Although artificial intelligence (AI) has long been part of our everyday lives, it still raises many questions. A research team from the University of Stuttgart now wants to discuss the possibilities and limits of the technology with members of the public.
Flawed digital facial recognition in train stations, automated application selection programs that tend to discriminate: there are various concerns about the use of AI. On the other hand, its use in everyday life has long been a matter of course, for example in Internet searches or email spam filters. A research team consisting of researchers from the University of Stuttgart and the Stuttgart Media University (HdM) now wants to discuss this widespread technology and the associated ideas and expectations with members of the public in the Fragen an Kollegin KI (Questions for Colleague AI) project. This is one of two projects that the German Federal Ministry of Education and Research (BMBF) is funding in the context of the 2022 Year of Science under the motto “In demand!”.
“People feel a strong need to have a say in determining the research agenda,” says Dr. Elke Uhl, Executive Director of the University of Stuttgart’s International Center for Cultural and Technological Studies (IZKT). As Uhl explains, this project is about “the fears, hopes, anxieties, and visions associated with the use of AI in the workplace and in everyday life. “Hence the name colleague AI: will we have to accept its widespread use in the workplace? What will be the consequences?”
Initiating a broad dialog about the use of AI
The project comprises three pillars in order to discuss these issues on a broad basis under the first of which experts will be presenting and discussing their use of AI applications with the public in a series of public discussions at the Stuttgart Public Library. The second pillar will involve a project seminar organized by the HdM and the University of Stuttgart, in which students from the natural sciences, engineering, and media sciences will collaborate to address issues relating to the use of AI. This might include, for example, video or multimedia reports that will be shown during the discussion series.
The aim is for students to learn how to refine their science communication skills in a more dialog-oriented way,” as Dr. Alexander Mäder of the HdM explains. Mäder will be leading the seminar together with Elke Uhl. “What I'm seeing is that there's an increasing interest in engaging with the audience,” says Mäder. Which is why these budding media professionals are not only expected to report on AI, but also to moderate, network, and elaborate on different positions. The project is being supported by the university radio station HORADS and the “future reporters” on the riffreporter.de journalism platform.
Supporting learning processes with new technology AI
may also be used as a teaching format during the seminar: Jun.-Prof. Maria Wirzberger and her team at the University of Stuttgart’s Department of Teaching and Learning with Intelligent Systems (LLiS) of the Institute of Educational Science are studying the development of adaptive educational technologies that provide effective support for the learning process. The smart teaching systems in question are based on AI. “But these are not threatening in any way,” Wirzberger says. “Nevertheless, one needs to put some thought into it: what data do the algorithms collect? What happens to it? And what long-term effects might this have on students?”
“What data do intelligent systems collect? What happens to it? And what long-term effects might this have on students?”Jun.-Prof. Dr. Maria Wirzberger
Opening a window on AI
The third pillar of the project will involve the use of student input on example AI use cases to initiate discussions, to which end the so-called AI Citizen’s Council will be bringing 300 representative people from all over Germany together in a virtual space. A team led by political scientist Prof. André Bächtiger, Executive Director of the Institute for Social Sciences, is hoping that the “Democracy Factory” program will encourage participants to explore AI in depth. All questions raised by the study group are sent in a distilled format to the experts, who then respond to them, which leads to the next discussion round.
“There’s always a risk when developing new technologies that the respective systems will be over and above what is actually needed,” explains psychologist Wirzberger. “The only way to avoid this is to involve the perspectives of future users in the process right from the outset. This is precisely where I see the project's great potential for opening a window on AI.”
Wirzberger is the spokesperson for the Stuttgart Research Focus “Interchange Forum for Reflecting on Intelligent Systems” (SRF IRIS), which was founded in 2021, and the director of the “Reflecting on Intelligent Systems In the Next Generation” (RISING) teaching and learning forum, which includes the seminar in its event calendar.
Researchers from almost all faculties of the University of Stuttgart are represented in the IRIS. The Questions for Colleague AI project was born out of this association and therefore benefits from an ongoing creative exchange of ideas. For example, Bächtiger, another member of the IRIS Board of Directors, is able to find specialists on the topics discussed in the Citizen's Council. “The unique thing,” he says, “is rather than presenting a number of paternalistic experts, who simply dictate to the public, our specialists have to respond to comments from the public.”
“What I'm interested in is how interacting with a virtual host influences the outcome, and how satisfied people are with it.”Prof. Dr. André Bächtiger
Promoting factual discussions
Students in the project seminar will host one of three Citizen's Council groups. Mäder plans to focus on how to promote factual discussions in a series of editorial meetings. “My envisioned goal is to work out the really critical issues after clarifying the purely technical ones,” he says. An artificial moderator named Sophie will host the second group in order to compare AI with this. The third group will serves as a control group and will not be overseen by a specific host. Sophie, the intelligent chatbot, will prompt people to read posts from others or respond to questions, among other things. The benefit of this would be a significant reduction in personnel costs. At the same time as Bächtiger emphasizes: “I want to know how interacting with the virtual host will affect the outcome and how satisfied people are with it.
At the beginning and end of the experiment, the participants in the council will respond in writing to the question: what would you like to see from colleague AI? “We will be able to run computational linguistic text analyses on the responses,” says Bächtiger, who is also scientific director of the University of Stuttgart's Center for Interdisciplinary Risk and Innovation Studies (ZIRIUS), which will evaluate the questionnaires and forward the results to the BMBF. In this way, the wishes of the public will also be taken into account in decisions on research funding. In addition, as Bächtiger says, it will be possible to determine the degree to which people have thought about AI. “If for example, we find that the group has developed a very complex view of the world but has not changed its mind, that would also be a very important finding. Policymakers will be interested to see what conclusions people arrive at when they consider the issues in depth.”
Because as Uhl emphasizes, the purpose of the AI Citizens' Council is not to provoke confrontation, but rather to encourage differentiated points of view. “Our concept is based on the notion of countering polarization and increasing resentment.”
Text: Daniel Völpel