The user interface (UI) is currently undergoing a significant transformation. In the course of machine learning (superordinate: artificial intelligence) and under the influence of universal interconnection of digital devices, the way in which technical functions of various applications are used is changing.

The theoretical part of the dissertation shows how the classic user interface, which Peter Sloterdijk once understood as the ‘make-up of machines’, must be called into question. Sloterdijk describes how over-complex functions are reduced by design until the user can be suggested to have their own sovereignty (Peter Sloterdijk, Der Welt über die Straße helfen). In the meantime, however, technical development has moved beyond the point where user input is still necessary at all, however much the interaction surfaces owe themselves to sophisticated simplification. Where the ISO standard assigns the UI the role of providing ‘information’ and ‘control’, the dissertation distinguishes between five historical perspectives on the user interface and explains how the current ‘anticipatory interaction paradigm’ can be viewed from two semantic perspectives.

Fundamentally, user interfaces no longer function as input masks that are used to make entries line by line, gesture by gesture, input by input. Learning applications, which sometimes anticipate inputs as ‘technoid counterparts’, set chains of action in motion and anticipate the decisions of their human counterparts. The work of designers, who have so far been relieved of training tasks in dealing with technical devices through semantic references, is thus inevitably put to the test.

Essentially, user interfaces no longer function as input masks, with the help of which inputs are made line by line, gesture by gesture, input by input. Learning applications, which sometimes anticipate inputs as ‘technoid counterparts’, set chains of action in motion and anticipate the decisions of their human counterparts. The work of designers, which has so far relieved people of learning tasks in dealing with technical devices through semantic references, is thus inevitably put to the test. It is shown that new design tasks lie less in the shaping of interfaces that structure access to the ping-pong game as a sequence of user input, device processing and subsequent display. Rather, the task is to design the character and framework conditions through which learning applications reveal themselves in use - that is, in interaction with people. This means that the user interface can no longer be designed sequentially. Human and machine learning are mutually dependent and permeate each other. Examples from everyday life, images from science fiction and stories of artificial-human counterparts throughout the course of human history are used to redefine the requirements for human-object interaction in this sense.

It is shown that where machine learning is used in industrial design products, humans are not automatically freed from all learning tasks. Learning still takes place on the human side. Perhaps in future, design will have to ensure less rapid accessibility and more careful comprehensibility through precisely designed stumbling blocks - and at the same time help shape the framework conditions for machine learning. In any case, design is an extremely demanding activity, and design education is once again a highly political matter.

The practical part of the dissertation is of a cumulative nature and comprises the collection and discussion of short design projects that were created during the course of the doctorate.  In three project formats, aesthetic requirements, areas of application of algorithms and machine learning tools as well as exercises and reflection work for university teaching are presented and discussed. The practical work is to be understood partly as an implementation of the theoretical findings and partly as a demonstration of the conclusions drawn in the theoretical work from a practical perspective. Simultaneously, it can be regarded as stand-alone work, too. The various projects are designed to form the basis for a reorientation of university teaching in the context of current, accessible “AI tools” and applications.

Supervisors:

Prof. Dr. Martin Gessmann

Prof. Dr. Klaus Klemp

Prof. Frank G. Zebner

Pia Scharf is a lecturer in design history and design theory at the FHNW Academy of Art and Design in Basel (Switzerland) since 2022 and supervises bachelor's and master's theses in industrial design. Previously, she was a substitute professor of design history at the Offenbach University of Art and Design (hfg) from 2021 to 2022. In April 2025, she submitted her dissertation on machine learning (AI) and design; the defense took place in November 2025.

Pia Scharf studied industrial design at hfg and completed an exchange semester at Tongji University in Shanghai, China. After graduating in 2016, she worked for several years as a research assistant to Prof. Dr. Klaus Klemp at the Chair of Design Theory and Design History at hfg, where she gained extensive experience in design research and teaching. She has held teaching positions in theory and practice at several German art and design colleges, including Mainz University of Applied Sciences (communication design), Saar College of Fine Arts (product design) and hfg (industrial design).

Pia Scharf is a founding member of the association design inclusion e. V. and a member of the Gesellschaft für Designgeschichte e. V. and the ernst-may-gesellschaft e. V.