ESDST’2026

Résumé

Innovation ecosystems in emerging regions often mirror and reinforce existing socio-spatial inequalities, particularly between well-resourced urban centers and marginalized inland territories \citep{unesco2021, oecd2019}. In Tunisia, such disparities limit equitable access to technological infrastructure, data, and innovation opportunities. This paper investigates how AI-driven, place-based hackathons can serve as instruments for advancing Diversity, Equity, and Inclusion (DEI) through the case of H12 Innovation, a multidisciplinary initiative implemented across three distinct regional contexts.

The study analyzes three editions of H12 Innovation: (i) Sfax, addressing urban waste management challenges, (ii) Kasserine, focusing on the underdevelopment of cultural heritage and structural regional inequality, and (iii) Gabès, tackling industrial pollution and associated public health risks. These cases illustrate how locally grounded problem framing can foster more inclusive and context-aware innovation processes, aligning with the capability approach to development \citep{sen1999}.

Methodologically, the initiative adopts a multimodal data approach, integrating environmental imagery, structured datasets, and socio-economic context. Participants leveraged Artificial Intelligence techniques including Machine Learning, computer vision, and Generative AI—particularly Large Language Models \citep{bommasani2021}—to ideate and prototype solutions. This hybrid approach supports both technical innovation and accessible knowledge production.

From a DEI perspective, three dimensions emerge. First, diversity is reflected in interdisciplinary collaboration, bringing together participants from engineering, artificial intelligence, architecture, and social sciences, thereby enhancing cognitive diversity \citep{page2007}. Second, equity is addressed through the decentralization of innovation activities toward underrepresented inland regions, contributing to more balanced territorial development \citep{oecd2019}. Third, inclusion is promoted by lowering barriers to participation via accessible AI tools and collaborative formats, supporting broader engagement in line with Responsible AI principles \citep{floridi2018, unesco2021}.

The findings demonstrate that AI, when embedded in participatory and context-sensitive frameworks, can act as a catalyst for inclusive innovation. Hackathons function as experimental socio-technical ecosystems bridging technological, disciplinary, and territorial divides. Their impact is multifaceted: social, by empowering local actors; environmental, through data-driven solutions; and educational, by democratizing AI literacy.

This study contributes to the literature on inclusive digital transformation by providing empirical evidence that localized, AI-enabled innovation models can operationalize DEI principles in emerging contexts, offering a scalable pathway toward equitable innovation ecosystems.

\end{abstract}