3 / 2024 // LEOPOLDINA / NEWS 9
NEW MEMBERS CLASS III – MEDICINE
Leopoldina President Gerald Haug ( centre ) with Members of Class III – Medicine ( from left to right ) elected in 2023 : Triantafyllos Chavakis ML , Pauline Wimberger ML , Wolfgang Weber ML , Karin Michels ML , Leticia Quintanilla‐Martínez de Fend ML , Wolfgang Henrich ML , Jan-Henning Klusmann ML , Peter Dayan ML and Michael Roden ML ( elected 2022 ).
Image : Markus Scholz | Leopoldina
Generative AI : Uncovering ambivalences
Discussion paper published on “ Generative AI – Beyond Euphoria and Simple Solutions ”
Since the release of
ChatGPT at the end of 2022 , generative artificial intelligence ( AI ) has been drastically changing vast areas of our lives and work . Numerous further developments and AIbased applications now exist that learn patterns based on training data and use them to create new linguistic or visual products . Many of these applications are now generally accessible , easy to use , and highly versatile .
In the recently published discussion paper “ Generative AI – Beyond Euphoria and Simple Solutions ”, ethicist Judith Simon , lawyer Indra Spiecker and computer scientist Ulrike von Luxburg ML take a differentiated look at the opportunities and threats for individuals , the economy , democracy and society that arise from the use of generative AI . In doing so , they counter both utopian promises of salvation and dystopian warnings that dominate the current debate .
The discussion paper outlines problem areas that are currently not sufficiently reflected in the general discussion , including the possibilities and limits of explainable AI , power asymmetries , deception about the performance of AI , diffusion of responsibility and control deficits . These are partly inherent in the functional logic of AI technology itself , but also result from the interaction between humans and technology in different contexts .
In the authors ‘ view , the European AI Act and other regulations such as the General Data Protection Regulation ( GDPR ) do not yet address the possible dangers posed by AI , or only do so insufficiently . Thus , it seems necessary to disclose these aspects of generative AI with all their ambivalence in order to make them accessible to critical public and political debate .
■ CWI
Image : AI | Leopoldina
Discussion paper “ Generative AI ” ( German only )