Mark Dingemanse is Associate Professor in Language and Communication at the Centre for Language Studies, Radboud University, Nijmegen. His studies of social interaction in diverse languages uncovered several universals of conversational structure. With Andreas Liesenfeld, he developed the first global index of open source generative AI. Central to his current research is a focus on language as humanity’s most important and most flexible technology, which leads him to develop new perspectives on artisanal and artificial ways of languaging. His research group Futures of Language is funded by a Vici talent grant from the Dutch Research Council.
„Why are LLMs so irresistible?“
Perhaps the most momentous recent development at the intersection of the „digital“ and the „humanities“ is the rapid rise of large language models. As humanities scholars well know, new technology never comes out of thin air. From haruspicy to horoscopes and from ELIZA to LLMs, I aim to trace some of the ways people have been thinking with things, outsourcing judgement, and making sense out of non-sense. Sheer ubiquity and ease of use have a way of obscuring the interactional and hermeneutic processes at play, so I will devote some time to digging up the interactional foundations of our engagement with technologies of language, and thinking critically about the consequences of outsourcing scholarly processes to stochastic next token prediction. I will detail some of our work on the European Open Source AI Index, in which we carry out evidence-based openness assessment so that model providers can be held accountable, scholars can scrutinise generative AI, and end users can make informed decisions for or against deployment. As experts in meaning-making and connoisseurs of constructivist approaches to knowledge, scholars in the humanities have an opportunity if not an obligation to be at the forefront of critical AI literacy.
"Fertig – vorerst: Unfertigkeit als epistemischer Wert in den Digitalen Geisteswissenschaften"
Wissenschaft ist permanent „under construction“. Anstatt Unfertigkeit als Defizit zu betrachten, argumentiert dieser Vortrag für ihre Anerkennung als produktives Prinzip. Digitale Forschungsprozesse sind iterativ, offen und oft vorläufig – sichtbar in wachsenden Editionen, dem „Permanent Beta“ digitaler Werkzeuge oder der Volatilität digitaler Objekte. Doch Unfertigkeit ist mehr als eine methodische Notwendigkeit: Sie ermöglicht Offenheit, Multiperspektivität und Kritik, indem sie Erkenntnisprozesse instabil hält. Gerade die Digital Humanities zeigen, dass Modulierbarkeit und Versionierbarkeit nicht nur Aktualisierungsbedarf begegnen, sondern auch alternative Perspektiven eröffnen. Der Vortrag beleuchtet diese Chancen wie auch die Herausforderungen und fragt, was nötig ist, um Unfertigkeit produktiv zu gestalten.
