In her talk “Large Language Models and Linguistic Insights” Anette Frank reflects on the question of how to gain insights about language from LLMs, and what new challenges and prospects this undertaking may generate – not only for AI researchers, but also for theoretical, empirical and computational linguistics.
She argues that bringing these fields closer together again could greatly impact future work in all areas of linguistics by asking new questions.
While most current research in this direction is devoted to syntax, she highlights aspects of this emerging paradigm in research conducted in the HD-NLP group: studying the generalization abilities of LLMs in language-based reasoning tasks; investigating the fusion of vision and language representation in Vision and Language Models, and how to integrate representations when combining language with Knowledge Graphs.
