Online Teilnahme:

https://zoom.us/j/95603412698?pwd=lWi9j5gEC1WV5VKIx8RbQOYKsaCJIQ.1

Meeting-ID: 956 0341 2698
Kenncode: 135860

Dr. Arno Simons, TU Berlin

Using LLMs for Interpretative Research in Science and Technology Studies

Abstract

This talk introduces large language models (LLMs) and examines how they can be used as tools for interpretative research in Science and Technology Studies (STS). It begins with an overview of what LLMs are, outlining different model architectures, tasks, and training paradigms, and how these elements interrelate in shaping model behavior and interpretative potential. The second part presents concrete methods for using LLMs in STS research, including classification, entity and relation extraction, text similarity analysis, topic modeling, and interpretative sparring. Each method is illustrated with concrete examples that show how LLMs can assist in analyzing and interpreting sociotechnical materials. Throughout, the talk sheds light on the specific opportunities and challenges of applying LLMs in STS, taking into account the field’s methodological commitments and the nature of its data—such as working with historical texts and situated discourses that differ from the contemporary language environments in which LLMs are trained. It concludes with five key lessons on how to employ and critically engage with LLMs in interpretative research.

CV

Arno Simons ist Soziologe, Politikwissenschaftler und Philosoph. Er erforscht, wie Wissenschaftler sowohl untereinander als auch mit Entscheidungsträgern, Journalisten und der breiten Öffentlichkeit kommunizieren. Dafür kombiniert er computergestützte Methoden der neuronalen Sprachmodellierung und Korpuslinguistik mit Verfahren der sozialen und bibliometrischen Netzwerkanalyse sowie qualitativen Ansätzen der Diskursanalyse und der Grounded Theory.