Large Language Models and Text Analysis – Scoping the Risks and Potential

The VMU Vytautas Kavolis Transdisciplinary Research Institute invites the academic community to a methodological workshop on Large Language Models for research tasks.

• Date – 10 December 2025 at 10 AM.
• Duration of the workshop – 90 minutes.
• Language – English.
• Format – online workshop via MS Teams.
• Please register for the workshop here.  All registered participants will receive the MS Teams link to the workshop one hour before the training.
• Certificates will be issued to participants upon request after the workshop.
• Trainer: Erkki Mervaala, Researcher at the Finnish Environment Institute, doctoral candidate at the University of Helsinki.

Content of training

In this workshop, we will cover use cases, risks and possibilities for generative artificial intelligence – especially large language models – in research tasks. This workshop partially addresses the findings of the article “Out of Context! Managing the Limitations of Context Windows in ChatGPT-4o Text Analyses” by Erkki Mervaala and Ilona Kousa (2025), but also expands the theme to current State-of-the-art (SOTA) models in use by both the public and the research institutions.

During the workshop, we run through some key concepts necessary to understand how LLMs work (f.ex. “context window”, “zero-shot”, and “tokens”), and what kinds of caveats there are considering automating and outsourcing text analysis tasks to generative artificial intelligence – including ways how to manipulate the results via  “LLM hacking” and uncover how f.ex. system prompts function in both local and privately owned LLM services.

And finally, the workshop will cover some of the environmental and ethical aspects of using such tools and services, from energy usage and the natural resource extraction industry to oppressive working conditions within the model training industry, and to the potential “cognitive death” and skill deterioration related to the extended use of such tools.

More about the trainer

The convenor of the workshop, Erkki Mervaala, has worked as a researcher at the Finnish Environment Institute since 2019, and is finishing his PhD in the University of Helsinki’s Interdisciplinary Environmental Sciences doctoral programme. His research has covered addressing and analysing caveats of GenAI usage in academic research, especially text analysis, but also utilising the use of natural language processing (NLP) and other machine learning -based tools for media analysis. His previous research has addressed the contradictions and representations of climate change and economic growth in the Finnish media. His background is in computer science and journalism, and he has previously worked as a foreign correspondent and in media consultancy.

The training will be moderated by Dr. Gintarė Pocė, Head of Project Activities at V. Kavolis Transdisciplinary Research Institute.

For more information about the workshop, please send an email to vki[et]vdu.lt.

This workshop is one of the activities of the project “Strengthening of R&D activities of the Vytautas Kavolis Transdisciplinary Institute of Social and Humanities Sciences (SOCMTEP)”. The project is financed by the Lithuanian Science Council and the Ministry of Education, Science and Sports of the Republic of Lithuania, contract no. S-A-UEI-23-13 (12/27/2023).