EAI addresses the ethical and epistemic challenges that limit the reliability of artificial intelligence. Central to our work are three obstacles: the reproducibility crisis, which undermines confidence in models and results; the intrinsic methodological limits of current AI, which constrain interpretability and robustness; and the fragmented dissemination of knowledge, which hampers collective progress and proper use across domains. We develop participatory and human-centered design frameworks that acknowledge these limitations, embedding fairness, accountability, and transparency into AI systems from the ground up. In parallel, we study the impact of emerging regulations such as the EU AI Act, building compliance tools that translate legal requirements into actionable practices for developers and institutions. By confronting these scientific and societal obstacles directly, EAI promotes more trustworthy, reproducible, and responsible AI, fostering both innovation and long-term public trust.

Andrea Ottolia, Andrea Vian