Answers are grounded in this paper and cite their sources.
Analyzes AI research dynamics in collaboration with local data holders, suggesting implications for science policies.
This article analyzes Artificial Intelligence (AI) research conducted at a Mexican university center as a situated dynamic of co-production, enabled by data-based associations between AI specialists and external actors. Based on sixteen semi-structured interviews with researchers affiliated with the center, we identify a distinctive mode of collaboration – data-enabled co-production – in which specialists rely on non-specialists who possess or mediate access to the data required to develop or adapt algorithms. These data holders often act as coauthors, evaluators, and potential beneficiaries of the resulting technologies, while researchers can test their developments in real environments. This dynamic anchors AI research in local contexts and may contribute to reconfiguring the relationships between technoscience and society. However, it rarely generates applicable knowledge despite sustaining academic production, reflecting a regional pattern of science that is applicable but not applied. We argue that this configuration constitutes a form of halfway technoscience, socially embedded yet constrained by academic evaluation systems and weak institutional linkages. Finally, the article outlines directions for more situated and context-sensitive science policies.
No takes yet. Share an insight, caveat, or question.
Sued et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: