Responsible AI for Researchers

Artificial Intelligence (AI) is becoming part of many stages of the research process, ranging from developing ideas and reviewing literature to collecting and analyzing data, writing code, preparing publications and grant proposals, conducting peer review, and communicating findings. Used responsibly, AI can support researchers in working more efficiently and exploring new possibilities. At the same time, its use introduces risks that are not always immediately visible. AI systems may produce fabricated or inaccurate information, reproduce biases, generate non-reproducible results, expose confidential information, or encourage an inappropriate delegation of scholarly judgment and responsibility. Responsible AI use therefore requires more than knowing how to operate individual tools. Researchers must be able to assess whether a particular use of AI is scientifically sound, ethically acceptable, secure, transparent, and consistent with applicable legal, funding, publishing, and disciplinary requirements.

In this workshop, we examine responsible AI use across the entire research cycle. Participants learn to distinguish productive AI assistance from questionable or prohibited practices, assess ethical, researchintegrity and governance risks, and select appropriate safeguards. In the hands-on part, participants apply these principles to their own research and develop a concrete plan for using AI responsibly in a current or planned project.

Learning Objectives
By the end of the workshop, participants will be able to:

1. Explain the main capabilities and limitations of generative and other AI systems and how problems such as fabricated outputs, bias, limited reproducibility and inappropriate reliance on AI can affect research quality and integrity.
2. Distinguish between research-ethics, research-integrity and research-governance considerations and identify relevant risks at different stages of the research cycle.
3.  Evaluate specific uses of AI in their own research and select appropriate safeguards concerning human oversight, verification, confidentiality, data protection, transparency, documentation and reproducibility.
4. Develop a project-specific responsible AI action plan that defines acceptable, restricted, and prohibited uses of AI and aligns with good scientific practice and relevant institutional, funding, and publishing requirements.

Contents

1. Foundations of responsible AI in research

• How generative AI works
• The limitations of generative AI
• Accuracy, bias, reproducibility, confidentiality, and human responsibility
• Research integrity, ethics, and governance requirementsAI across the research cycle

2. AI across the research cycle

• Responsible use in research design, literature review, data collection, analysis, writing, publication, and peer review
• Questionable practices and risks associated with AI-assisted research
• Verification, transparency, documentation, and disclosure

3. Managing AI-related risks

• Assessing whether an AI use case is appropriate
• Protecting personal, sensitive, confidential, and unpublished information
• Selecting suitable tools and maintaining meaningful human oversight

4. Hands-on application

• Assessing an AI use case from participants’ own research
• Developing practical safeguards and rules for responsible AI use
• Creating an individual responsible AI action plan

Format
Interactive workshop combining input sessions and discussion of realistic research scenarios. Throughout the workshop, participants work with examples of AI use at different stages of the research cycle and assess them in terms of research integrity, ethics, data protection, transparency, reproducibility, and human oversight.

Prerequisites
The workshop is open to researchers with and without prior experience using AI. The relevant technical concepts will be introduced in an accessible way; no programming knowledge is required.

Trainer
André Walter is a lecturer at the University of St. Gallen and a political scientist with a background in comparative politics, political economy, and computational social science. His work combines quantitative social science, historical and contemporary political data, text-as-data methods, and reproducible computational workflows. He has extensive experience with research data management, data documentation, code-based analysis, and the development of reusable research datasets. His work is closely
connected to Open Science principles, particularly transparent research design, reproducibility, and the responsible sharing of data and code.

Sabou Rani Stocker is the Open Science PhD candidate at the newly established Open Science Office at the University of St. Gallen and a psychologist by training. Sabou is currently a research associate and PhD student at the Institute of Behavioral Science and Technology at the University of St. Gallen. She is passionate about open and collaborative research practices.

Target Group
The workshop is open to PhD candidates and postdoctoral researchers at the University of St.Gallen, the Eastern Switzerland University of Applied Sciences (OST), the St.Gallen University of Teacher Education (PHSG), and other member universities of the ENGAGE.EU alliance. Are you from another institution? Please write to fd@unisg.ch to express your interest.

Time and place
Tuesday, 8 December 2026 | 08:00-10:00 (Room 83-1235, House Washington)

Language
English

Register here

Registration is open until 30 November 2026.