World Wide Models: Literary Tools for Cultural AI
Large language models currently risk creating a monolingual AI landscape that overlooks cultural diversity.
The research explores how literary tools, such as comparative reading and translation theory, can help build more culturally literate models. It suggests that integrating “world literature” approaches allows AI to better handle cultural nuances and concepts that are difficult to translate.
For the public sector, this is important for ensuring AI services are inclusive and accurate across different regions. If models only reflect a single worldview, they may fail to serve diverse populations or misinterpret local contexts. Using these literary frameworks helps create AI that respects linguistic variety and cultural heritage.
How can we better balance global AI standards with local cultural identities?
#AI #PublicSector #CulturalDiversity #LLMs #DigitalGovernance
https://arxiv.org/abs/2607.02369v1