Paper short abstract:

In this presentation, I explore Artificial intelligence (AI) as a form of future-making. Using ethnofiction as a genre of ethnography, my presentation offers a glimpse into the politics of AI research and asks what future is being reproduced through datasets used to train artificial intelligence?

Paper long abstract:

In this presentation, Artificial intelligence (AI) is framed as a form of future-making. Exploring ethnofiction as a genre of ethnography within anthropology, my presentation offers a glimpse into the politics of AI research and how it accounts for a limited envisioning of different and desirable futures. AI is a computational technique that makes determinations without human direction (Crawford 2021). Those determinations are based on the analysis of large quantities of data (big data). AI is trained by different processes, one of which is Machine Learning, which ultimately learns from data on how to act or decide on different topics and in very specific contexts. AI is also a way of sorting through large quantities of data and making decisions based on this assembled information (Birhane 2021). AI can only predict based on what has already occurred or what has already become data. With this in mind, this presentation focuses on predictive data models and asks, "what future is being reproduced through datasets used to train artificial intelligence?" The argument highlights how AI can represent a limited potential for imagining new situations and scenarios, especially as database updates always work with the past and imagine the future using information exclusively from the past.