How to make it fAIr - Participatory design methods for the field of artificial intelligence (AI)
"What is fair AI?" and "How can fair AI be implemented in practice?" these and other relevant questions about human-centered AI research were the focus of an online event on May 15 2021. At the event, organized by AIT and the University of Salzburg, four FFG-funded projects and their key findings were presented and reflected upon in a discussion round with experts. Anita Thaler (IFZ) presented project results from the dAIalog.at project led by IFZ and referred in particular to the importance of involving stakeholders in the development process at an early stage (e.g. works council members and gender- and inclusion-representatives). How this can be done with concrete participatory formats and at which points fairness in AI model development, AI training, AI implementation and data storage has to be taken into account, can be read in the report available for download at www.dAIalog.at (open access).
In the following discussion, project member Christopher Frauenberger (University of Salzburg) and other discussants reflected on questions of technology philosophy and ethics, e.g. to what extent AI really has to be 'human-centered' and whether 'fair' or perhaps 'just' is the better term, because this has to be negotiated again and again between the participants.