Algorithmic bias and discrimination through digitalisation in education: a socio-technical view

Book section : Chapter

There is a long-standing concern that the use of digital technologies in education risks maintaining, and in some cases exacerbating, existing social and educational inequity. Such debate has intensified due to the increasing use of AI in education that typically relies on advances in machine learning. There are multiple ways that AIEd can lead to unjust practices and outcomes. Employing a socio-technical perspective, this chapter focuses on the way that such systems encode certain biases and values that lead to discrimination of particular groups. It provides an overview of some of the main sources of bias that emerge when digitalising education, from the outset of model creation to their application in practice, and outlines the strategies that can be made by data scientists to mitigate such bias. It then highlights the limitations of these numerical approaches, which largely ignore fundamental questions of justice and power. The chapter concludes by emphasising the importance of governance and regulation to mitigate discriminatory practices, calls for a more nuanced vision of how to best conceptualise and address social and educational inequity in current times, and the need for multi-stakeholder engagement to determine how, if at all, AI should be used in education.

Authors: Eynon, R

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Routledge
• Host title: World Yearbook of Education 2024: Digitalisation of Education in the Era of Algorithms, Automation and Artificial Intelligence
• Chapter number: 15
• Place of publication: Abingdon / New York
• Publication date: 2023-11-30
• Edition: 1
• DOI: 10.4324/9781003359722
• EISBN: 9781003359722
• ISBN: 9781032417905
• Language: English
• Subtype: Chapter