Abstract

Advanced computational technologies suffer from a growing catalogue of failure and abuse and heightened concern over their use to to extract and harvest personal data, seemingly with consent, to build systems that commodify, disempower or manipulate ‘users’. Yet are standard large-scale social research instruments such as surveys sufficiently better? In Global South contexts, citizens are too often commodified research subjects, disempowered in the very act of handing over personal information to an opaque research exercise with obscure policy or intervention outcomes.

This seminar explores learnings from Africa’s Voices Foundation, a non-profit start-up spun out combined socio-political and technological research at Cambridge. From Nairobi and Cambridge, AVF works with partners such as UNICEF , Oxfam, the MasterCard Foundation and many others to help them “listen intelligently”, with speed, scale and depth, to the views of the citizens whom they serve. To do this, we design socio-technical systems that subvert the extractive, commodifying and dehumanising logics of mainstream artificial intelligence and research instrument approaches. This requires accepting the methodological and technical burden that comes from engaging with citizens as social agents rather than extracting from subjects as sources of data. To listen_rather_than_sense, we use AI to augment human capabilities for interpreting social evidence rather than to automate data-driven decision-making. AVF ’s learnings shed a light on what technological innovation, driven by socially-conscious applications of AI, might look like if not tied to commodification, power or control, but aimed at humane goals to advance social good.