Tag
Article argues against using generative AI in design and typography, emphasizing the millennia-long human evolution behind letterforms and warning that AI tools freeze cultural development by relying on limited training data.
This paper reframes model collapse in LLMs as a cultural transmission phenomenon, showing that iterated learning theory predicts a non-monotonic trajectory of compositionality under self-training, confirmed across multiple languages and models.
This study analyzes 130,882 images from 368 Artbreeder 'remix parties' over 13 months, finding that collective human-AI co-created images become simpler and converge toward common thematic attractors, while users paradoxically prefer to remix less novel images despite novelty producing more engaged children.