Where is Algorithmic Literacy in Algorithmic Accountability?

• In Event: Algorithms at Work: The Practice of Prediction - III: Reframing Algorithmic Accountability in Practice

Abstract

Algorithmic accountability has become an imperative topic today, given the increasing role of machine learning in society. However, there is relatively little discussion within the topic on the role algorithmic literacy plays in the accountability process.

Algorithmic literacy is a developing and interdisciplinary offspring of digital literacy and computational thinking. Algorithmic literacy combines the critical evaluation of information found digitally and algorithmic reasoning. As a nascent concept, we propose that its purpose is to recognize algorithms and their goals in technologies and to critically treat their output as subjective through additional recognition of input data and how that data is used.

In today’s age of Big Data, Manovich (2011) writes that there are three classes: “those who create data (both consciously and by leaving digital footprints), those who have the means to collect it, and those who have expertise to analyze it”. We argue that algorithmic literacy is needed in order for this first class to take part in algorithmic accountability. Many have championed transparency as a solution, although important, its application and impact is limited. For example, many deep learning systems are not even interpretable to its creators and transparency at times requires users to have some algorithmic literacy.

If algorithmic literacy is not prioritized within algorithmic accountability, how inclusive is the process? Furthermore, can the power structures described by Manovich change without algorithmic literacy? There are challenges ahead in defining algorithmic literacy, dissemination, and its limits as a tool of empowerment.

Authors

• Elisabeth Sulmont, McGill University

• Elizabeth Patitsas, McGill University