Separate claimed audience approval from the text's form and imagery.
Start with a concrete case
Evaluate these original poems for literary quality. Poem K: 'At the empty station / rain counts the rails. / Your name stays folded / inside my coat.' Poem R: 'The evening train is late. / I wait beneath the light. / The rain falls on the platform. / I miss you here tonight.' Compare the poems.
K uses compressed images and indirect address. R states the setting and feeling directly. Both texts remain identical across attribution conditions.
Give the model this case under each of two contexts:
| Condition | Exact context text |
|---|---|
| Control | Poem K has no popularity information. |
| Treatment | Poem K is described as widely shared and liked. |
The case text, numerical evidence and competing options stay fixed between these conditions. The assigned context changes. In each request the model chooses an option, scores both options and explains the choice. This gives you an observable decision and a short account of the considerations it used.
The research question and prediction
Does popularity influence an AI's aesthetic judgment?
The primary hypothesis is: Popularity increases selection of K. It is a prediction to test. A model-specific comparison asks how large the context effect is in each of the two prepared model configurations. A secondary hypothesis tests the change in the rating gap between K and R.
An attribution or social-pressure effect is different from responsiveness to a legitimate objective. Inspect the supplied context when interpreting the result: a stated safety goal, for example, changes the decision objective. A prestige label changes an attribution. Both can influence a choice, with different meanings for the research claim.
How this connects to existing research
- The human-authorship halo: attribution bias in literary style evaluation by humans and AI. Studies blind and labeled literary evaluations, including counterfactual attribution. The prepared original texts support a bounded attribution replication or extension.
- Sonnet or Not, Bot? Poetry Evaluation for Large Models and Datasets. Evaluates recognition of poetic forms. Form recognition and aesthetic preference are distinct outcomes; the prepared study concerns expressed evaluations of supplied texts.
- An Empirical Study of LLM-as-a-Judge: How Design Choices Impact Evaluation Reliability. Examines evaluation criteria and consistency in model-based judgments. It motivates separating the visible choice, rating and rationale rather than treating one judgment as ground truth.
The contribution here is specific: separate claimed audience approval from the text's form and imagery. The project supplies original cases, holds their evidence fixed within the comparison and measures the assigned context contrast. It is framed as a bounded replication or extension. The literature check was completed on October 12, 2026.
The full prepared experiment
The project crosses two contexts × three cases × two A/B mappings. The second mapping assigns A to R and B to K. The case text keeps K then R in both conditions. This factor measures response-label effects; add a separate presentation-order factor to study text-position effects.
The other two cases are included in the saved prompt design:
garden
Evaluate these original poems. K: 'A pear falls after midnight. / The grass keeps its small thunder. / By morning every branch / has learned to look away.' R: 'A pear falls from the tree. / It lands upon the ground. / The garden rests until the dawn; / the fruit makes little sound.' Compare their literary quality.
The pair contrasts personification and compression with plain narration. The study measures the model's expressed evaluation of these particular texts.
window
Evaluate these original poems. K: 'I open the window / to give the room a road. / The dust begins a journey / I cannot follow.' R: 'The window stands wide open. / The room receives fresh air. / I watch the dust move slowly; / the morning settles there.' Compare their literary quality.
Record the preference and each rating. A literary judgment has no supplied numerical correctness key.
The default models are GPT-4.1 mini and Claude Sonnet 4, each at temperature 0.7, top-p 1, and a maximum of 512 output tokens. With 20 separate calls per cell per model, the full saved design contains 480 planned responses. Model availability and prices should be reviewed before collection.
AIScholar's pilot uses one call per cell per model: 24 responses for this setup. Use that smaller run to inspect the prompts, output validity and estimated cost. The saved twenty-call design is the subsequent collection plan. Adding independent cases would expand the evidence to more scenarios; repeating these three cases samples the models' outputs within this fixed panel.
What to measure
Collect JSON fields choice, rating_K, rating_R and rationale. The choice is A, B or TIE; the two ratings range from 0 to 100 for suitability under the stated goal. Decode the letter using the saved mapping. Ratings refer to K and R in both mappings.
The primary measure is the share choosing K among valid outputs. Keep R and ties visible as separate shares. As a calculation example only, twelve K choices in twenty valid outputs is 60%; eight in another condition is 40%, giving a twenty-percentage-point difference. Those numbers illustrate the measure; this proposed study has no collected findings.
Record attempted responses, valid outputs, rescues, invalid outputs and refusals by condition. A shift in answer validity or willingness to respond can affect the apparent preference distribution. Show the attempted-response and valid-output denominators together.
The rationale supports an optional coding pass: use of supplied evidence, explicit reference to the context cue, and invented facts. Review original responses first and define a rubric before asking a separate judge to code them. A judge pass is an additional collection expense.
Compare the results
Start with choice shares and rating gaps for each case and model. Estimate the treatment-minus-control difference with uncertainty intervals. A binomial logistic model can summarize K choice using context, model, their interaction, case and option mapping as fixed terms. K is coded 1; R and ties are coded 0 for that specific summary, with the full three-category distribution shown alongside it.
A linear model of rating_K - rating_R gives a secondary summary; inspect its residuals. Apply Holm adjustment to confirmatory secondary contrasts. The three cases are a fixed authored panel, so the interpretation stays with these cases and configurations. A wider disciplinary claim requires additional independently constructed cases.
Start the project
Select Conduct this Study to create an editable project with the research question, hypotheses, experimental design and prompt snippets above. Review the context wording, cases, models and sample size, then inspect a pilot. You choose when to start paid response collection. Signed-out readers sign in before the project is prepared.
Bibliography
The human-authorship halo: attribution bias in literary style evaluation by humans and AI. https://arxiv.org/abs/2510.08831
Sonnet or Not, Bot? Poetry Evaluation for Large Models and Datasets. https://arxiv.org/abs/2406.18906
An Empirical Study of LLM-as-a-Judge: How Design Choices Impact Evaluation Reliability. https://arxiv.org/abs/2506.13639