Skip to main navigation Skip to search Skip to main content

What AI Doesn't Know: Technological Innovation, Generational Knowledge, and Global Development. Lessons from Water Innovation in India and Clean Cooking in Uganda.

Activity: Talk or presentation › Invited talk

Description

Abstract
A growing assumption is that artificial intelligence (AI) can improve access to knowledge and, in turn, support technological innovation. Within global development contexts, however, this assumption is highly problematic. Much of the knowledge that shapes innovation and everyday practices in the Global South is not formally documented or easily available to AI. Rather, this knowledge and practice has developed locally, often over generations, and is embedded within communities and specific social contexts.

This talk presents preliminary findings from ongoing research in India and Uganda which explores the importance of generational knowledge for technology innovation. It also considers what may happen as AI becomes more widespread across the Global South. The first project examines filtering water pots for water treatment and storage in India. The second examines clean cookstoves in Uganda, where the use and adoption of improved practices derives from existing cooking traditions and knowledge developed over many years.

For both projects, we are finding that AI does not significantly improve these processes and technologies. Indeed, in some cases, it can make them worse. This may be because AI systems tend to prioritise knowledge from the Global North, with local and generational knowledge from the Global South being less visible. The resulting recommendations from the AI models can therefore miss important aspects of how these technologies are actually used and understood within particular social and cultural contexts.

The talk considers what this means for the future of ICT4D. In particular, it questions whether increasing reliance on AI may weaken or ignore some of the knowledge systems that global development interventions depend upon. The main argument is not that AI has no role in these settings, but that we need to think much more carefully about whose knowledge is represented, whose is prioritised, and whose may be lost.
Period8 Sept 2026
Held atUniversity of Oslo, Norway
Degree of RecognitionInternational