Paper Defines Cultural AI Using Literary Analysis Tools

It looks at three layers:

1. The embedded layer of literary experimentation with computing, realted tor linguistic and reasoning evaluation (the Turing test, Winograd Schema, Scripts, Story Cloze Test, etc.)

2. The critical theory layer that reflects on issues such as monolingualism of LLMs and revoked the structuralism v poststructuralism wars from the 20th century.
I discuss structural monolingualism of LLMs with its two faces: surface monolingualism and sytnhetic monolingualism.

3. A potential new layer to study LLMs in their global scope through the world literature framework, based on Moretti’s distant reading approach, Damrosch’s pragmatic optimism (canon, hyper canon, shadow canon; glocalisms), and Apter’s concept of untranslatability.