On November 2, Julia Stoyanovich led an interactive workshop at Berlin Science Week titled “Reim, Rhythmus und Risiko: Was Sprachmodelle ausmacht” (“Rhyme, Rhythm, and Risk: What Makes Language Models Tick”). The session introduced a broad audience to how large language models (LLMs) work, explaining how these systems learn patterns from large collections of text and where their strengths and limitations lie. Through discussion, examples, and humor, the workshop explored both the capabilities and blind spots of modern AI systems while encouraging participants to think critically about how these technologies shape the way we create, interpret, and share information.

Julia’s contributions to Berlin Science Week also included an article in Tagesspiegel, “Weit mehr als ein bloßes Werkzeug” (Far More Than Just a Tool), available in German and English. In it, she reflects on how AI has quietly become part of our social infrastructure—shapig how we learn, heal, and govern—and reminds us that responsibility rests not in the code itself but with the people and institutions who build and deploy these systems.

Berlin Science Week is an annual international festival that brings together researchers, innovators, policymakers, and the public to explore major scientific developments and their impact on society.