The backroads of AI: the uneven geographies of artificial intelligence and development

Artificial intelligence is being reconfigured as a set of technologies that can address poverty with precision. The impacts of AI will both exacerbate and ameliorate the conditions of uneven development. Recent debates focus on the disruptive effects of AI, particularly to replication of development trajectories that have had success in reducing poverty. In this paper we further these debates by analysing the backroads of AI. The backroads serve as a metaphor for understanding the ways AI will travel from the sites of incubation to the frontlines of uneven development. We explore dialogues between AI and development in two arenas: the World Bank's Development Impact Evaluation initiative (DIME) and the Government of India's national AI strategy, #AIforAll. We argue that both these arenas serve as hubs from which AI will travel out along the backroads to remote, poor, and fragmented polities. While the World Bank utilises AI as technology to empower experts and mobilise a techno-political authority, what we refer to as precision AI, India seeks to function as a second-tier AI hub, making AI cheaper and more accessible domestically and for other developing countries, what we refer to as populist AI. We conclude by discussing the interrelations of precision and populist AI along the backroads, and the potential of backroads research for mapping AI, uneven geographies of development and technology and the impacts of AI's disruptions at different scales.