Post by Karya

Across geographies, a similar challenge is becoming visible. AI systems scale quickly. Context does not. Over the past year, Karya has been working with the Mastercard Center for Inclusive Growth to co-create a digital work global toolkit for distributed data work across the Global South. The focus is not only on scale, but on ensuring that data collection remains grounded in context, language, and lived experience. This toolkit is now being implemented in Kenya and Indonesia with our regional partners, with pilots scaling to 1000+ contributors by mid-2026. The work spans regional languages, dialects, and code-mixed varieties, including Maasai, Bahasa Indonesia, and Sheng. Communities lead evaluations across domains, grounded in speech and language data collection. What is emerging is a model where participation and data quality are closely linked. Systems become more representative when the people shaping them reflect the environments in which they are deployed. Grateful to the Mastercard Center for Inclusive Growth for a partnership grounded in a shared approach to building inclusive digital systems.

Video Content