Rise of the machines? The evolving role of AI technologies in high-stakes assessment

Abstract

Our world has been transformed by technologies incorporating artificial intelligence (AI) within mass communication, employment, entertainment and many other aspects of our daily lives. However, within the domain of education, it seems that our ways of working and, particularly, assessing have hardly changed at all. We continue to prize examinations and summative testing as the most reliable way to assess educational achievements, and we continue to rely on paper-based test delivery as our modus operandi. Inertia, tradition and aversion to perceived risk have resulted in a lack of innovation (James, 2006), particularly so in the area of high-stakes assessment. The summer of 2020 brought this deficit into very sharp focus with the A-level debacle in England, where grades were awarded, challenged, rescinded and reset. These events are potentially catastrophic in terms of how we trust national examinations, and the problems arise from using just one way to define academic success and one way to operationalize that approach to assessment. While sophisticated digital learning platforms, multimedia technologies and wireless communication are transforming what, when and how learning can take place, transformation in national and international assessment thinking and practice trails behind. In this article, we present some of the current research and advances in AI and how these can be applied to the context of high-stakes assessment. Our discussion focuses not on the question of whether we should be using technologies, but on how we can use them effectively to better support practice. An example from one testing agency in England using a globally popular test of English that assesses oral, aural, reading and written skills is described to explain and propose just how well new technologies can augment assessment theory and practice.