IRIS Research

“Reflecting Intelligent Systems for Diversity, Demography, and Democracy (IRIS3D)”, started January 2023 with three new Research Groups funded by the Ministry of Science, Research and the Arts of the State of Baden-Württemberg (Ministerium für Wissenschaft, Forschung und Kunst Baden-Württemberg)

• Computational Digital Psychology - Group Sindermann (IRIS3D)
• Diversity-Aware NLP Intelligent Systems (DANIS) - Group Falenska (IRIS3D)
• Ethics of Generative AI - Group Hagendorff (IRIS3D)

Applicants: Prof. Dr. Steffen Staab, Jun.-Prof. Dr. Maria Wirzberger

Duration: 04/2022 - 12/2026

Funding: Ministry of Science, Research and the Arts Baden-Wuerttemberg (funding scheme "Artificial Intelligence & Society")

Report by the University: 3,400,00.00 €

In search of explainable and interpretable machine learning with philosophy and physics

Machine learning (ML) algorithms are permeating our everyday and public lives with increasing intensity.  They make predictions, but why they ‘decide’ one way and not another often remains unintelligible to us:  in a sense they are “opaque”. In our project, we want to understand how this opacity arises, and whether or how it could be retroactively reversed. To do this, we want to interpret the nature of the (implicit) abstractions that ML inherently generates, using insights from physics and other theories of complexity. Our working hypothesis is that the complexity of ML and the difficulty of understanding certain components of the learning process together give rise to this opacity problem. In this sense, a solution does not call for simply “more understanding”, or “less complexity”, but for a sensible reduction of complexity. By this we mean abstractions that are adequate and simplifications that are non-trivial, to ensure access to well-grounded understanding. In our project, we will develop tools to analyze the complexity of ML algorithms in new ways and find reductions that make sense from the perspectives of many-body physics and philosophy.

Funding: Heidelberg Academy of Science and the Humanities

IRIS Member Involved: Dr. Miriam Klopotek

More Information: Project descritption

Funding: European Research Council (ERC Advanced Grant)

Applicants: Prof. Dr. André Bächtiger

Duration: 04/2022 - 03/2027

More Information: Project description

Institute of Aircraft Systems

Flying has been one of the safest modes of transportation for decades due to high-quality standards throughout the development, testing, operation, maintenance, and inspection of aircraft as well as the comprehensive training of pilots. Aircraft operations, particularly those involving passengers, constitute highly complex safety critical situations and, thus, put substantial mental demands on pilots. Therefore, pilots usually receive support from a co-pilot who can take over control in potentially critical situations. With technological developments advancing, AI systems could enable single-pilot operations, where one pilot is in command of the entire flight operation. The system could thereby serve as digital co-pilot – provided a safe and reliable interaction between AI and human can be established. As a crucial prerequisite for such collaboration, the system needs to provide information in a way that the human pilot can easily understand and interpret to act appropriately, for instance, by taking over control. Likewise, the system needs to be equipped with the ability to understand and interpret human actions correctly to infer when to act and provide support or completely take over control in critical situations. To address this challenge, we provide insights into ongoing research on requirements for ensuring safe operations when human and AI collaborate in a cockpit setting. By conducting controlled behavioral experiments in a flight simulator, we evaluate the scope of information exchange on safety. Our results can inform strategies for facilitative AI-human collaboration and simultaneously provide critical food for thought on arising challenges.

IRIS Member Involved:

• Jun.-Prof. Dr. rer. nat. Maria Wirzberger, Teaching and Learning with Intelligent Systems
• Prof. Dr.-Ing. Zamira Daw, Institute of Aircraft Systems
• Patrick Lorrig, Institute of Aircraft Systems
• Alina Schmit-Hübsch, Teaching and Learning with Intelligent Systems
• Nadine Koch, Institute of Software Engineering
• Michelle Fini, Institute of Aircraft Systems

Student Thesis:

• David Tsoi: Development of a Test Environment for Evaluating AI-Human Collaboration in the Microsoft Flight Simulator
• Lennart Lux: Pilot Study for Trust in Human-AI-Teaming
• Michelle Fini: Trust in Human-AI Teaming: Analyzing the Interdependencies of Team Performance, Situation Awareness, Workload and Trust.

Publications:

• Lorrig, P., Tsoi, D. A., Wirzberger, M., & Daw, Z. (2024). AI-Human collaboration in the cockpit: Towards safe single-pilot operations by explainable information exchange. In 53^rd Congress of the German Society for Psychology / 15^th Congress of the Austrian Psychological Society.
• Lorrig, P., & Zaw, D. (2024). Advances and Challenges Towards Enabling Human-AI-Teaming Applications for Flight Deck Operations. In Digital Avionic Systems Conference 2024.

Institute for Natural Language Processing

This research focuses on Natural Language Processing applications to support (and understand) deliberative processes (from political discourse to e-deliberation) and the Interpretability (and cognitive plausibility) of Natural Language Processing models.

IRIS Members Involved: Dr. Gabriella Lapesa

More Information: Powering up E-DELIBeration: towards AI-supported moderation

Institute for Modelling and Simulation of Biomechanical Systems

This research focuses on human physical interaction with the environment using digital human body models.

Research Questions:

• Do these models allow for unbiased and ideology-free decision-making?
• Do these models enable improved human-centered and barrier-free design of the built environment?

IRIS Members Involved: Prof. Dr. Syn Schmitt

More Information: Research in Computational Biophysics and Biorobotics

Funding: Volkswagen Foundation (funding scheme "Artificial Intelligence - Its Impact on Tomorrow's Society")

Applicants (related to IRIS): Prof. Dr. Jonas Kuhn

Duration: 10/2021 - 09/2025

More Information: Project description

Department of Teaching and Learning with Intelligent Systems

The equal professional participation of people with special needs (e.g. people on the autism spectrum or attention deficit disorder) enriches collaboration through diverse perspectives and strengths. However, difficulties in social interaction often lead to the failure of professional life plans despite high cognitive abilities. Innovative technology-supported training measures can take a preventive approach at this point and strengthen interpersonal skills through learning experiences in a protected setting.

IRIS Members Involved: Jun.-Prof. Dr. Maria Wirzberger

Duration: 09/2021 - 08/2024

Funding: The project is funded by the Federal Ministry of Education and Research within the funding scheme "Interactive systems in virtual and real environments - Innovative technologies for the digital society".

More Information: UFO Project