Text is a technology in itself, a critical form of infrastructure, and a product and medium of culture.  Now, after 6,000 years of handcrafted text production, fluent, human-like text is produced in part or entirely by statistical machines, whose generative capacities are unlimited in pace and scale. With funding from DNRF Grundforskningsfond for up to ten years, the TEXT Center will create a new model for interdisciplinary basic research on text in the new era of Large Language Models (LLMs).

• The TEXT Center fosters interdisciplinary collaboration to address the multifaceted implications of LLMs. Recognising that LLMs represent a significant technological and cultural shift, the Center unites researchers from diverse fields to investigate their impact on communication, creative practice, knowledge work, media dynamics, and our understanding of human language and cognition. This collaborative approach aims to move beyond a purely technological perspective to consider the broader human and societal dimensions of this rapidly evolving technology.
• A core focus of the TEXT Center is the identification and evaluation of ‘LLM-hard’ problems across domains. The Center’s research is structured around four work packages that delve into specific challenges where LLMs currently fall short. These include exploring the evolving nature of creative practice and writing, developing rigorous evaluation techniques for foundation models, refining practical interventions in education, business, and research, and probing the linguistic performance of LLMs as both a tool and an object of study in linguistics and cognitive sciences. By focusing on the challenging problems, the TEXT Center aims to drive innovation in both understanding and effectively utilising LLMs in these fields, evaluated against existing theories, models, and practices .
•  By adopting a human-centred and critically engaged approach to LLM research, prioritising responsible research practices, ethical considerations, and knowledge exchange, TEXT aims to provide scientifically-grounded insights on the affordances and limitations of LLMs for our collective text culture.