Opening

• Stephanie Schworm (HIDA)
• Lea Maria Ferguson (Helmholtz Association, Helmholtz Open Science Office)

Welcome address

• Karin Lochte (Zentrale Ombudsperson der Helmholtz-Gemeinschaft)

In order to make AI a useful fool for science, all actors involved have to contribute: -political actors to provide appropriate legal frameworks, -research organisations to develop guidelines and train their staff, -researchers to keep up with the latest developments in AI technology, indicate when AI was used in their output and always remain honest in research. But what is good, honest science? These questions are the core of ethics of science and rules to good scientific practice. The emergence of AI presents new challenges to both, partly because completely new fields of violation of good scientific practice emerge and partly because it is not easy to detect any misuse. Therefore, AI guidelines have to be linked to the rules for good scientific practice and for ethics in science. Training of staff and advice on best practices is essential as well as being alert to new developments.

Ensuring Research Integrity in the Age of AI: Recommendations, Trends, and Challenges

• Katrin Frisch (German Research Ombudsman)

With the introduction of free and easy-to-use generative AI applications like ChatGPT, Bard/Gemini or Perplexity, their use among researchers has exploded in the span of less than two years. According to recent studies, between 54% and 76% of researchers have already used AI. Just as swift as the uptake of AI was the emergence of fears and concerns about its effects on matters of integrity, ethics, and honesty within the context of higher education and science. While at first the focus lay on student use of AI, it has increasingly shifted to researchers and the question of what good research practice looks like in the age of AI. In this presentation, I will approach AI from a research integrity perspective and devote myself to the following questions: where does consensus exist? What still needs to be settled? And what are going to be persistent challenges?

Guidelines in the use of Generative AI in Research

• David Arranz (European Commission)

The guidelines focus on the responsible use of artificial intelligence (AI) in research, and has been developed in collaboration with the countries and stakeholders in the European Research Area Forum. These guidelines emphasize the importance of ethical considerations, transparency, and accountability in the use of generative AI in research activities. They flag challenges and provide simple recommendations for ensuring that generative AI tools are used in ways that uphold scientific integrity, safeguard privacy, and promote public trust. The guidelines provide recommendations for researchers, research organisations and funders to foster an environment where AI can contribute positively to scientific progress while mitigating risks associated with bias, misuse, and ethical concerns. The guidelines are designed to fit and align with the broader EU framework on AI, ensuring that Europe remains at the forefront of responsible AI innovation in science.

Helmholtz’s recommendations for the use of AI in research and administration

• Michael Schmuker (Helmholtz Association)

In summer 2024, a committee from across the Helmholtz association has developed recommendations for the use of AI in Helmholtz. This document is intended to provide guidance to users of AI in research and administration. Its main intent is to highlight opportunities and point out risks inherent to the use of AI tools. The current version is the first instance of a “living document”, to be continuously updated to cater to developments in the field. As Germany’s largest research organization, Helmholtz has the ambition to lead the way in the application of AI services for research and administration.

Ethical Implications of Generative AI in Scientific Research

• Theresa Willem (Helmholtz AI - Consultants, Piraud Group)

The adoption of generative AI (genAI) to create scientific content carries significant risks for multiple stakeholders both on the input and the output level. On the input level, the scraping of copyrighted data used for genAI training and the unpaid labor hidden beneath can harm data and intellectual property owners. On the output level, the potential spread of misinformation, fraud, and the reinforcement of systemic biases can have detrimental consequences for individuals and society at large. This impulse highlights these concerns and provides us with ethical considerations to keep in mind for our workshop.

Workshop Session

• Steffi Genderjahn (Helmholtz Open Science Office)
• Stephanie Schworm (HIDA)
• Lea Maria Ferguson (Helmholtz Association, Helmholtz Open Science Office)

This hands-on workshop session is based on the “Living guidelines on the Responsible Use of AI in Research”, using a world-café setting. Three virtual tables will host a deep dive into the guidelines through guided discussions and exchange of experience, capturing different perspectives on the topic. We will use miroboards for documentation.

Joint discussion & Wrap-up

• Stephanie Schworm (HIDA)
• Steffi Genderjahn (Helmholtz Open Science Office)
• Lea Maria Ferguson (Helmholtz Association, Helmholtz Open Science Office)