Worked example · Art · Year 11
DCF: Citizenship → Digital rights, licensing and ownership
One pupil's completed sheet, arguing FOR the motion because that is the side they were given — their honesty box shows they personally disagree, which is exactly what the task is designed to produce. Show it after the debate, while pupils are drafting their closing speeches, so they can see what "evidence" means here. The ownership answers are factually checkable and should be marked as right or wrong.
| Claim | Evidence | Predicted rebuttal, and my answer |
|---|---|---|
| A competition is a comparison, and generated entries break the comparison. | The competition judge on the stimulus sheet: "if one entry took four hundred hours of oil work and another took eleven prompts, I am no longer comparing like with like." He is the one person quoted who runs a competition. | They will say a good picture is a good picture. My answer: a prize for painting is a prize for a named skill, the way a 100 m sprint is a prize for running. Nobody argues a bicycle is cheating — they argue it belongs in a different race. |
| Competitions with no rule about it end in public disputes, so the safe default is exclusion until there is a category. | The Colorado State Fair digital art category in 2022 had no rule and the result was widely disputed. In 2023 Boris Eldagsen won a Sony World Photography Award and refused it, saying he entered to test whether the competition was ready. | They will say the fix is disclosure, not exclusion. My answer: Eldagsen did disclose — after judging — and the award still collapsed. Disclosure without a separate category just moves the argument to the prize-giving. |
| The training data was taken without consent, so a prize rewards work built on other artists' unpaid labour. | The illustrator: pictures "scraped off the web without anybody asking me… I never licensed that." The UK's text and data mining exception covers non-commercial research only, and the 2024–25 Government consultation on Copyright and AI exists precisely because the commercial position is unsettled. | They will say the law has not decided, so I cannot call it theft. My answer: I do not need to. A competition sets its own entry rules and can decline to hand a trophy to a process whose inputs nobody can account for. |
Criterion 1: every claim is anchored to a named quotation or a dated fact from the sheet. Criterion 4: each row carries the rebuttal and the answer, so nothing is left for the other side to land free.
| Who owns an AI-generated image in UK law? | UK copyright normally needs a human author. For a work that is genuinely computer-generated with no human author, the Copyright, Designs and Patents Act 1988 gives copyright to the person who made the arrangements necessary for the work to be created. That is an odd rule and it is one of the few of its kind in the world. It is also not a rule about prompts: a prompt is usually far too short to be protected as a literary work, so "I wrote the prompt so I own the picture" is a much weaker claim than it sounds. |
|---|---|
| How long does it last, and how is that different from a painting? | 50 years from the end of the year the work was created. A painting is protected for 70 years after the artist's death, so a painting by a 17-year-old could be in copyright for well over a century, while the generated image beside it expires on a fixed date. |
| Does a CC licence give permission to train a model? | No, not by itself. A licence grants permission for the acts copyright controls, and Creative Commons has published guidance saying that much of how AI systems use content sits outside what copyright, and therefore what a CC licence, was built to address. So a CC BY photograph is not a "train on me" sign. That gap is why CC is now working on a separate signalling framework, and why the illustrator's complaint is not answered by telling her to license her work differently. |
Criteria 2 and 3: the 50-year term, the "arrangements necessary" rule and the limits of a CC licence are all stated correctly and contrasted with the 70-years-after-death rule for a painting.
Nobody on this side has argued that these tools should be banned from studios, and nobody has said the pictures are ugly. Our case is narrower and it has not been answered. A competition is a comparison, and the judge on our own stimulus sheet admits he can no longer compare four hundred hours of oil painting with eleven prompts. The Colorado State Fair found that out in public in 2022 because it had no rule; Boris Eldagsen proved the same point in 2023 by winning and then refusing. The other side says the skill simply moves. Perhaps it does, into art direction and into judgement, but then enter it in a category that judges those things. Our opponents' best point is that disclosure would be enough. Disclosure tells a judge what happened; it does not tell them how to rank two entries against one criterion. Until it does, keep generated work in its own class.
Criteria 1, 4 and 5: two dated facts, a direct concession to the other side's best point before answering it, and the argument stays on the competition question instead of drifting into whether AI art is "real art".
| The side I actually agree with | Against the motion. I use generative tools for mood boards myself and I think the designer is right that the skill moves rather than disappears. |
|---|---|
| The strongest argument against my own real view | The illustrator's transfer point. "The skill moves" is easy to say when you are the one holding the new tool and somebody else's eleven years of drawings are inside it. |
| One thing I would need to find out to settle it | Whether a named artist's work was actually in a given model's training set. Nobody in the debate could show that, and it is the fact the whole consent argument rests on. |
Criterion 5: separates the consent-to-training question from the quality question, and identifies the missing evidence rather than pretending the argument is settled.