Experimental Data on Cognitive Processing Differences in AI-Generated Art Evaluation

Three experiments tested the Pleasure-Interest Model of Aesthetic Liking to locate the source of bias against AI-generated art. The data, shared by Yongquan Wang on figshare under CC-BY-4.0, indicates the bias emerges predominantly during controlled, not automatic, cognitive processing. Providing interpretive semantic cues significantly mitigated the negative bias.

Use Cases

• Analyzing bias magnitude under automatic vs. controlled processing conditions based on the experimental design
• Testing the effect of interpretive semantic cues on bias mitigation as described in the results
• Modeling aesthetic evaluation responses using the Pleasure-Interest Model of Aesthetic Liking framework

Strengths

• Data is from three controlled experiments testing a specific cognitive model
• Results show a clear, measurable difference in bias magnitude between processing conditions
• License is permissive (CC-BY-4.0) for reuse

Limitations

• Dataset size is extremely small at 502.0 B, indicating limited scope
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

figshare

Collection Method

Data collected from three psychological experiments where only creator labels for AI-generated artworks were manipulated.

Freshness

Last updated 2026-06-03 04:26:49