Prof. Dr. Jeremy Knox (University of Oxford & The University of Edinburgh)
Abstract
In a similar way to previous education technologies, recent developments in artificial intelligence (AI) are typically discussed in terms of their ability to enhance learning or make the educational process more efficient. While this dominant discourse is sometimes countered with concerns about the potential harm to ‘humanistic’ notions of learning, general debates are often stymied by an opposition between ‘benefits and drawbacks’. Discussion of the use of Large Language Models (LLMs) in education, for example, often seems to oscillate between predictions of grand transformations in knowledge production and concerns over cheating in assessments.
This talk will suggest that, rather than a habitual focus on the potential enhancement or impairment of learning, a richer understanding of the impact of AI on education might be developed through an analysis of power. This will focus on the ways AI systems increasingly produce, collect, and organise data, drawing on Isin and Ruppert’s (2020) discussion of four overlapping forms of modern power – sovereign, disciplinary, regulatory, and sensory. It is in this way that AI might be understood, not as an instrument of power, or a ‘tool’ through which better or worse educational outcomes can be determined, but rather as a technology through which power produces effects. The use of AI in education will be suggested to both consolidate historical forms of governance through education, as well as enact new assemblages of power in an era of datafication. Resistance to such power will be suggested to come from a return to educational notions of subversion and opacity.
