AGRI402361 Generative AI: Skills for robust and responsible GenAI usage in scientific work

Art
Vorlesung und Übung
Semesterstunden
1.5
Vortragende/r (Mitwirkende/r)
Hummel, Sebastian , Zitek, Andreas , Büschl, Christoph , Groß, Angelina Sarah
Organisation
Bioanalytik und Agro-Metabolomics
Angeboten im Semester
Wintersemester 2026/27
Unterrichts-/ Lehrsprachen
Englisch
Lehrinhalt

Generative AI (GenAI) in the form of large language models (LLM) and transformer-based image generators are reshaping research by accelerating tasks such as model complex systems, find and summarize literature, generate ideas, and even draft entire manuscripts. While these tools are immensely powerful, they also come with inherent limitations/risks and require informed, critical human oversight to ensure reliability, reproducibility, and scientific integrity.
This lecture equips students – especially in the life sciences – with the critical AI literacy needed to use GenAI responsibly and productively. Participants of the course will learn about A) the technical foundations of GenAI, B) interacting with GenAI tools in the scientific processes, C) Principles of mindful and responsible AI application/usage, and D) the impact of GenAI on human society as well as the environment and also themselves (Tadimalla et al. (2025), AI literacy as a core component of AI education, doi:10.1002/aaai.70007). By the end, students will not just be able to use GenAI tools, they will be able to scrutinize them, document and validate their outputs, reflect critically if GenAI-usage is advantageous and justified, and integrate them into rigorous, accountable scientific practice.
Inhaltliche Voraussetzungen (erwartete Kenntnisse)

Preliminary experience of large language model (LLM)-based AI tools such as Scopus-AI, ChatGPT/Gemini/Claude/Academic AI or similar.
Basic knowledge of matrix/vector/scalar operations and statistics are advantageous.
Lehrziel

By the end of this course, students will have acquired critical AI literacy and have learned competences for a reflected use of GenAI in scientific work (as required by the EU AI act).
More specifically, students will be able to:

A) The technical foundations of GenAI
- Explain the technical foundations of GenAI systems, including tokens, embeddings, attention mechanisms, transformers, probability distributions over datasets, model architecture, and quantization
- Understand basics of matrix/vector multiplication, chain rule, probability distribution in the context of GenAI
- Differentiate between training and inference processes
- Compare cloud-based and local deployment setups
- Apply simple language models in hands-on tasks
- Evaluate the creative potential and limitations of GenAI outputs

B) Interacting with GenAI tools in scientific processes
- Apply GenAI for literature review, generate scientific ideas (ideation), assist in programming and writing, as well as automated peer review
- Choose the appropriate tools and deployments
- Master common strategies for prompting (e.g., one-shot, iterative refinement, chain-of-thought, AIM-approach)
- Understand hallucinations and that output of LLMs must be manually verified for correctness in the primary literature
- Apply Retrieval Augmented Generation (RAG) and Model Context Protocol (MCP) extensions

C) Principals of responsible AI application/usage
Explain key ethical principles of AI (e.g., the Vienna Manifesto on Digital Humanism)
- Examine the use of generative AI in relation to research integrity (e.g., The European Code of Conduct for Research Integrity)
- Identify and analyze different types and sources of bias in AI systems
- Assess known risks of GenAI use in scientific contexts
- Apply principles of research integrity by critically reviewing, verifying, and refining AI-generated content (human-in-the-loop)
- Explain copyright and licensing issues related to AI-generated outputs
- Classify AI use cases according to the framework of the EU AI Act

D) The impact of GenAI on human society as well as the environment
- Describe the societal impacts of GenAI, including biases, human oversight, and socio-economic implications
- Understand the resource-usage, environmental impact, and the ecological footprint of GenAI systems
- Recognize and discuss trade-offs between performance, scalability, and sustainability
Noch mehr Informationen zur Lehrveranstaltung, wie Termine oder Informationen zu Prüfungen, usw. finden Sie auf der Lehrveranstaltungsseite in BOKUonline.