Following the onset of the Covid-19 pandemic the over-indebted Kenyan state was unable to stabilise a nose-diving economy. Faced with the risk of bankruptcy and widespread defaults on loans, financial credit providers such as banks and microfinance institutions restricted access to credit. The resulting credit-crunch forced many Kenyans to turn to digital lenders and friends or kin for access to credit to pay off pre-existing debts and make ends meet. This collection comprises ethnographic data on how individual borrowers and informal savings-and-credit groups navigated this credit-crunch. In particular, the collection features information on the social consequences of financial technologies and the quantification of creditworthiness through digital data in informal and domestic economies.Artificial Intelligence (AI) is in the ascendant the world over. This is especially the case when it comes to machine learning and big data, which are said to offer a technical fix to human questions of trust. Such rhetoric obscures its embeddedness in a specific socio-cultural context, while downplaying the extent to which trust is an ethical and political issue rather than a strictly technical one. But most social sciences, unlike mathematics and computing, have had little to say about the trust that such technologies are said and designed to foster. My proposed fellowship marks a step towards addressing this imbalance, through research activities as well as by building an interdisciplinary network of social scientists seeking to develop publicly engaged scholarship on AI. More broadly, the fellowship will help consolidate my doctoral research on trust, by bringing it to bear on recent developments in AI, as well as by integrating some additional research into my existing material, with a view of developing a monograph within the next two years. In my thesis, I argued that - in Kenya as elsewhere - popular narratives about trust link up with the historical reproduction and transformation of social and economic inequalities. Having situated local narratives of trust and faith in a history of missionary and colonial projects, I now wish to consolidate my historical analysis with additional archival data and to weave in fresh ethnographic data on digital infrastructures and their consequences for spiritual and political-economic concerns in the everyday lives of ordinary Kenyans. For example, I am especially interested in how global narratives on artificial intelligence and social trust are impacting the social lives of the microfinance groups I worked with in my doctoral research. If, as argued in the thesis, microcredit borrowers draw on a theologically-diverse language of faith to negotiate the terms of trust and cooperation, it is less clear whether this language of faith can also account for how people respond to and make sense of the data infrastructures that increasingly profile them through machine-learning algorithms and credit-scores. Thus, the fellowship asks: 1. How do algorithmic metrics of creditworthiness and trustworthiness affect formal and informal credit arrangements in rural Southwest Kenya? 2. How do people encounter AI algorithms in a context where trust is said to be elusive, declining, fraught by inequality, and often expressed in the language of religious faith? 3. How do religious commitments influence notions of trustworthiness and relations of trust in social, political, and economic life?

The data was collected through semi-structured interviews with 24 respondents. Repeat interviews were arranged at least a month apart, partly to fill in gaps in information but also to get a sense of how respondents’ situations and experiences progressed. Respondents were recruited primarily through snow-ball sampling, with a view to ensure a variety of socio-economic backgrounds were proportionately represented. Research participants thus included young as well as middle-aged men and women, with a third in formal employment and the vast majority self-employed in the informal and gig economy as farmers, merchants, contractors, technicians, motorcycle taxi drivers, shopkeepers or small-business owners. Most participants owned a smartphone or a mobile phone. Interviews were conducted remotely, through a research assistant, due to travel restrictions and the risk of spreading Covid in rural Kenyan communities.