Related Experiment Videos
Resisting epistemic loss in AI image generation
Abdullah Hasan Safir1, Alan F Blackwell2, Ramit Debnath3
1Collective Intelligence and Design Group, University of Cambridge, Cambridge, UK.
Patterns (New York, N.Y.)
|June 22, 2026
Summary
Current AI image generators produce generic outputs. Viewing AI art as cultural processes and considering marginalized artists
Area of Science:
- Artificial Intelligence (AI)
- Sociotechnical Systems
- Digital Art and Culture
Background:
- Current vision-language models (VLMs) exhibit a tendency towards generating overly generic outputs, limiting creative and nuanced expression.
- Existing research often focuses on the technical aspects of AI image generation, overlooking broader societal and cultural implications.