AI Platforms and Writing Instruction: Educational Equity in New Learning Ecologies

• In Event: Artificial Intelligence (AI) and Literacy Education

Abstract

In this review paper, we situate the recent flurry of news articles, briefs and op-eds about AI’s implications for literacy education through a systematic review of research on the use of AI platforms in writing instruction. We conceptualize platforms as digital apps, services, and infrastructures for facilitating social and economic exchanges (Gillespie, 2010) and AI platforms as a category of platforms that use data-processing to generate, or respond to, predictions using computational inferences (Bechmann & Bowker, 2019; Williamson & Eynon, 2020). Following scholarship in “critical platform studies” (Author, 2022; Decuypere et al., 2021) we approach AI platforms as “ecologies” – dynamic assemblages of social, technical, and political-economic relations – and we develop an analytical framework for considering the interplay of these relations with multiple valences of “writing”: cognitive, sociocultural, critical, semiotic, and embodied. We use this framework to understand how extant educational research has characterized the (a) function and (b) positioning of AI platforms in writing instruction, as well as (c) their implications for educational equity.

Guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standards (Moher et al., 2009), our review included empirical educational studies published between January 2006 and December 2022 (n=2284, which we narrowed to 188 after removing duplicate and irrelevant results). Through descriptive coding (Saldaña, 2016), we first identified ‘functional’ categories of AI platforms. We then narrowed the pool of studies to those focused on K-12 classrooms (n=28) for closer analysis, using our conceptual framework, of how these AI platforms were ‘positioned’ in relation to writing instruction in a common setting. Finally, we conducted an additional round of descriptive coding to trace references to ‘equity’ (and adjacent terms, like ‘access’ or ‘diversity) to identify if, and how, the literature has engaged matters of educational justice.

Our findings revealed three ‘functional’ categories of AI platforms in writing instruction– assistive, assessment, and authentication – with key differences in how each was taken up empirically. The assessment and authentication categories, for instance, focused almost exclusively on postsecondary applications and contained few studies of how teachers and students actually used AI platforms. We also found the ‘positioning’ of AI platforms tended to focus on the technical dimensions of platforms and their interplay with the cognitive dimensions of writing – leaving other intersections of platform and literacy environments unexamined. Moreover, while many of the 28 studies included culturally diverse student populations, ‘equity’ (and related terms) were virtually absent from our corpus. Taken together, these findings clarify the contours of existing research on AI platforms in writing instruction – what has been included, de/emphasized, and omitted – at a moment when scholars, policymakers, and practitioners are reaching for resources to make sense of new developments in AI. In concluding, we suggest that these findings are instructive not only for grounding the lofty claims now circulating about AI and the future of literacy, but also for charting an agenda for research and pedagogy that confronts aspects of platform and writing environments that have, to date, been omitted from the empirical record.

Authors

• Amy Stornaiuolo, University of Pennsylvania

• Jennifer Higgs, University of California - Davis

• T. Philip Nichols, Baylor University

• Robert Jean LeBlanc, University of Lethbridge

• Roberto de Roock, University of California - Santa Cruz