Hoppa till huvudinnehåll

Kalendarium

25

March

Large Language Models, RAGs, NLPs for Research and Education at Lund University

Use of LLMs at LU
Lund University and LLMs
Tid: 2025-03-25 10:00 till 15:30 Konferens

LU eScience Hub fika-to-fika event: LLMs, RAGs, NLPs for Research and Education at Lund University

We are excited to announce an engaging and inspiring event arranged by the LU eScience Hub and supported by AI-Lund, where the forefront of AI research meets education. This fika-to-fika event invites researchers, educators, and enthusiasts to dive into the transformative world of Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), and Natural Language Processing (NLP). Join us as we explore innovative applications in research and education, unlocking new possibilities for learning, collaboration, and discovery.

Starting with a morning fika to spark connections, the day unfolds with thought-provoking discussions and demonstrations. A light lunch will provide a refreshing pause, followed by more exciting sessions, collaborative ideation and an afternoon fika to cap off the day.

Whether you're an AI enthusiast, a seasoned researcher, or an educator curious about integrating cutting-edge tools into your teaching, this is the perfect opportunity to learn and connect with like-minded individuals. Don’t miss out on this unique chance to shape the future of AI in academia and beyond at Lund University.

Registration for the event is now closed 

Space is limited - please register to secure your spot! Let’s explore, learn, and innovate—together.

10:00 - 15:15 (on-site only)

Program and Speakers:

  • 10:00 - 10:15 Fika
     
  • 10:15 - 10:55. Rachel Forsyth - Educational development 
    Title. Revolution, Resignation or Rebellion?
     
  • 10: 55 - 11:55. Jan-Fredrik Olsen and students - Mathematics
    Title. AI in Mathematics Education: Perspectives from Both Sides of the Classroom
     
  • 12:00 - 13:00  - Lunch/Mingle
     
  • 13:00 - 13:45 Xuan-Son Vu - Media & Language
    Title. Generative AI in Education: Perspectives, Challenges, and Reliability Considerations
     
  • 13:50 - 14:35 Sverker Sikström - Cognitive Psychology
    Title. Generative AI Tools in Clinical Psychology
     
  • [CANCELED] - 14:40-15:00  - Sonja Aits -Experimental Medical Science 
    Title. Scientific text mining with EasyNER for medicine, life and environmental science
     
  • 14:35-14:45 Melvyn B. Davies - Mathematics
    Reflections and Discussions
     
  • 14:45-15:15 - Fika and Mingle
     

More about the presentations:

  • Rachel Forsyth - Educational development 
    Title. Revolution, Resignation or Rebellion?

Abstract. The availability of user-friendly tools to exploit large language models (LLMs) is having a marked impact on university education, with increased discussion about how they might affect teaching, learning, and the broader purposes of universities. Should institutions embrace the Revolution, embracing these tools to reimagine education for a digital age? Will they drift into Resignation, making only superficial adjustments to teaching and examination as LLMs quietly disrupt established norms? Or should they opt for Rebellion, resisting the integration of AI to safeguard human-centred learning and critical inquiry? This talk examines the transformative potential of LLMs, their impact on academic practices, and the critical choices facing educators and students as they navigate this transformation.

  • Jan-Fredrik Olsen - Mathematics

Title. AI in Mathematics Education: Perspectives from Both Sides of the Classroom
Abstract. This fall semester, I tried to adapt to and make use of AI in teaching my first-semester calculus course at Lund University. Given that current AI chatbots can solve most of the problems students are expected to work on, I saw this as a necessary step. However, I was surprised to find that students were initially reluctant to use AI as part of their studies, with some expressing frustration about being stuck for long periods of time because they could not get help from instructors.

In this talk, I will describe how I changed the course structure to help students use AI in a productive way, as well as analyse and discuss data from course evaluations, surveys, and the outcome of examinations. Based on these insights, I will reflect on when AI can be a valuable tool in a course - and when it might not be - and try to offer some general guidelines for the use of AI in teaching along the way.

  • Xuan-Son Vu - Media & Language

Title. Generative AI in Education: Perspectives, Challenges, and Reliability Considerations

Abstract: As Generative AI (GenAI) reshapes educational landscapes, its potential to enhance learning experiences is undeniable. However, ensuring reliability in AI-generated content remains a critical challenge. This talk explores the intersection of AI research and education, focusing on the technical foundations of GenAI, its mechanisms for content generation, and the inherent uncertainties affecting accuracy, bias, and trustworthiness. I will discuss key reliability concerns e.g., hallucinations, data quality, and model interpretability - and their direct implications for educational integrity. By examining current advancements and mitigation strategies, we highlight paths toward more dependable AI-assisted learning environments.

  • Sverker Sikström - Cognitive Psychology

Title. Generative AI Tools in Clinical Psychology
Abstract. Generative AI is particularly applicable in clinical psychology as language is the most important tool for psychologists to assess and treat mental illness. This presentation demonstrates TalkToAlba.com tools that are practically applicable to patients and clinicians. These tools include making automatic assessments based on transcriptions of patient-clinician meetings, and chatbots that take the role of psychologists making clinical interviews or CBT treatments. These tools improve accuracy, standardize, and speed up the assessment of mental health. It also allows a wider group of professionals to take on challenging patients. The experiences of patients and clinicians in using these tools are discussed.

  • Sonja Aits - Experimental Medical Science

Title. Scientific text mining with EasyNER for medicine, life and environmental science
Abstract. EasyNER is an automated text mining pipeline designed to extract and connect key information from the vast medical literature. The system tackles the challenge of processing over 35 million PubMed articles by using Named Entity Recognition (NER) to identify entities such as diseases, cells, chemicals, genes/proteins, and species. It employs deep learning models fine-tuned on the HUNER corpora and also supports dictionary-based NER, especially for COVID-19-related topics. Users can integrate their own models and dictionaries, and the pipeline produces ranked lists, graphs, and annotated text files for further analysis. The pipeline has been successfully applied to autophagy-related abstracts from PubMed and to a large collection of COVID-19 abstracts from the CORD-19 dataset.

 

Speaker Details:

  • Rachel Forsyth 
    Rachel Forsyth is a senior educational developer at Lund University in Sweden and was previously head of the University Teaching Academy at Manchester Metropolitan University in the UK. She has a background in curriculum design, digital learning, and assessment design and management. She is the author of Confident Assessment in Higher Education (Sage, 2022) and co-author of the forthcoming Generative AI in Higher Education: Transforming Teaching, Learning, and Student Experience (Bloomsbury, 2025). Rachel is also a Principal Fellow of the Higher Education Academy in the UK.
    https://portal.research.lu.se/en/persons/rachel-forsyth 
  • Jan-Fredrik Olsen 

    Dr. Jan-Fredrik Olsen is a senior lecturer in the Centre for Mathematical Sciences at Lund University, Sweden. He earned his Ph.D. from the Norwegian University of Science and Technology in 2009, with a dissertation focusing on the boundary properties of modified zeta functions and function spaces of Dirichlet series.  Dr. Olsen’s research interests include operator theory, time-frequency analysis, and complex analysis.  He has published extensively in these areas, contributing to the advancement of mathematical understanding in functional analysis and related fields.
    https://portal.research.lu.se/en/persons/jan-fredrik-olsen

  • Xuan-Son Vu

    Dr. Xuan-Son Vu is a senior researcher at the WASP Media & Language Arena and the founder of DeepTensor AB, a spin-off from the Department of Computing Science at Umeå University, Sweden. He received his Ph.D. from Umeå University, focusing on privacy-guaranteed machine learning with big data. Prior to this, he earned an M.Sc. in Computer Science from Kyungpook National University in Korea, specializing in Natural Language Processing and Machine Learning. Dr. Vu’s research centers on developing trustworthy machine learning and deep learning solutions, particularly in processing ubiquitous data from IoT applications and user-generated content. He has co-authored several neural models, including ppRNN, MGTN, and SGTN, aimed at enhancing data processing capabilities. In addition to his research, Dr. Vu serves as a reviewer for numerous journals and conferences and is a board member and publication chair of the Vietnam Language and Speech Processing (VLSP) association, which organizes international workshops to promote research in related areas. 
    https://www.umu.se/en/staff/xuan-son-vu/

  • Sverker Sikström

    Professor Sverker Sikström is a faculty member in the Department of Psychology at Lund University, Sweden, where he also serves as the head of the cognitive division.  His research primarily focuses on computational models aimed at understanding memory and cognition in the brain. Professor Sikström has developed platforms for assessing psychological constructs related to mental health and well-being, such as wordDiagnostics.com and semanticExcel.com. His work contributes significantly to the field of cognitive psychology, offering insights into the quantification of verbal responses and the underlying mechanisms of memory and learning. 
    https://portal.research.lu.se/en/persons/sverker-sikström 

  • Sonja Aits

    Sonja Aits is a research team manager and associate senior lecturer at the Department of Experimental Medical Science, Lund University, Sweden, where she leads the “Cell Death, Lysosomes and Artificial Intelligence” group. After earning her PhD from Lund University in 2010, she advanced her career with postdoctoral research at the Danish Cancer Society Research Center and a Visiting Senior Scientist position at the Peter MacCallum Cancer Centre in Melbourne, Australia. Dr. Aits’ work uniquely blends experimental cell biology with cutting-edge artificial intelligence techniques. Her research focuses on deciphering the mechanisms of lysosomal membrane permeabilization in cancer cells while developing deep learning and natural language processing tools for advanced biomedical image analysis and medical text mining. Through collaborative initiatives such as EpiHealth, LUCC, AI-Lund and eSSENCE—and her active role in several interdisciplinary profile areas—Dr. Aits is instrumental at integrating AI into experimental medicine. 
    https://portal.research.lu.se/en/persons/sonja-aits


Om händelsen
Tid: 2025-03-25 10:00 till 15:30

Plats
Lundmarksalen

Kontakt
alexandros.sopasakis@math.lth.se

Sidansvarig: webbansvarig@math.lu.se | 2017-05-23