MGK Colloquium: An AI workshop by Frank Richter

The next MGK Colloquium will host a workshop on AI as Research Infrastructure offered by Frank Richter. For Zoom link please contact MGK Coordinator Dr. Derya Nuhbalaoglu-Ayan via E-mail: Nuhbalaoglu@em.uni-frankfurt.de

Abstract:

AI as Research Infrastructure: LLMs, Coding Agents and Linguistics

“The talk asks a practical question: what role can current AI tools play in academic research, and where should we be careful? I will start with a high-level picture of what large language models are, how they are trained, and why this matters for working with them.

The aim is to make the discussion concrete. We will look at current systems, from chat interfaces to tools that can work with files and programs, and at examples where AI seems especially useful because the work can be checked: programming, data analysis, formal or semi-formal tasks, and parts of linguistic research workflows.

A central part of the talk will be a small case study from psycholinguistics. We will investigate to what extent a coding agent can produce an auditable statistical analysis and a draft results section from the background, hypotheses, materials, and experimental data of a published study, without being given the published results or the original analysis scripts. This should show where expert judgment remains essential: in checking the analysis, code, model choices, and interpretation.

Along the way, I will address questions from members of the GK, including AI use for coding, LaTeX and R, corpus work, experimental materials, prompting, model choice, academic writing, disclosure, and institutional rules.

This will be an informal talk, at times from a deliberately personal perspective, with room for questions and open discussion. I hope we leave with a clearer sense of what is useful, what is fragile, and what we need to learn to judge for ourselves. If you are curious, ask me how this abstract was written.”