Why my robot is still stupid: SCIoI member Vito Mengers at the Science Slam of the Berlin Clusters of Excellence at Berlin Science Week 2025

Rethinking what it means for a robot to act intelligently

Vito’s slam centers on an intriguing question: Maybe robots are stupid because we assume the wrong things about what it means to be smart?

In robotics, intelligence has traditionally been linked to planning, like calculating future states, selecting optimal actions, and executing carefully ordered sequences. Yet when we look at the natural world, many animals that we do not usually label as “intelligent” cope remarkably well with complex, changing environments. By contrast, robots that rely heavily on predefined plans often struggle outside carefully controlled settings.

Rather than trying to make plans more complex, Vito’s research asks whether intelligent behavior might be possible without plans at all. But if intelligence is not primarily about having a plan, then what else might it be about?

©MATH+/Kay Herschelmann

What Vito works on beyond the stage

Behind this performance lies Vito’s everyday research as a roboticist. In his work, he explores how robots can act robustly in uncertain, real-world situations, where information is incomplete, environments change, and small disturbances can disrupt carefully designed behavior.

One of the approaches he works with is a framework called AICON (Active InterCONnect). Instead of instructing a robot with fixed sequences of actions, this method relies on continuous feedback between perception, internal estimates, and movement. The robot does not follow a stored script but rather adjusts its behavior based on whether its current actions improve the situation relative to a goal.

This way of working has been tested in physical tasks like opening a drawer under uncertain conditions. In these scenario, the robot adapts, corrects itself when something goes wrong, and continues even when the environment changes.

Why this matters

Within SCIoI, this perspective is central. Vito’s work shows how goal-directed behavior can be grounded in real-time perception and action, rather than in abstract representations and rigid plans. This connects artificial systems with biological ones, where organisms continuously adjust their behavior based on feedback rather than long-term foresight.

In this sense, his research contributes to a broader scientific question that defines the cluster: how intelligent behavior can emerge from simple principles when agents are embedded in a complex world.

The Science Slam makes it possible to share such ideas beyond academic audiences. In six minutes, stripped of technical detail, what remains is the core question: are we building intelligent machines based on the right assumptions?

As part of Berlin Science Week’s 10th anniversary under the theme “Beyond Now,” SCIoI contributes three events—an exploration of collective intelligence, the inter-cluster Science Slam, and an Excellence Pub Quiz dedicated to the nature of intelligence. Together, they reflect the cluster’s commitment to bringing fundamental research into dialogue with the public.

And while Vito’s performance does not aim to explain his methods in detail, it opens a window onto the kind of thinking that drives his research, and invites a diverse audience to reconsider what intelligence might look like, in machines and beyond.