Participation and Practices in Open Artificial Intelligence: A study of the Sociotechnical Networks of Open-Source Software Sharing and Development

My research concerns the social dynamics of openness in artificial intelligence (AI) research and development, with a particular focus on the social and technical networks of open source software sharing and development. My research will blend social theory with a suite of data science methods to interrogate uneven participation and influence in this community of practice, operationalised through three separate but interrelated studies on this subject matter. The first study will examine the social and technical network structure of open source AI software development on the software hosting platform GitHub, due to its popularity and data availability via its Public API. Upon the collection and preparation of data recording ties between developers and software projects, I will use network science techniques to examine the network structure of collaboration in AI software projects as well as who and whose software have emerged as central authorities in this community. The study will contribute to academic scholarship of social networks in science and technology fields as well as practical outputs oriented at raising awareness of uneven participation and influence in the field of AI.

The second study will examine geographic contours and concentrations of open source AI software development and the factors that can explain global disparities. Similarly, I will programmatically collect geo-coded data on software development activities and enumerate these activities on a country-basis. I will use cartographic analysis to map global activity and use multivariate regression analysis to analyse what factors can explain disparities between countries. This analysis will inform policy recommendations targeted at national governments and international organisations on how to broaden global participation in the AI community. The third study will zoom into a handful of practitioners who in semi-structured interviews will be asked whose values and interests do they perceive to influence norms, dependencies, and possible futures in the open source AI software community. Using purposive sampling methods, I will conduct interviews with users identified in the first study. This study will contribute to academic theory on the imaginaries of members of this scientific community of practice. Overall, this project will contribute to work in the academy on inequality in software development as well as inform policymaking aimed at widening participation in this increasingly pivotal field.