The conference was opened by Giuseppe Primiero (Milan), who, in his talk ‘Formalising and Verifying Trustworthiness in AI’, highlighted ways of formally modelling trustworthiness in AI systems and, in particular, of making fairness properties technically verifiable. Katherine Dormandy (Innsbruck) then examined, in her talk titled ‘Digital Whiplash: The Case of Trust’, the dynamics of digital information environments and the question of how repeated shifts between trust and mistrust affect epistemic practices.
Further presentations focused on the ethical and social implications of AI-supported interactions. In ‘In the Loop, out of Sync, analysed changes in moral cognition in human–AI interactions, whilst Christiane Elisabeth Ernst and Kathrin Figl (both from Innsbruck) presented empirical findings on how transparency disclosures influence trust in AI-generated deepfakes in the healthcare sector. Eleonora Catena (Erlangen) discussed the implications of personalised AI assistants for personal autonomy and decision-making.
The conference concluded with a focus on fundamental philosophical questions regarding understanding and attribution. Camilla Quaresimini (Milan) addressed the limits of algorithmic gender classification, before Matteo Baggio (Turin) and Federica I. Malfatti (Innsbruck) explored, in their presentation *Can AI understand?*, , explored whether, and in what sense, understanding can be attributed to AI systems. The discussions made it clear that trust in AI is not merely a technical challenge, but above all an epistemic and ethical one, which can only be adequately addressed through the interplay of philosophical and empirical perspectives.
(Author: Federica I. Malfatti)