What’s new in the field of Human-Machine-Interaction? (in german)
Chair: Sebastian Schleidgen (Hagen/Munich)
Gabriele Gramelsberger (Aachen): Machine-Rational Sociality. Critique of the Sociality Inscribed in Machine Learning Procedures
Machine learning, unlike other algorithmic structures, is characterized by the fact that a specific sociality is inscribed in them. This reduced form of sociality is already evident in Alan Turing’s 1950 essay Computing Machinery and Intelligence, where Turing postulates that machines can have intelligence even though the Turing tables of human machines are unknown.
Turing’s postulate is based on the reverse conclusion that the statement “being regulated by laws of behavior implies being some sort of machine” also applies to the intentional motivation of humans. Turing elaborates on this by claiming that the intentional motivation of rule selection and deployment can also be understood as learning, primarily involving the exteriorization of learning mechanisms. He has in mind the classical behaviorist learning strategies in the form of negative and positive reinforcement, which he claims for learning in “child machines.” The contribution examines this line of development, which is used today in so-called reinforcement learning and which, due to its machine-rational sociality, enables ML algorithms to train themselves.
Manfred Hild (Berlin): On haptic interaction with self-referential machines
If one breaks with the generally prevailing view that autonomous robots must be equipped with separate sensors and actuators, all connected to a central digital processor, and instead uses special local sensorimotor loops, so-called “cognitive sensorimotor loops” (CSLs), then surprisingly rich and divers patterns of movement suddenly emerge. These unfold through the close coupling between body and environment and, in particular, through haptic interaction with humans.
As it turns out, CSLs do not require a digital processor, but can be realised through simple analogue circuits. This naturally opens up the possibility of using novel synapse-like components (so-called memristors) – both for implementing adaptive processes (movement optimisation) and for storing implicit models during the body’s self-exploration.
Christopher Coenen (Karlsruhe): The Digitalisation of Taylorism. From Alterity to Embodiment and Back Again?
The contribution is an attempt to use elements of Don Ihde’s post-phenomenological approach for a comparative analysis of historical and digital Taylorisms. The assumption is that using his approach for this purpose requires its broader extension to collective action and interpersonal relations.