Artificial Intelligence technology is advancing quickly, to the point that its underlying functionality and ‘soft’ impacts are challenging to grasp ex ante. Due to AI’s reliance on being designed by programmers, they can become manifestations of human bias. On the other hand, as AI can hold inherent and latent qualities, it can shape individuals’ perceptions, decisions and behaviors. This effect poses serious threats of distortions from and to society and technology. Therefore, this research project focuses on investigating antecedents, risks, and opportunities of technological mediation of modern AI applications (problem analysis) and developing necessary mitigation strategies (mitigation analysis).
The research team will study the epistemic and normative distortions existing in the reciprocal influence between humans and technology and to what extent these distortions can be attenuated. This project will address these questions from two distinct directions (i.e., human-to-tech and tech-to-human influences) and with a mix of theoretical and empirical methodology. Expected practical outcomes are threefold: (1) several publications providing a holistic overview and empirical analysis of technological mediation; (2) a new and joint student module in order to share research insights and create knowledge transfers in real-time; (3) practical recommendations for assessing and mitigating distortions that can emerge during the design and use of AI. Therefore, the findings will serve and aim to impact the skills and actions of various stakeholders such as academia, policymakers, students, and technology companies.
Formalizing ethical principles within AI systems: experts’ opinions on why (not) and how to do it