Introduction
The rapid expansion of generative artificial intelligence (GenAI) since 2022 has confronted our societies with questions that extend far beyond the technological sphere. Governments, international organizations, companies, universities, professional organizations, and social actors are debating not only its uses but also its ethical, political, and social implications. Even within religious spheres, such as the Roman Catholic Church, the encyclical Magnifica Humanitas stresses the need for the development of artificial intelligence to remain subordinate to human dignity, social justice, and the common good (León XIV, 2026).
The spread of these tools has been extraordinarily rapid. ChatGPT reached 100 million users within two months of its launch (UNESCO, 2024), while various studies point to their growing adoption in activities related to information seeking, content production, and cognitive support. This process has generated expectations regarding their potential to broaden access to knowledge and improve productivity, but also concerns related to privacy, algorithmic bias, misinformation, transparency, task substitution, and the transformation of human cognitive capacities (Goldkind et al., 2024).
In Uruguay, the National Survey on the Adoption of Generative Artificial Intelligence shows that its use is particularly concentrated among younger people and those with higher levels of education and income. ChatGPT is the most widely used tool, followed by Gemini (CEPE y Fundar, 2025). In education, this expansion requires us to reconsider traditional ways of teaching, learning, and assessment. The 2026 Global Report on Higher Education Trends by UNESCO-IESALC identifies digital transformation and artificial intelligence precisely among the dimensions reshaping the contemporary higher education landscape (UNESCO-IESALC, 2026).
For social work, the debate cannot be reduced to choosing between prohibiting or adopting these technologies, nor to teaching instrumental skills for their use. From the Southern Cone, we propose shifting the discussion from learning how to use GenAI toward learning how to critically understand the algorithmic world in which our professional, institutional, and everyday practices unfold.
Findings accumulated through research we have conducted in Uruguay and in the cross-border region of Salto (Uruguay)–Concordia (Argentina) show that GenAI is already being used by students, faculty (Suárez et al., 2025), and social work professionals (Del Prado, 2026). Its adoption occurs largely through processes of exploration and self-directed learning and coexists with unequal levels of training, limited awareness of institutional guidelines, and concerns about its limitations and effects.
The issue, therefore, is no longer whether or not artificial intelligence should be incorporated. The challenge is to transform spontaneous and unevenly informed uses into processes of critical, ethical, and professionally situated appropriation.
From Banking Education to Augmented Banking Education
Paulo Freire (2014) used the term “banking education” to describe a pedagogical conception in which knowledge is treated as content that can be deposited by those who supposedly know into those who supposedly do not. The expansion of GenAI requires us to ask whether its uncritical incorporation into higher education risks updating and deepening this logic.
In this sense, we might speak of augmented banking education. We deliberately use the adjective “augmented,” which is common in contemporary discourses on artificial intelligence and education, in order to critically invert its meaning: what promises to augment human capacities may also expand and accelerate pre-existing pedagogical logics. By this we mean a mode of education in which artificial intelligence increases the speed of searching, organizing, and producing answers without necessarily transforming pedagogical relationships or strengthening learners’ autonomy.
The problem does not lie in using these tools, but in using them in ways that allow speed, efficiency, and the immediate availability of answers to replace questioning, reflection, and dialogue. Increasing the speed of a practice does not necessarily increase its critical capacity. Education mediated by GenAI may be technologically sophisticated while simultaneously reproducing passive forms of learning. It may produce more texts, more answers, and more information, but not necessarily a deeper and more complex understanding of the reality in which we are immersed.
For this reason, in the face of a possible augmented banking education, social work needs to promote educational processes capable of questioning not only what we can do with artificial intelligence, but also what artificial intelligence does to our ways of knowing, deciding, and acting.
Toward an Algorithmic Liberatory Education
From the perspective of Latin American cultural studies, García Canclini (2020) has argued that the data produced by individuals are reorganized through algorithms capable of shaping patterns of consumption, participation, and citizenship, thereby configuring an increasingly opaque public sphere concentrated in the hands of major technology corporations. Complementarily, Crawford (2024) helps denaturalize the very idea of artificial intelligence by showing that these systems depend on material infrastructures, natural resources, human labor, data, institutional decisions, classifications, and power relations. From this perspective, AI ceases to appear as an autonomous technology and is instead understood as part of complex social, economic, and political arrangements.
Within this context, we propose the notion of algorithmic liberatory education as a conceptual framework for thinking about the relationship between GenAI, education, and social work. The proposal draws on Paulo Freire’s tradition of critical pedagogy (2014, 2018a, 2018b), the contributions to algorithmic literacy developed by Aparici, Bordignon and Martínez-Pérez (2021), and the conception of social work as an ethical-political practice developed, among other Latin American authors, by Margarita Rozas Pagaza (1998).
Algorithmic literacy is a central component of this proposal, but it does not exhaust it. It is not simply a matter of learning how algorithms work or recognizing their biases. Algorithmic liberatory education places these forms of learning within a broader pedagogical, ethical, and political project aimed at strengthening autonomy, situated judgment, professional responsibility, and the collective capacity to intervene in the conditions under which technologies are designed and used. Drawing on the Freirean tradition, it also entails articulating reflection and action within a praxis oriented toward transforming unequal conditions of social reality, including those that algorithmic systems may reproduce, deepen, or render invisible.
We therefore understand algorithmic liberatory education as an educational perspective aimed at developing critical capacities to understand, interpret, and contest the power relations that run through contemporary algorithmic systems. These three movements are not rigid or separate stages, but interconnected dimensions of the same critical praxis.
Understanding involves denaturalizing artificial intelligence. It means recognizing that behind an apparently automatic response lie data, models, infrastructures, human labor, corporate decisions, and classification criteria. Understanding means ceasing to view GenAI as an isolated tool and beginning to see it as part of a social, economic, political, and cultural configuration.
Interpreting involves situating these systems within concrete social contexts. For social work, this means asking which forms of knowledge they privilege, which populations may be rendered invisible, which inequalities they may reproduce, which conceptions of normality they embed, and how they interact with territories, institutions, and subjects historically shaped by unequal power relations. Interpreting therefore entails confronting algorithmic classifications with the complexity of the social realities in which we intervene.
Finally, contesting involves going one step further: moving from critical interpretation to the capacity for intervention. It means rejecting the idea that algorithmic architectures constitute an inevitable or technically closed scenario. It entails participating in debates over their purposes, regulation, forms of use, and criteria of legitimacy, while asking what technologies we want, for whom, and in the service of which projects of society.
From this perspective, education in artificial intelligence for social work should not be organized exclusively around the question, “How can we use GenAI more effectively?” Rather, it should be structured around a broader set of questions: What problems can it help us address? What new dependencies does it create? What forms of knowledge does it validate or exclude? What effects might it have on professional autonomy? Who controls the infrastructures that mediate these practices? What ethical responsibilities do we assume when we delegate decisions or forms of production to algorithmic systems?
Walking, Not Arriving
Speaking of walking toward an algorithmic liberatory education means acknowledging that we are not proposing a finished model. Technologies, their uses, and their consequences are changing at a pace that makes any attempt to formulate definitive answers imprudent. What we propose is a direction for what must necessarily be a collective construction within social work.
Academic institutions, professional councils and associations, national and international organizations, faculty, students, and practitioners are all called upon to participate in this discussion. The challenge cannot be limited to learning how to use new technologies effectively. We need to build capacities to understand, interpret, and contest them, bringing to the debate on artificial intelligence the political, ethical, and professional frameworks that have historically guided critical social work.
Walking toward an algorithmic liberatory education means educating professionals who are capable not only of using artificial intelligence systems, but also of asking what classifications they produce, what forms of knowledge they privilege, what power relations they reproduce, and what possibilities exist for transforming them.
In the face of a technological transformation of global scope, social work can contribute a distinctive perspective by placing human dignity, social justice, human rights, autonomy, and the collective construction of more democratic societies at the center. Perhaps this horizon can be expressed by returning to the well-known maxim of the Chiapas Indigenous movement: contributing to the construction of “a world in which many worlds fit.” In the contemporary algorithmic landscape, this aspiration also requires us to ask what technologies we want, for whom, and in the service of which projects of society.
Bibliography
Aparici, R., Bordignon, F. R. A., & Martínez-Pérez, J. (2021). Alfabetización algorítmica basada en la metodología de Paulo Freire. Perfiles Educativos, 43(Número especial), 36–54. https://doi.org/10.22201/iisue.24486167e.2021.Especial.61019
Centro de Evaluación de Políticas basadas en la Evidencia (CEPE) & Fundar. (2025). Encuesta Nacional sobre Adopción de IA en Uruguay: Relevamiento individuos. CEPE–Fundar.
Crawford, K. (2024). Atlas de IA: Poder, política y los costos planetarios de la inteligencia artificial. Fondo de Cultura Económica.
Del Prado, L. (2026). Usos de la inteligencia artificial generativa en el colectivo profesional de Trabajo Social de Salto (Uruguay) y Concordia (Argentina) [Proyecto de investigación]. Departamento de Ciencias Sociales, CENUR Litoral Norte, Universidad de la República.
Freire, P. (2014). Pedagogía del oprimido. Siglo XXI Editores.
Freire, P. (2018a). La educación como práctica de la libertad. Siglo XXI Editores.
Freire, P. (2018b). Pedagogía de la autonomía: Saberes necesarios para la práctica educativa. Siglo XXI Editores.
García Canclini, N. R. (2020). Ciudadanos reemplazados por algoritmos. Bielefeld University Press.
Goldkind, L., Wolf, L., Glennon, A., Rios, J., & Nissen, L. (2024). The end of the world as we know it? ChatGPT and social work. Social Work, 69(1), 103–105. https://doi.org/10.1093/sw/swad044
León XIV. (2026, 15 de mayo). Magnifica humanitas: Carta encíclica sobre la custodia de la persona humana en el tiempo de la inteligencia artificial. La Santa Sede.
Rozas Pagaza, M. (1998). Una perspectiva teórico-metodológica de la intervención en Trabajo Social. Espacio Editorial.
Suárez, M., Robaina, N., Pou, L., Noboa, A., Del Prado, L., Nieto, E., Anzuatte, N., Prospero, E., González, M., Dorrego, A., & Cabrera, M. (2025). Inteligencia artificial: usos y desafíos para la mejora de la enseñanza de grado en Ciencias Sociales [Proyecto de investigación]. Universidad de la República, Comisión Sectorial de Enseñanza.
UNESCO. (2024). Guía para el uso de la IA generativa en educación e investigación. UNESCO.
UNESCO-IESALC. (2026). Informe mundial sobre tendencias de la educación superior: Hacia una educación inclusiva, equitativa y de calidad en un panorama de movilidad internacional. UNESCO.