Literacy in the Age of <scp>AI</scp>

Abstract

A Google Trends search for “artificial intelligence” and “literacy” tells us something important about our field's entrance into the age of AI: a long stretch of no measurable interest interrupted by a dramatic spike in late 2022, when OpenAI published ChatGPT. Since then, debates across diverse communities and domains—dinner table conversations, government reports, op-eds, podcasts, and journal special issues like this one—have worried over AI's impact on academic integrity, copyright law, the future of writing, and the very nature of authorship and humanity. At the same time, generative AI has begun to upend established pedagogical practices and approaches to assessment across levels and disciplines, leaving many educators grappling—and sometimes groping—with how to respond (Figure 1). Yet even as our community of scholars and educators scrambles to make sense of this disruption to literacy learning and living, it's worth remembering that the infrastructures that power AI were already everywhere in late 2022, operating well below the flat line and the lack of interest it reflects. Consider the Google Trends tool itself, for example. The clear, familiar-looking graph belies complex systems of data capture, storage, and retrieval—systems increasingly shaped by AI-powered computational techniques like machine learning, the same methods enabling chatbot platforms like ChatGPT. These techniques, which are often as opaque to their creators as they are to those of us outside the black boxes of the platform industry (Hassenfeld, 2023), have been quietly at work in our lives for years, sifting our email inboxes; recommending new music, television, and films to us; and gluing us to precision-guided, short-form video content on social media platforms. They have also been working on students as they learn to read and write—assessing their essays for plagiarism and grammar, personalizing their reading education experiences with adaptive algorithms, and mediating their access to the internet through school-based surveillance platforms. All to say, literacy studies had entered the age of AI long before many in our field became aware of it in late 2022. But to modify an old saying, if the best time to reckon with AI was 20 years ago, the second-best time is now. As we noted in the call for this special issue, “the practice of literacy—whether it be reading algorithmically-recommended books on an Amazon Kindle or composing with predictive text suggestions—is increasingly threaded to sociotechnical factors that are and will continue to shape the boundaries of literacy learning and living in the age of AI.” These factors are not limited to the computational techniques that constitute AI, but, as surfaced in recent literacy scholarship, extend to their entanglements with broader social and political economic forces (Nichols & Garcia, 2022). Studies have shown, for example, how automated writing platforms can encode normative assumptions about language that perpetuate racialized and gendered notions of “good writing” (Dixon-Román et al., 2020), while predictive analytics across ed-tech platforms raise questions about student privacy and algorithmic bias (Williamson & Eynon, 2020). Moreover, the development and deployment of AI in education, literacy education included, is largely driven by a handful of tech companies operating locally and globally, with profit imperatives that may not always align with humanizing educational processes and aims. These systems are not only designed by humans—most of them male, White, and from the Global North—with specific values and objectives (West et al., 2019), but they can amplify these values to the detriment of marginalized communities and the environment. To be clear: The values and objectives guiding deep learning neural networks and large language models cause harm, particularly regarding race, environment, and labor. A growing body of literature continues to critique algorithmic bias embedded in computational systems (Benjamin, 2019; Buolamwini & Gebru, 2018). Benjamin (2019) details the new Jim Code that drives the “engineered inequality” built into AI systems which (re)produce oppression. AI-supported facial recognition systems—as just one example—have misidentified numerous dark-skinned men—including Robert Williams, Michael Oliver, and Nijeer Parks—leading to wrongful arrests, detainment, and imprisonment. Moreover, recent work has called attention to the environmental impact of AI data centers (Li et al., 2023), noting, for instance, that a “large AI model can consume a stunning amount of water in the order of millions of liters for training” (p. 8). Both the “when and where” of AI training matters as “on-site” water efficiency is related to numerous variations, including local weather conditions and grid energy fuel at the moment of training. Furthermore, the dehumanizing aspects of AI training have become increasingly transparent, particularly to workers in the Global South (Crawford, 2021; Irani, 2015). Irani (2015) describes the early stages of massively mediated microlabor of Amazon's Mechanical Turk, for instance, and how workers become transformed into computational resources. This kind of heteromation—or the use of computational systems to harvest financial value from life, human or non-human—feeds AI. Humans, land, and water are now “indispensable mediators” in contemporary capitalist society (Ekbia & Nardi, 2017). In introducing this special issue, we center the “geometries of power” (Massey, 1988) that service AI technologies. AI is a technological, social, and spatial phenomenon that binds places and people, local and global, together. In line with this thinking, Selwyn (2022) urges educators and researchers to adopt a relational approach to AI, particularly in terms of the social harms it causes: “Any instance of some people being disempowered and disadvantaged by the implementation of AI technologies in education is accompanied by others being empowered and advantaged” (p. 624, emphasis original). Importantly, Selwyn cautions against the rise of seemingly harmless AI centers, groups, and organizations—AI for “Good,” “Responsible” AI, “Ethical” AI, and the like—given this relational orientation. Drawing on the work of Golumbia (2021), Selwyn argues that names like “AI for Good” often act as “inadvertent (if not outright) dishonest way[s] of silencing more complex discussions around racism, ableism and other forms of social discrimination” (p. 624). The articles in this special issue attend to these complexities in a variety of ways, opening up possibilities for literacy studies along three thematic strands: (a) philosophical and theoretical expansions, (b) critical and transformative possibilities, and (c) speculative and humanizing approaches for literacies studies in the age of AI. Burriss and Leander (2024), for example, draw on posthumanist theorizing to re-imagine critical literacy for AI, noting that “posthumanism provides us with alternative modes of thinking about the nature of ‘things’ (and ourselves); with an understanding of agency as not a human possession but an accomplishment among/within many human and non-human actors” (p. 560). Kumar and colleagues (2024) further examine the posthuman in the AI era. With an assist from Barad's agential realism, they center the concept of “response-ability” as learners engage with AI systems, moving us away from focusing too heavily on how AI is “approximating or displacing what we take to human capacities” to exploring how we might “reposition such systems as companions” (p. 575). Alternatively, thinking with pluriversal literacies—an approach that embraces forms of knowing, being, and expression beyond the text-centric forms privileged by colonial and Western epistemologies—Bradley and team (2024) use the moment of AI's disruption as an opportunity to clarify, and critically so, the dominant paradigms of “intelligence” in higher education, asking what kinds of intelligences educators should value and nurture for sustainable, anti-colonial futures. Such conceptual work makes important contributions to literacy studies reckoning with AI, inspiring us to continue reimagining the philosophical and theoretical ideas that animate our thinking about the relations between and among diverse forms of power, agency, and intelligence. Authors also explore the critical and transformative possibilities of literacy in the age of AI. De Roock's (2024) article, for instance, engages the concept of heteromation head-on, directly dialoguing with ChatGPT in order to probe its possibilities and limits, particularly around linguistic white supremacy. Cortez et al. (2024) offer alternative ways of subverting dominant ideologies baked into AI through their development of AlgoRitmo literacies. Their article shows how ChicanX communities “can harness AI for world-making and social transformation to potentially subvert the tool's biased mechanisms” (p. 611). Nash (2024) considers the relational aspects of generative AI, its capacity to simulate social interactions in particular, and discusses the potential for young people to (re)conceptualize human relations in response to such relational AI, including their implications for linguistic justice. Sharing findings from a qualitative study of young people's everyday experiences with AI-mediated writing technologies, Higgs and Stornaiuolo (2024) show that young people use AI deliberately and thoughtfully—perhaps more so than many anxious adults suggest—using it both in and out of school to manage daily routines, entertain themselves, and spark their writing and thinking processes. Through a self-reflexive autoethnographic study of Alice and Sparkle, a children's book co-authored with AI, Wandera (2024) surfaces tensions between the individualistic discourses embedded in and outputted by generative AI writing technologies and the communal worldviews of global communities. Finally, Aléman and Martinez (2024) present findings from a Youth Participatory Action Research study involving the Nayah-Irú curriculum, a speculative learning experience designed to support critical data, algorithmic, and AI literacies for young people attending an alternative school in Uruguay. They show participants examined the influence of AI-powered platforms in their lives and developed counter-narratives in response. Taken together, this work highlights the potential for creative engagement with AI alongside the need for ongoing critical awareness not only of its limitations but also its power to amplify and diminish the perspectives and experiences of particular groups, often along racial, economic, ideological, and geographic lines. Finally, authors considered speculative and humanizing approaches to literacy in the age of AI. Mcbride et al. (2024) emphasize the potential for social transformation through humanizing data expression (HDE). As the authors explain, HDE describes a way of “writing with and about AI-powered, data-driven technologies that disrupts pervasive patterns of dehumanization occurring with and beyond digital tools” (p. 679). In their graphic novel-inspired work, Shaw et al. (2024) challenge algorithmic antiblackness through speculative visual storytelling alongside interdisciplinary perspectives on literacy and AI, concluding that “we must ensure that even if the default algorithm is antiblackness, we assist young people in reading analog and digital words and worlds, so they learn to look up, fight, and, eventually, construct algorithms coded for Black liberation” (p. 703). Beck and Levine (2024) investigate how speculative fiction can support students in imagining the harms and benefits of generative AI. Guided by Graham's Writers-in-Community model, they discuss how AI introduces new and potentially destructive ways for historical, political, institutional, and social forces—encoded in the massive language models that power these technologies—to govern individual writers and their communities. Finally, in response to the fervor surrounding the rise of generative AI, Nichols et al. (2024) invite us to interrogate the widespread calls that “literacy pedagogy ought to respond or adapt to changes in our sociotechnical landscape—be it through the integration, critique, or obstruction of new communicative tools” (p. 212). They offer speculative capture to make visible how literacy's “attachment to imagined futures leaves it susceptible to cooptation by, and enrollment in the parallel speculative project that animates AI technologies—thus remaking the former in the image of the latter” (p. 212). In the end, they call for an “ethico-political intention” across the purpose, design, implementation, and use of sociotechnical systems to “enable an alternative worlding, welcome to indeterminacies and ontologies otherwise” (p. 217). [Please note that Nichols et al., 2024, is very much a part of this special issue although technically published in Reading Research Quarterly, Volume 59, Issue 2, pages 211–218.] Collectively, these speculative and humanizing perspectives help us envision novel formations of literacy that center human agency, criticality, and social justice in the age of AI. In concluding, we turn back to Ruha Benjamin and her recent calls for a collective imagination (2023). This collective imagination is realized when “we imagine different worlds together, writing shared stories and plotting futures in which we can all flourish” (p. x). As each new generation expands their collective imagination, she writes, “let them also develop a keener ability to detect bullshit. Each of us can foster the kind of discernment that tells us the difference between New Stories of collective well-being and Faux Fables deciding our collective fate” (p. 21). So let's detect some bullshit as we draw this editorial to a close. In the summer of 2024, Google aired an Olympics-themed commercial for the company's Gemini AI platform. The visuals for the ad included footage of a young runner training with her father, interspersed with images of Sydney McLaughlin-Levrone, a gold-medalist runner at the 2020 and 2024 Summer Olympics. The father, who narrates the ad, begins by saying, “My little girl's always been a runner,” before proudly describing his daughter's admiration for McLaughlin-Levrone and ambition to be a world-beating runner. These expressions take a twist toward the commercial's end, however: “So Gemini, help my daughter write a letter telling Sydney how inspiring she is and be sure to mention that my daughter plans on breaking her world record one day. She says sorry, not sorry.” To review: A father prompting an AI to write a letter to a celebrity on his daughter's behalf. Perhaps unsurprisingly, the ad went viral, although probably not in the way Google hoped, with voices across social media criticizing the company for its problematic depiction of helicopter parenting, inauthentic communication, and the automation of human creativity and expression (Holmes, 2024). What caught our attention, however, was how the company's casual instrumentalization of AI implies its own sort of speculative model of literacy, along with the practices that sustain it. It reflects a new autonomous model of literacy (Robinson, 2023) where, in this case, writing is a productivity task to be offloaded and automated as opposed to a situated, meaningful practice rooted in identity and human connection. The breezy use of AI imagined by the ad disrupts the intimate binds among voice, subjectivity, and agency, replacing it with an extractive, transactional exchange where the affective labor of expression is smoothed over by proprietary algorithms. As we noted earlier in this introduction, and as the Gemini ad illustrates, the opaque yet powerful operations of AI are reshaping literacy in ways that have gone underexamined for too long. The tensions that emerge as AI platform logics bind to situated literacy practices surface in many of the contributions to this special issue, including de Roock's (2024) attention to the extractive nature of AI and Nichols and colleagues’ (2024) caution about the risks for the speculative capture of literacy by platform capital in the age of AI. If the community of scholars and educators that make up the field of literacy studies are to resist such capture and preserve the transformative potential of literacy and literacy education, we must become better attuned to the sociotechnical systems shaping literacy today and committed toward bending (or resisting) such systems for more just, equitable futures. The articles collected in the special issue offer possibilities and provocations for such work as they call attention to many of the challenges and opportunities (and some bullshit) our field must reckon with as literacy evolves in the age of AI. None. Bradley Robinson (corresponding author) is an Assistant Professor in the Department of Curriculum and Instruction, Texas State University, San Marcos, Texas, USA; email: [email protected]. Ty Hollett is an associate professor in the Department of Learning and Performance Systems, The Pennsylvania State University, University Park, Pennsylvania, USA; email: [email protected].

Topics & Concepts