Towards Inclusive and Sustainable Artificial Intelligence: Case Studies on User-Centric Business Models for Development in Low-Income Countries
While the use of artificial intelligence (AI) for sustainable development is growing at a rapid pace, recent studies show that the way in which AI systems are designed and deployed poses threats to sustainable AI. The current dynamics of this industry, driven by the mantra ‘the more data the better’, dictate the extraction of data from human lives. Consequently, power asymmetries arise or amplify between those who possess and control AI systems and those from whom the data are extracted (‘data subjects’). These mechanisms pose significant threats, particularly when manifested in the context of low-income countries (LICs). The literature suggests that there is a need for research into a different possible paradigm in the data business, as well as for approaches that position the data subjects at the heart of the company.
This research explores the approaches of businesses that position those who serve as data sources as a central stakeholder of their business: ‘user-centric businesses’. Case studies were performed to examine the potential of these approaches to address the concerns of the unfolding dynamics of the data industry for development in LICs. Informed by business model theories, a framework was designed to systematically capture the contexts and approaches of user-centric businesses, and to guide interviews with founders and employees. Within-and cross-case analyses revealed three themes: engagement (direct social value return), participation (indirect social value return), and self-reflection across the business model approaches.
Although direct social value returns – such as using data primarily for understanding people’s contexts and views, providing possession or control over data, and non-identifiability – hold potential to contradict the extractive logic of data markets, the data suggest that a tension persists between delivering services to customers and returning social value to the data subjects. Moreover, the perception of individuals as to whether social value returns truly compensate for extraction remains unclear. Furthermore, this study reveals that forms of ownership over data and participation practices can provide opportunities for agency and empowerment to data subjects, but challenges remain in building robust and reliable systems for offering possession and privacy to individuals. Recommendations are made to researchers, practitioners and policy makers for driving change towards more inclusive and sustainable AI.