en
Enter the Blackbox transfers the technological problem of AI's lack of transparency into the space of the theatre black box: the black box becomes a laboratory in which probability and vector logic, training processes and decision-making architectures are performatively ‘de-blackboxed’. Deep Society focuses on the social consequences of a probability-based order: structural inequalities, bias, exclusions and the accountability gap that arises when automated systems act without being able to take responsibility for their actions. Finally, Artificial Theatre Intelligence attempts to develop, train and test a ‘theatre AI’ as a multi-agent system (AGI) on stage: When text, lighting, sound and stage are controlled by different AIs, theatre itself becomes a testing ground for delegation, control and artistic authorship.
Together, the three projects form a developmental arc from the performative ‘de-blackboxing’ of AI, through the negotiation of its social consequences, to the scenic testing of AI-controlled theatre practice.
Enter the Blackbox is a critical performance and laboratory setup that asks how the black-box effects of contemporary AI systems shape social realities—and why structural opacity enables political, ecological, and psychosocial harm. Drawing on current debates on opacity, power, and authoritarian tendencies (e.g., Rainer Mühlhoff), the project treats AI as a contested, economically driven apparatus that is vulnerable to manipulation and abuse. Where transparency is missing, democratic oversight becomes difficult or impossible; discrimination, environmental damage, and concentration of power remain concealed. This produces a “known-unknown” condition: we sense that knowledge is being produced, yet it remains inaccessible—a diffuse insecurity that can be politically exploited.
The theatrical black box becomes an experimental space for strategies of making-visible. The audience enters a black-box theatre space and, symbolically, the AI black box. In collaboration with set designer Sandra Fox, we build a walkable spatial diagram: tensioned strings draw axes, networks, and nodes; fog and laser lines mark zones, distances, clusters, and “probability clouds.” The evening unfolds as a sequence of experiments (Input → Mapping → Training → Output). The audience contributes keywords and examples around themes such as security, labour, intimacy, resources, and publicness—and in certain passages can become “tokens” themselves, positioning their bodies in space so that proximity and distance turn into meaning relations. From this, an analogue neural network emerges: connections are created and collectively weighted (strengthen, weaken, delete).
Performers and audience participants take on functions familiar from machine-learning training (filtering, annotation, moderation): inputs are sorted, prioritised, “cleaned,” overrepresented, or excluded. In this way, bias, norm-setting, and the power effects of “safety/ethics” moderation become physically legible. The “trained” network produces condensed outputs—spatial arrangements, rules, and scenes—performed live and repeatedly reshaped. The black box is not “solved,” but staged as a contested field of knowledge, control, and responsibility—inviting the audience to reclaim agency through embodied presence.
Manus Nijhoff (1993) is a Dutch media artist and developer based in Berlin. Trained as a graphic designer, he has been working at the intersection of design, code and performative art for years. With the Berlin theatre group Interrobang, he developed interactive, AI-explorative and critical works such as Commune AI, Chatbot Challenge and AI Democracy, in which he was involved both as a developer and on stage. As a technical artist at game studio Moving Castles, he was responsible for This Cursed Machine and rat.fun. His practice combines technical depth with a sense of the comical relationships between robots, flowing text and human experiences.
INTERROBANG
Till Müller-Klug
till@interrobang-performance.com
+49 (0)174 407 3606
Jack Willenbacher
Company Management
jack@interrobang-performance.com
+49 30 3259 4991
HAU HEBBEL AM UFER
Sarah Reimann
s.reimann@hebbel-am-ufer.de