How could a mechanical architecture, no matter how ingenious, alone solve the inherent complexity of robotic vision systems in the long term? It's a too narrow vision that forgets the structural forces at play over a decade. The secular trajectory of artificial intelligence shows us that it is not the flexibility of an arm that will define sensor integration, but the capacity of algorithms to interpret complex data in real-time, regardless of the shape of the manipulator. For example, future anticipate the movements of an object solely based on understanding its environment, even without touching it, rather than focusing on how to hold it.