On February 27, 2025, the Leipzig University hosted its bi-monthly KI-Netzwerktreffen at our Living Lab in Leipzig. The KI-Netzwerktreffen is a hybrid colloquium bringing together educators, researchers, and students interested in the integration of AI in teaching. Organized by the university’s E-Learning team, this event offered participants the opportunity to learn about innovative AI applications in education, network with professionals in the field, and discuss support needs and questions.

The event began with a warm welcome from Doreen Klein (E-Learning, Leipzig University) and an opportunity for participants to get to know each other. This networking session set the stage for meaningful connections and collaborations.

Introducing ScaDS.AI Dresden/Leipzig

Matthias Täschner and Vanessa Kuhfs took the opportunity to introduce ScaDS.AI Dresden/Leipzig to the attendees, showcasing our mission, vision, and the innovative projects we’re working on. It was a proud moment to highlight our contributions to the community.

The Magic Mirror: A Glimpse into the Future

For the first time, the KI-Netzwerktreffen took place at the Living Lab in Leipzig, posing the opportunity for a hands-on session on the Magic Mirror Demonstrator. It is an interactive system that allows users to provide spoken prompts, such as “show me in a different outfit” or “apply a virtual makeover” and instantly receive a digital projected image in accordance with the request.

The interaction with the Magic Mirror allows users to experiment with various looks and styles in real time. A dynamic and engaging experience. The discussion delved into how we communicate with AI and how such interactions can influence our self-image. There were concerns expressed that AI reproduces and displays stereotypes and the beauty ideal it has learned from its training data. This beauty ideal is often not healthy or inclusive, promoting narrow standards that can negatively impact users’ self-esteem and body image. Critics argue that by reinforcing such ideals, AI may contribute to unrealistic body expectations and perpetuate harmful stereotypes.

Despite these criticisms, AI applications can be a trigger for discussions about beauty ideals, stereotypes, and images. By allowing users to interact with their own body display using their own voice, the Magic Mirror – as an example of such AI applications – highlights the users internalized beauty standards, but also provides an opportunity to break these standards up.

Generative AI: Simplified

Later, Matthias Täschner provided a simplified explanation of how generative AI works, breaking it down into key steps:

• input
• tokenization
• embeddings (semantics)
• token positions

This helped demystify the technology for the attendees.

Who Designs for Whom in AI: Exploring Power Dynamics, Feminism, and Ethics

The presentation “Was KI mit Machtverhältnissen, Feminismus und uns allen zu tun hat” (What AI has to do with Power Dynamics, Feminism, and All of Us) by Vanessa Kuhfs explored the critical question of who designs AI and for whom it is designed. Highlighting the power dynamics and biases that are embedded in AI systems, which often don’t reflect the whole of society. This can result in the reinforcement of existing inequalities and the marginalization of underrepresented communities.

The talk emphasized the importance of diverse representation in AI development to ensure that the technology is inclusive and equitable. By critically examining the training data and the worldviews it perpetuates, we can work towards creating AI systems that are fairer and more just. The speakers called for a greater focus on ethical AI development, which takes into account the perspectives and needs of all users, rather than just a privileged few.

AI as a Mirror of Society?

The conclusion was a holistic view of AI as a reflection of society. It’s essential to consider the ethical implications and strive for a cycle of non-discrimination in AI development.

After the formal event was over, people gathered to stay a bit longer since the need for discussions was great. The lively conversations continued, reflecting the enthusiasm and curiosity sparked by the presentations. The success of the KI-Netzwerktreffen lies in its ability to create a community of practice, where members regularly connect to discuss digital teaching methods. With a blend of in-person and virtual participation, it ensures that everyone can contribute, regardless of their location.

Preview

Due to the positive feedback and high demand, there will be a follow-up KI-Netzwerktreffen on April 30, 2025. This event will focus on the human element behind AI, delving into the often unseen labor that drives AI systems. The working title for the upcoming event is: “Click, Label, Repeat – The Invisible Work Behind Artificial Intelligence.”

For more information and to access past event results and presentation slides, visit the KI-Netzwerktreffen TaskCards.