The human-centric approach in scientific research: the AI Act and the new frontiers of research ethics

Chapter 6 analyses how data-driven methods and AI are transforming scientific research and giving rise to new ethical challenges. Researchers now rely heavily on datasets and algorithmic processing, which creates a mediated relationship that can obscure the origins, limitations and societal implications of data. The author critiques current AI ethics approaches for blurring the lines between ethics and law, transferring ethical principles across domains without considering the context, and focusing on high-level values that are difficult to implement in practice, issues that are exacerbated by industry-driven 'ethics washing'. Although the EU AI Act introduces important safeguards, its research exception means that ethical assessment is largely outside its scope. The author argues that this gap requires dedicated methodologies for evaluating the social and ethical impacts of AI-driven research. Drawing on the example of biomedical ethics committees, he proposes the establishment of expert, multidisciplinary and participatory structures to guide the responsible development of AI. Ultimately, the chapter calls for EU-level methodological guidance to support a robust, human-centric research framework.