do, and it costs nothing.
• Never paste a finished block into a shared document. Write in it, or paste in pieces with edits, so the history shows work rather than an arrival. A large block appearing at once is what triggers questions, whoever wrote it.
• One person owns each section, named at the top of the draft.
• One person does a consistency pass at the end, and says so.

The reference list is where it goes wrong

Combining four bibliographies is the moment fabricated citations enter a document, because the person merging did not read anybody else’s sources and assumes they were checked.

The rule that fixes it: whoever cites it, opens it, and says so. Before submission, each person confirms in the group chat that every reference in their section was opened and that they can say which page they used. That message is your record, and it takes two minutes.

If someone will not confirm, the reference comes out. This is not distrust; it is exactly the standard everybody would want applied to their own name on the cover.

Dividing the work, revisited

From the study-groups lesson, but with the assessment consequence attached. Splitting topics so each person only learns a quarter is efficient for producing a document and poor for learning, and in any group assessment with an individual component (a viva, a reflective piece, a question in an exam) it is directly costly.

The arrangement that survives: split the writing, share the understanding. Everybody reads the whole draft, and everybody should be able to defend any section. Half an hour before submission, take turns explaining each other’s sections. It is the fastest possible check for anything nobody actually understands, which is also the fastest possible check for anything nobody actually wrote.

If it goes wrong

If you find, late, that a section contains work you are not comfortable submitting: raise it with the group first, immediately, in writing. If it cannot be resolved, tell the module leader before submission rather than after. Disclosing early is treated completely differently from being found later, at every institution, and it is also the only version you will feel fine about afterwards.