�����Ƶ

Finland’s pioneering “one-stop shop” health data sharing model could fast-track AI innovation and enhance patient care. However, for many emerging economies, overcoming regulatory, structural, and trust barriers remains a significant challenge.

Risk prediction algorithms have become central to decision-making, despite known biases. Mitigation techniques too often focus on being statistically “fair,” overlooking the broader societal disparities these biases reflect. We need to expand the conversation to include the structural roots of bias for more transformative interventions.

The rise of artificial intelligence is transforming public administration in Québec, particularly in the management of social benefits. The dematerialisation of social assistance files promises efficiency and agility, but also poses major challenges: the dehumanisation of service provision, reduced discretion for agents and the embodiment of a new form of algorithmic governance.

Equitable and accessible health data is essential for global health innovation. A collective approach and global frameworks would help share the benefits and unlock the exponential value-creation opportunities of artificial intelligence for health care.

The static nature of algorithmic impact assessments makes them ill-equipped to fully protect against embedded gender biases. A holistic approach that includes feminist methodologies and perspectives, like the “woman question”, is necessary to preemptively tackle these issues.

Current AI methods lack an understanding of legal decision-making. Legal meanings are often lost and expertise is at risk. We need new design frameworks to reintroduce lost meanings, prevent expertise loss, and regulate AI system design effectively.

Avec le Règlement européen sur l’IA, la norme de conformité CE est passée d’un outil de garantie de sécurité à un véhicule de protection des droits fondamentaux. Néanmoins, la logique du standard technique présente de nombreuses limites pour assurer une protection effective des droits.

Automated tools for rental practices are promising convenience, objectivity and fairness. But can they avoid reproducing the deep structural inequity underpinning housing markets? In Australia and elsewhere, policymakers need to address the underlying political, social, and material conditions of these seemingly ‘objective’ automated tools.

The deployment of generative AI systems is raising fears of an increase in disinformation campaigns, particularly in the context of elections. The EU’s Digital Services Act could help limit some of these practices.

A universal “one-size-fits-all” AI governance approach will do more harm than good. The unique socio-economic, political, and cultural realities of the Global North and Global South call for a balanced approach that promotes innovation while ensuring ethical and responsible AI deployment tailored to regional needs.