Assignment 4: The Slop Report
An independent investigation into a real, self-chosen case of suspected synthetic content, written for a public audience
Due: 28 May 2027 · Weight: 40%
The brief#
Find a real, current case where whether something is AI-generated actually matters to somebody — a disputed image, a suspicious workplace report, a platform’s labelling decision — and investigate it properly: multiple methods, a documented chain of reasoning, and a recommendation.
This is the capstone, and it is deliberately the least scaffolded piece of work in the course: no assigned passage, no assigned side, no assigned methods. The room it leaves is entirely in the choice of case and the judgement about which of the eleven weeks’ methods that case actually calls for — the strongest responses read as an investigation shaped by its subject, not a syllabus applied to whatever subject was nearest to hand.
What you submit#
A public-facing investigative report (1200–1800 words), written for a reader who was never in this course: the case, why it matters, every method applied and what it found, a stated confidence level, and a concrete recommendation — to a platform, an institution, or a reader — that the evidence actually supports. A five-minute version of the same report is presented in week 12’s casework session before the written version is due.
An unscoped week for an unscoped assignment is a bad combination, so it isn’t left until week 12: pitch your case, ungraded, in week 10’s casework session. Your tutor will tell you then if it’s too broad, too narrow, or not actually contestable — cheaper to hear two weeks out than to discover it yourself the night before the deadline.
The spec
The brief poses the problem and leaves room for a response. The spec is the fixed part: what a marker checks. Some lines can be checked mechanically; the rest call for judgement.
- submitted by the deadline as a public-facing investigative report
- investigates a case the student chose independently, not one assigned by the course
- stands on its own, without the author present to explain or defend it
How it is marked
Judged as a whole investigation rather than a checklist of methods — whether the case chosen was genuinely worth investigating, whether the methods applied were the right ones for that specific case rather than a default toolkit, whether the confidence claimed matches the evidence gathered, and whether the report would stand up if handed to a stranger with no access to its author.