Emergent uses of generative AI: human agency as a central question?

Abstract

Generative artificial intelligence (GenAI) has become a highly visible phenomenon in education, notably in the form of chatbots, which we refer to here as “conversational artificial intelligence” (cAI). cAI has prompted numerous critical perspectives that emphasize the risks associated with its use, both for users and for teaching, some of whose very foundations are seen as potentially threatened. In this article, we therefore focus on human agency and on how cAI, conceived as an instrument for learning, can contribute to educational success. We first contextualize and situate this notion, which is broader than that of academic success as measured by statistical indicators reflecting performance on examinations. We then present examples of interactions with cAI, developed from narratives collected as part of a Franco-Québec collaboration conducted within the 2IFPER working group. We analyze these interactions in terms of their potential for intellectual emancipation or functional subjection, according to the place human agency occupies within them: humans may “play first violin”, while cAI assists or accompanies them. Conversely, delegation to cAI may reduce human agency within the interaction and, when such delegation becomes predominant, open the way to subjection.