InterSpaces

Random Matrix Theory (RMT) stands out as a unifying framework at the core of seemingly unrelated fields ranging from signal processing, theoretical ecology, economics and finance, to the theory of learning and the statistical physics of disordered classical and quantum systems. Far from being a mere set of computational tools, RMT provides deep theoretical insights and a foundational understanding of various complex systems. It is key to understand the stability or fragility of complex systems with numerous interacting agents, providing a theoretical framework to capture their responses to perturbations, that may lead to significant changes in the system’s configuration with substantial rearrangements in the state of the individual components (thus describing scenarios of crises or extinctions).


In this episode physicist and capital fund manager Jean-Philippe Bouchaud Paris), quantum physicist Valentina Ros and Tobias Galla from the Institute for Cross-disciplinary Physics and Complex Systems (IFISC) at Palma de Mallorca talk about their research group at ZiF and discuss the potential of Random Matrix Theory and the challenges of interdisciplinary collaboration. 


 


Links Jean-Philippe Bouchaud 


Valentina Ros


Tobias Galla Homepage of the research group at ZiF 

Since September 2025 the long-term research group “Robust Decision Heuristics for Natural and Artificial Intelligence” is working at ZiF. The group seeks to bridge the study of decision-making under uncertainty in natural systems and the design of robust AI systems. Here, "natural systems" is primarily a reference to humans but we will also consider insects because they have simpler decision processes. The goal of the researchers more specifically is to explore the design of algorithms for robust and transparent AI systems by drawing on models of decision processes under uncertainty in natural systems; and to explore the design of AI systems driven by decision heuristics to gain new understanding about decision-making in natural systems.


In this episode the conveners of the group, Mohan Sridharan (Chair in Robot Systems, Edinburgh), Özgür Şimşek (Professor of Computer Science, Bath) and psychologist Gerd Gigerenzer (Berlin) talk about the many facets of their research, covering human decision making as well as robot soccer. They also discuss the necessity and the challenges of interdisciplinary collaboration. 


 


Links Mohan Sridharan 


Özgür Şimşek 


Gerd Gigerenzer ZiF research group “Robust Decision Heuristics for Natural and Artificial Intelligence” 

In der Literaturgeschichte hat die Einfühlung in Täter:innen mit den Mitteln der Kunst eine lange Tradition. Dies bedeutet jedoch nicht, dass die literarische Auseinandersetzung mit Täter:innen zwingend zur Anerkennung von Unrecht, Entschuldigung, Entlastung oder gar zu einer Täter-Opfer-Umkehr führen muss. Gerade Marcel Beyers Roman Flughunde über die Familie des NS-Propagandaministers Goebbels und Nora Bossongs Reichskanzlerplatz, der Magda Goebbels ins Zentrum stellt, verdeutlichen, dass die Annäherung an Täter:innen in der Literatur auf eine Weise möglich ist, die eine komplexere, über Einzeltäter:innen hinausgehende Perspektive auf Schuld und Verantwortung ermöglicht. Aber warum erzählen wir uns überhaupt Geschichten über Täter:innen, welche soziale und moralische Funktion erfüllen sie? Was ist die ‚Aufgabe‘ der Literatur in diesem Zusammenhang? Und wie weit reicht ihre Autonomie? 


Der Literaturwissenschaftler Sebastian Schönbeck moderiert die Lesung der beiden Autor:innen und führt das anschließende Gespräch mit ihnen. Die Lesung war Teil des Workshops "A Case for Empathy? Facing the Perpetrator in Literature, Culture, and History", den die Literaturwissenschaftlerin Saskia Fischer (Hannover) am ZiF veranstaltet hat.


Links: Marcel Beyer 


Nora Bossong


Sebastian Schönbeck Saskia Fischer  Workshop "A Case for Empathy?"


 


Literatur: Nora Bossong: Reichskanzlerplatz, Berlin 2024 


Marcel Beyer, Flughunde, Frankfurt 1996

With the increasing adoption of artificial intelligence (AI) methods in society, discussions about associated risks and how to mitigate them have also intensified, as evidenced by efforts to regulate AI applications in the EU and industry attempts to provide “safe AI”. However, while there appears to be a general consensus that AI-supported systems ought to be safe, we also observe fierce disagreement about which risks should be considered and what constitutes viable mitigation strategies. In other words: there is an urgent need to understand what both risk and safety mean in the context of AI.


In this episode cognitive scientist Benjamin Paaßen (Bielefeld), computational neuroscientist Elizabeth Herbert (München) and Olle Häggström, Professor of Mathematical Sciences at Gothenburg University, talk about the difficulties of defining risk and safety, about the range of risks that should be considered, as well as strategies to enhance AI-safety across technical, social and legal dimensions. They also reflect on the necessity and the challenges of interdisciplinary collaboration to make progress with this endeavor. Links: Olle Häggström Elizabeth Herbert Benjamin Paaßen  ZiF-Workshop "AI Risk and Safety Landscape"