Send message to Author

Short abstract:

Nowadays, decision-making relies on predicting future events. New algorithms offer predictions but their accuracy remains uncertain. This contribution explores how they shape future predictions amidst high uncertainty, focusing on socio-material affordances and onto-epistemological implications.

Long abstract:

In the era of poly-crises, effective decision-making is more and more relying on predicting and speculating about future events. In an increasingly complex world, societal actors require knowing as much as possible about the future, precisely they need information about possible futures without being able to determine the probability of these events occurring with accuracy. Predictive algorithms are increasingly being used as a tool of choice, as they promise not only to perform more objective analyses using complex computations and heterogeneous data sets. Furthermore, they have become producers of future predictions as well.

This contribution aims to change the perspective on future predictions in times of high uncertainty and when one can only make speculations about possible futures. Instead of focusing on (socio-technical) imaginaries and their performative effects, as most approaches do, this contribution focuses on the relationality and materiality of future predictions. This opens up the perspective of how predictions are produced and stabilized in and through socio-material arrangements and their affordances. Specifically, the transformation process of possible futures (speculations) into probable futures (predictions) will be traced to make uncertain futures actionable.

Drawing on empirical work on anticipatory action programs using predictive algorithms in the field of humanitarian crisis forecasting, this contribution (a) will analyze the socio-material interplay of actors and algorithms in the production and transformation process of actionable future knowledge, and (b) discuss the onto-epistemological status of the future in the face of new technologies like predictive algorithms.