Matthew Kirschenbaum and Rita Raley's clarion call to evolve our language disciplines to respond to the transformational effects that AI and large language models (LLMs) are having on the profession is both timely and urgent. It is vital to assess the structural reorganization this technological shift will represent for institutions of higher learning, and more specifically to evaluate repercussions for the MLA and for the study of languages, literatures, and cultures, broadly understood.
I want to respond from my own position as a specialist in Latin American and Iberian cultural studies (situated within a languages and literatures department), a field shaped predominantly around the study of those geographies, cultures, and communities from the nonmonolithic Global South—the majority world—whose cultural products are primarily in Spanish or Portuguese, as well as indigenous languages. Like other language disciplines, mine is financially undergirded, at least in part, by basic language programs that are facing grave challenges from machine translation, text-to-speech technologies, and foreign language online learning tools, aggravating preexisting decreasing enrollment trends. The additional stress placed by AI and LLMs on basic language programs and the study of languages generally might serve as a canary in the coal mine, prefiguring challenges facing the discipline in the main: as language study goes, so will the study of literatures and cultures, and the humanities as a whole.
From this vantage point, I am not only concerned about the deleterious consequences that AI and LLMs are having on our profession at large, their imprint on the study of literature and language, but equally unsettled by the amplifying effect these technologies could have on furthering the existing marginalization of languages “other” than English, by buttressing English dominance and further eroding linguistic diversity in higher education and beyond, as well as enforcing long-standing North/South power asymmetries, the legacy of colonialism, capitalism, class inequality, and preconceptions about the “proper” directionality of knowledge production. These issues are glaringly obvious in the implementation of AI in the Global South, where the technology risks becoming an apparatus for imposing a new form of coloniality. There is a body of research showing that standardized English is favored even within anglophone nations like the United States, as evidenced by the existence of AI-driven accent modification technologies, based on reprehensive raciolinguistic theories (Payne et al.; Roche).Footnote 1 These accent “neutralizing” tools are also in widespread use in Global South customer service sectors (such as call centers). If AI can energize language discrimination against racially minoritized groups, what might we expect as similar technologies and linguistic forms of oppression are directed to and against other languages, regions, and cultures, particularly from and in the marginalized South? And how might such forms of discrimination extend to academia and research, affecting scholars working on and in foreign languages and cultures? A comprehensive response to these questions necessitates the involvement of foreign language scholars and cultural critics that specialize in these languages and regions.Footnote 2
What are the specific challenges posed by AI and LLMs for those of us researching cultural production in languages other than English, those so-called foreign languages, often also spoken in majority anglophone nations (as with the more than fifty million Spanish speakers in the United States)? If, as Kirschenbaum and Raley lucidly diagnose, our profession's participation in the discussion surrounding AI is always already belated in relation to technical fields (data science and programming, engineering, informatics), then the participation of scholars working in and from non-English spaces and languages risks being especially lagging, if it arrives at all—my own intervention a mere opening salvo, which ideally would be followed by contributions specifically targeted to the analysis of AI and LLMs in other regions, languages, and cultures by foreign language scholars (as is already happening in other fields like education and the law). Scholars of foreign languages and literatures should spring to action, as Holly Dugan and Dolsy Smith state in their contribution to this issue: “if we are to grapple with the effect of AI on the profession, the time is now.”
The imperative is not just to rethink the MLA's collective response to AI as a “threat” or an “opportunity” but to ensure that response is inclusive of the general interest represented by our professional organization's membership and that sufficient space exists for diverse perspectives to engage in the conversation. This means fostering exchanges that consider what AI and LLMs signify not just for English, but in the context of other global languages, literatures, and cultures; thinking about what scholars of other languages and cultures might offer in terms of distinct critical AI methodologies; and considering the relationship and positionality of these scholars vis-à-vis the MLA as a body. A way to encapsulate these points might be to ask what a non-Anglo-American critical AI studies might look like, how the participation of foreign language scholars might be encouraged, and what these global AI studies could contribute to understanding the effects of AI and LLMs on the profession and higher education more broadly.
The development of such an “accented” critical AI studies necessitates overcoming a degree of inertia, both with regards to the way AI technologies themselves are being implemented and in terms of how the critical response to AI is shaped within academia. On the one hand, the amplification of Anglocentrism by AI and LLMs is technically determined, as a result of documented biases in algorithms, models, and processes (Noble; Everett; Hajri). Algorithmic processes also tend to homogenize cultural and linguistic difference and overemphasize English as the lingua franca of technology (Curry and Lillis; Hohti and Truman; Rosa and Flores), adding to language oppression (Roche).Footnote 3 On the other hand, the totalizing dominance of English could be unintentionally reproduced within our disciplinary efforts at addressing the effect of LLMs, AI, and other emerging technologies, because prevailing epistemological paradigms might place less emphasis on contributions from foreign language scholars, especially when published in other languages (machine translation could prove beneficial in this case). It is incumbent on critical AI studies to remain inclusive of these other disciplinary outlooks, attuned not only to the intersectionality of geography, race, ethnicity, caste, class, gender, religious status, and ability but also to linguistic diversity.Footnote 4
At present, efforts to understand AI are being determined and defined primarily by the anglophone sector of our disciplines, perhaps a reflection of the aforementioned belatedness with which foreign language scholars engage with potentially transformative cultural shifts. The inaugural issue of the journal Critical AI (Oct. 2023), which contained essays addressing broader questions of data justice, equality, racism, and capitalism, also called for “critical AI studies to forge an interdisciplinary community of practice, alert to ontological commitments, design justice principles, and spaces of dissensus” (Bode and Goodlad). Increasing the representation of perspectives from the Global South (and especially from Latin America) will further “globalize” AI studies while also enabling more of the conversation to focus on AI's specific impact on foreign languages, non-English cultural studies, decoloniality (relevant work is emerging on this subject already), and its material effects on these regions. Significant work on data colonialism (Thatcher et al.), algorithmic coloniality (Mohamed et al.), or the coloniality of technological power (Ricaurte) could be expanded on by scholars of literatures and languages other than English who examine, for example, the intersection of AI and cultural production, further diversifying research topics while addressing concerns specific to these disciplines and geographies.
I enjoin scholars in these “other” disciplines (including my own, Latin American and Iberian cultural studies, but also from other languages, area studies, and ethnic studies programs) to become involved in shaping the AI studies field, further diversifying it linguistically and culturally, reinforcing decolonial perspectives, as well as representing the interests of these other regions’ underrepresented populations and those of their diasporic communities in the West. Otherwise, critical AI studies risks replicating the disciplinary asymmetry exemplified by the digital humanities, where early adoption by English-language scholars and low participation by other languages (and the lack of circulation of untranslated work) resulted in a relative paucity of geographic and linguistic diversity of both projects and researchers, furthering universalist and monocultural perspectives and damaging the field's growth potential (Brown; Nilsson-Fernàndez and Dombrowski; Spence and Brandao; Fiormonte).
AI is only the latest challenge to foreign languages, under siege in the academy for decades. Even the teaching of major or most-studied languages (such as Spanish, French, Italian, Japanese, and Russian) is in freefall in terms of US enrollments, and, as shown by the MLA enrollment report published in 2023, these declines worsen every year (Lusin et al.).Footnote 5 Working within our institutions, which still prioritize English, we might wonder what foreign language scholars can do to reverse this decline. The excitement with which university administrators are including AI- and data-related studies in curricular and business initiatives seems to work in tandem with a desire to deemphasize certain humanistic disciplines and methods.Footnote 6 Paradoxically, a solution might be to appropriate the technology and study its impact: to maintain relevancy and avert disaster, it behooves the foreign languages to adapt by incorporating AI into research, learning to integrate it into our methodologies, and knowledgeably critiquing its effects. Then again, this adoption of AI, which can provide a temporary respite to our disciplines, may also hasten a process of devaluation of humanistic approaches, as Kirschenbaum and Raley hypothesize, since “it is precisely this tokenizing of language—both its subordination to technical processes and its symbolic devaluation—that promises to render universities vulnerable to the market logics by which the neoliberal institution has staked its primary claim to a sustainable, if not survivable, future.”
Yet we cannot bury our heads in the sand and pretend nothing is happening. As Katherine Elkins proposes in this issue, “it won't be immediate, but there's no question that AI research, writing, and translation will automate some of what we have typically thought of as our domain. While some of us engage more directly with AI, others had best start imagining what the rise of AI means for our everyday practice of teaching language and literature.” Simply ignoring the presence of AI and carrying on will not ameliorate the fare of language disciplines either, and in fact the technology could exacerbate the gravity of the situation—for example, as the need for learning a foreign language is rendered obsolete by machine translation (Piller). Speech recognition tools will facilitate the acquisition of speaking and listening skills for students who opt to learn a language on their own, further reducing the need for language teachers—and irrespective of actual results (at the same time, these tools could help people with disabilities in various ways). Recent advances in machine translation—a technology that has been around since the 1960s—threaten at least the more “mechanical” aspects of language learning, leaving (for now) the area of global cultural competence and critical analysis as the sole space for the foreign languages scholar to hold out, as linguistic competence is surrendered to AI technologies, regardless of whether actual competence is achieved.Footnote 7 Coupled with the proliferation of online courses, AI will be a “natural” fit for language teaching even for those institutions that elect to maintain foreign language requirements against administrative and economic pressures. With a promise of personalized teaching, instant feedback (however inane), and self-paced study, corporatized AI applications will be marketed as the “ideal” solution for those looking to learn a language with minimal effort and time investment, again regardless of the deleterious effect such overreliance on translating tools might have for actual language learning. In practice, AI could make it more difficult to learn a language, not less, because it does the work for you. Similarly, this tool might make it more difficult to learn how to write, to think critically, or to be creative if we allow it to wholly supplant our own efforts (our labor, our laboring) in any of these arenas.Footnote 8 The response by foreign language scholars cannot be to reject AI and LLMs wholesale (impossible) and refuse to be a part of their analysis and development. Quite the opposite; as Lauren M. E. Goodlad and Samuel Baker suggest, we must turn the tables and “disrupt AI,” and scholars of foreign languages and cultures can and should do a significant amount of disrupting. Perhaps the relevance of languages could be bolstered, rather than eroded, through a meaningful scholarly engagement with critical AI studies, in addition to diversifying it as a field. Our strength as humanists remains, as Dugan and Smith see it, in “the practices of humanistic pedagogy—of close reading, of compassionate listening, of impassioned but circumspect speaking and writing.” In an equally optimistic view of how we, as scholars, can reclaim agency, Aarthi Vadde observes in this issue that “[w]e can of course imagine the nightmare scenario of austerity politics and writing instruction by app, but university-based literature and writing instructors can still shape whether and how these technologies are used in the classroom.”
One specific way to bring in the foreign language scholar is through community-based research and public humanities work—for example, through projects that analyze the effects of AI on Spanish-speaking populations in the United States, or by examining the potential use of the technology to preserve indigenous languages or studying how machine translation might distort indigenous cultures (Chandran), or by considering whether AI-driven translation could be beneficial for immigrant communities. We could envision cultural studies research analyzing the use of AI in narratives from various regions—for example, its adoption by Global South filmmakers for a variety of tasks.Footnote 9 The situatedness and material specificity of AI in the Global South can provide concrete grounding for at-times abstract concepts developed by critical AI studies in the North, while also generating particular autochthonous forms of critique that focus on countering new digital-AI configurations of the coloniality of power that enforce Anglo- and Eurocentric systems of knowledge production and economic control, as defined by Latin American subaltern studies (Quijano and Ennis; Mignolo; Mignolo and Walsh; Ricaurte). Undoubtedly, there is a pressing need for more scholars from the so-called periphery or working on “peripheral” literatures and cultures to contribute to AI research, especially those housed in foreign languages departments. The point is that we have the necessary skill set to intervene from a variety of angles (technical, critical, theoretical, ethical), since, as Seth Perlow maintains elsewhere in this issue, “Humanities scholars, especially those studying interpretive theories and methodologies, are uniquely equipped to address the challenges that large AI models pose. We embrace literature for its resistance to rigid logic, its nuance and ambiguity, and modern criticism offers a robust tool kit for rigorous thinking at the limits of knowledge.”
In The Fall of Language in the Age of English, Minae Mizumura argues that the survival of local languages in the face of the onslaught of English-language culture depends on making the case that certain types of knowledge can be accessed only in their original language, in particular those dealing with literary-cultural forms. Global English (“Globish”) exists because the status of English as an “elite” language is bolstered by transnational capitalism and technological homogeneity (McCrum). Or as Raley frames the question of the effect of English dominance in the academy, “Literature in English, global English, global studies, and cosmopolitanisms can be read as new universalisms that are merely simulations of the old, that themselves contain a homogenizing and totalizing impulse, and that signify an epistemic and literal violence that the academy cannot afford to ignore” (52). Perhaps paradoxically, the primacy of English is in some small measure under pressure by AI itself, although in response to market demands. Thus, while currently most LLMs are trained primarily on English-language data, since the bulk of text scraped from the Internet is in that language, this situation is evolving: promoted by the government of Spain, plans are underway to create LLMs that are trained on Spanish. These diversifying moves, however, are bolstered by capitalist industries that stand to profit from (exploiting) Latin American markets, reengaging with the coloniality of (digital) power. The Spanish-trained LLMs will inject some competition and linguistic diversity to a US-dominated sector, but also reintroduce (in a new bottle) age-old colonialisms, resource extractivism, and labor exploitation (Crawford). As more efforts to capitalize on AI for exploitation emerge in other languages and regions, scholarly attention will become necessary to expose, document, and counter these trends.
These are just some areas of research in which scholars of and from the Global South, including foreign language scholars in the MLA, could further intervene—as some of us are already doing. The concept of a more collaborative approach to design and to research, attuned to place-based considerations, aligned with human interdependence, and reconciled with ecological concerns and social justice—as laid out, for example, in Arturo Escobar's Designs for the Pluriverse, Maurizio Tinnirello's edited volume The Global Politics of Artificial Intelligence, and Kate Crawford's Atlas of AI—can serve as inspiration toward not only a more just and equitable implementation of AI (through the examples discussed in these texts) but, more to my point, an inclusive approach to scholarly research about AI, in particular by the MLA and its membership. At stake is nothing less than the survival of our language disciplines, in their full and diverse existence, in their capacity to understand global cultures, and in their assurance that regardless of the current AI-LLM technological shift (or others that will come), they will maintain a vibrant multilingual organization that remains committed to its foreign language scholars and initiatives and that counters trends of institutional disinvestment in language departments. Approaches to studying literatures and cultures will need to adapt to new conditions, but our disciplines remain essential to higher education and will imagine innovative ways of conducting research and teaching that are critically responsive to AI while staying anchored in the humanities. It is not only up to the MLA as an organization to facilitate the inclusion of these scholars in the nascent field of AI (although that organizational effort is needed), but also incumbent on scholars from our language disciplines, especially those working on, from, and in the Global South, to take up the challenge, become AI literate (Bali), and join the conversation. Our long-standing and well-honed expertise in analyzing “other” cultural texts from non-Anglocentric perspectives, situated in their specific linguistic, geographic, historical, and political frames, can now be applied to further empower critical AI studies and usher in a global critical AI studies—for what is AI if not another “text,” another “language,” perhaps another cultural form altogether?
