Towards a New Learning Ecology: The Mediator Role of Teachers Facing Generative AI in the University Context of Peru, Chile, and Colombia

1. Heuristic Phase (Individual): Each researcher performed open coding and data segmentation in their respective countries, identifying preliminary patterns.

2. Hermeneutic Phase (Collaborative): Through researcher triangulation (Denzin, 2012), the multinational team met synchronously to reach a consensus on emerging categories. An intersubjectivity criterion was applied, selecting for the final report only those findings and codes that showed cross-sectional recurrence across the three national contexts, discarding non-representative singularities. It is important to note that data involving discrepancies among researchers were excluded from the generation of this text.).

1. The Collapse of Taxonomies and Cognitive Development

One of the most disruptive findings is the notion that traditional taxonomies, such as Bloom’s, are undergoing an erosion of their foundations. P. #47 points to a radical paradigm shift: “Bloom’s taxonomy and others are becoming ineffective because AI achieves all those levels. For students, learning as previously assumed is no longer enough; with GenAI, we must go further, promoting higher-level skills that AI cannot achieve, such as critical thinking, creativity, and socio-emotional skills.”

This observation suggests that the pedagogical purpose can no longer be information retention, but rather students’ cognitive development. This stance aligns with authors such as Bearman et al. (2024), who suggest that education must transition toward a pedagogy of critical evaluation, where value resides in judging the quality of output produced by synthetic agents. As P. #3 indicates, the purpose is “to develop and strengthen critical thinking to resignify what GenAI presents… the time and effort invested are not in generation but in the analysis and construction of thought.” Under this lens, GenAI should be integrated to liberate students’ brains from low-value tasks, allowing them to focus on complex problem-solving.

2. Personalization, Leveling, and Pedagogical Equity

A recurrent theme was the vision of GenAI as a tool to equalize learning opportunities. P. #28 highlights that the technology allows “each student to receive an education adapted to their individual needs,” a view shared by P. #8, who adds that the purpose is “to provide personalized attention to each student… ensuring the student is the center of the process.”

In the Latin American context, this purpose takes on a nuance of educational justice. P. #24 reinforces this idea by conceiving GenAI as a “leveling tool for students who bring learning gaps from secondary school.” This aligns with Ouyang and Jiao’s (2021) contributions on adaptive learning. However, it is essential to understand that the pedagogical meaning provided by educators is not limited to facilitating study; it aims to allow the student to develop a “better autonomous learning process” (P. #42) and improve their “capacity to learn” (P. #38), implying that GenAI should be integrated to enhance students’ academic self-efficacy.

3. GenAI as a Catalyst for Innovation and Professional Development

For educators, the pedagogical purpose also resides in the enhancement of research and digital competence. P. #1 indicates that the goal is “to improve academic production quality and facilitate the search for information and sources in research.” This is complemented by the vision of P. #13, who seeks “to empower skills for educational innovation… to generate changes in teaching strategies.”

This finding aligns with Crompton and Burke (2023) regarding how GenAI acts as scaffolding that accelerates literature reviews. P. #46 exemplifies this practice by using it for “article analysis, developing bibliographic references… and search expressions for databases.” The ultimate proposed purpose is “to link the student with the technology they will use in their professional world” (P. #18), ensuring that higher education is not an isolated bubble from students’ technical and professional realities.

4. The Ethical Sense: From the Correct Answer to the Correct Question

Finally, a pedagogical sense oriented toward epistemic validation emerges. P. #35 mentions that the purpose is “to foster critical thinking by contrasting AI-generated information with contextual reality.” This implies that the sense of teaching is no longer delivering truths, but teaching how to “question the certainty of the information obtained” (P. #15) and “verify that the data found are real” (P. #10).

Educators warn that the pedagogical sense is at risk if not accompanied by solid ethical training. P. #11 states that “integration must always have an educational purpose,” and P. #4 warns that the purpose should be “to support and serve as aid… without neglecting the fact that they must live a life away from [social] networks.” This reflection coincides with UNESCO (2024) guidelines, which urge maintaining human agency at the center, transforming the classroom into a space where the goal is not to consume AI results, but to learn to co-create with it responsibly and critically.

5. National Variations in the Purpose and Pedagogical Meaning of GenAI

Although a regional consensus exists perceiving GenAI as a catalyst for efficiency, the “ultimate purpose” varies depending on national educational agendas and institutional contexts.

In Chile, stances (e.g., P. #3 and P. #12) show a marked inclination toward the resignification of intellectual effort. The purpose is not merely to use the tool but to demand a “pedagogical lucidity” (P. #5) that allows GenAI to be a tool for meaningful learning. Similarly, a shift in emphasis from “generation” to the “construction of thought” is observable. As P. #3 indicates, saved time must be invested in the “analysis and construction of critical thought.”

Conversely, Colombian educators present a stance strongly linked to research and high-quality curricular design. For P. #1 and P. #14, the primary purpose is “to improve academic production quality” and streamline complex processes like “data tracking and identification of research trends” (P. #20). This orientation toward academic productivity resonates with Crompton and Burke (2023). Likewise, there is a concern for efficacy in knowledge searching (P. #31) and strengthening curricular design without losing the “teacher’s protagonism” (P. #20).

In Peru, the predominant stance is operational streamlining to connect with new generations. Professors like P. #41 and P. #45 see GenAI as a way to “complement pedagogy tailored to the student generation of these times.” The focus is on making “learning more meaningful” (P. #38) and achieving “efficient information organization” (P. #40). There is a notable emphasis on AI serving for “more efficient self-taught learning” (P. #36), coinciding with the “Adaptive Learning” perspectives of Ouyang and Jiao (2021).

Finally, in Mexico, Venezuela, and El Salvador, the purpose is perceived as a “support for basic pedagogical labor.” In Venezuela, P. #2 highlights the purpose of “improving didactic strategies inside and outside the classroom,” suggesting a vision of GenAI as an extension of the physical teaching space.

1. Curricular Redesign and Learning Architecture

A prominent finding is the transformation of educators into “experience architects.” P. #42 highlights the use of GenAI for creating high-quality digital content and didactic curricular activities that strengthen autonomous learning. Similarly, P. #6 uses the tool to generate test examples, recommended bibliographies, images, and presentations, facilitating constant syllabus updates.

This capacity for immediate response aligns with what Chiu (2024) terms a “dynamic curriculum,” where AI allows teaching materials to evolve alongside scientific production. Integration enables the rapid planning and structuring of study topics (P. #36), granting flexibility to instructional design. This coincides with Mollick and Mollick (2023), who suggest that the teacher acts as a curator of learning pathways, delegating base structuring to the machine to focus on pedagogical intent.

2. Support for Diversity and Socratic Tutoring

An emerging high-impact use lies in supporting students with specific needs. P. #21 indicates using GenAI to adapt content based on students’ learning difficulties or disabilities. This sense of inclusion is complemented by AI’s function as a permanent tutor. P. #8 mentions that the tool allows for the preparation of material tailored to the target group, applying active methodologies that ensure complex content remains accessible to diverse profiles.

This approach engages with the personalization paradigm (Ouyang & Jiao, 2021), where AI acts as a mediator reducing knowledge access gaps. P. #31 translates this into the classroom through role-playing simulations for clinical cases or problem-based situations, transforming theory into an immersive experience where the student enters a dialogue with knowledge.

3. Transformation of Assessment: From Product to Process

Assessment is the area of greatest experimentation. P. #47 uses the tool for analyzing complex topics and developing rubrics, shifting the focus toward higher-order skills. Likewise, P. #7 reports that the technology supports creative processes to achieve more comprehensive evaluative activities. This suggests that educators use AI to design instruments that no longer evaluate data repetition but rather the student’s critical capacity.

Lodge et al. (2023) warn that assessment must become “AI-resistant,” focusing primarily on the process. P. #44 applies this principle by generating executive summaries of extensive books, not to substitute reading, but to facilitate roundtable discussions where learning validation occurs through live debate rather than the final product.

4. Sophistication of Scientific Production and Documentary Management

In research, GenAI use is becoming increasingly technical. P. #46 describes an “augmented research praxis,” using the tool to define current topics, analyze articles, and draft research questions, objectives, and hypotheses. This level of assistance makes academic labor more rigorous and agile, optimizing bibliographic searches through the generation of database-specific search strings (P. #18).

This specialized use coincides with Crompton and Burke (2023), who position GenAI as an engine for scientific democratization. P. #20 highlights this by employing the technology to translate high-impact articles, breaking linguistic barriers that have historically limited academic updates in Latin America.

5. Comparative Analysis of Everyday Uses by Country

In Chile, the focus is on curricular engineering and pedagogical equity. For P. #12, GenAI is a fundamental resource for working with students with lower comprehension skills, allowing them to refocus on meaningful learning. The Chilean daily routine centers on constructing methodological scripts and assessments to prioritize university pedagogical work over tedious administrative processes (P. #14).

In Colombia, the stance is strongly linked to research productivity. GenAI is perceived as a “thinking partner” (Mollick & Mollick, 2023) that allows for contrasting the syllabus with international trends and updating bibliographies via advanced databases (P. #19).

In Peru, the predominant application is oriented toward didactic agility and multimodal resource creation. Educators use AI to complement pedagogy for today’s generation (P. #41), utilizing tools for graphic and visual generation to optimize time (P. #38).

In Venezuela and El Salvador, the purpose is seen as support for strategy design and a catalyst for teacher creativity. In Venezuela, P. #2 highlights that the technology helps overcome creative blocks, assisting in classroom didactics through the design of innovative strategies.

1. From Knowledge Transmission to Hybrid Orchestration and Curation

The predominant stance (Professors #2, #3, #6, #11, and #47) is the transition toward a facilitator and mediator role. P. #3 notes that educators must accompany the learning process, which requires deeper subject knowledge to supervise the machine. Authority no longer stems from information access but from the capacity to discern quality, aligning with the notion of pedagogical curation (UNESCO, 2023; Crompton & Burke, 2023). Educators are evolving toward “hybrid orchestration” (Weng et al., 2024), where they lead while AI assists (P. #15). This implies knowing more systematically to evaluate if AI output is correct or a “hallucination” (P. #3).

2. Dialogue Engineering, Critical Validation, and Algorithmic Responsibility

An emerging dimension is the mastery of dialogic interaction with technology. P. #22 describes their function as a critical guide teaching students to ask the right questions (prompts) rather than just searching for answers. This shift teaching competence from delivering finished solutions toward formulating complex problems. P. #1 emphasizes that this requires being proactive and responsible, applying tools to evaluate authorship and critical source management. This demand for “algorithmic responsibility” coincides with the ethics of human supervision (Bearman et al., 2024).

3. Operational Efficiency and the Recovery of the Reflective Intellectual

GenAI has brought tangible changes in productivity, allowing for resource diversity and time efficiency (P. #14). Delegating mechanical tasks allows educators to reclaim their role as researchers of their own practice and as reflective intellectuals (P. #14). Professionals can now focus on aspects requiring human sensitivity, such as emotional accompaniment and ethical guidance (P. #13). According to Crompton and Burke (2023), teaching is shifting from “execution” to “design.” However, this generates tensions; P. #7 warns of the risk of “deprofessionalization” (Selwyn, 2024), stressing that teachers must reaffirm their human value through empathy.

4. Co-agency, Vulnerability, and Challenge Design

Transformation also implies accepting pedagogical vulnerability. P. #41 suggests that technology requires educators to learn alongside students, breaking traditional hierarchies in a process of mutual learning. This dynamic is defined as “human-AI co-agency” (Weng et al., 2024), where the teacher acts as a designer of challenges. P. #40 indicates that efficient information management frees up space for reasoning and critique, elevating teaching from basic instruction toward fostering higher cognitive functions (P. #45).

5. Teaching Role and Professional Transformation: Regional Comparative Analysis

In Chile, transformation is seen as an evolution toward “pedagogical lucidity.” The teacher acts as an active, personalized learning facilitator (P. #12), with a marked concern for avoiding deprofessionalization due to automation (Selwyn, 2022).

In Colombia, the role is defined as the “architect and supervisor of scientific production.” The teacher is the protagonist integrating technology (P. #2), assuming a researcher-designer role (Crompton & Burke, 2023), with an emphasis on algorithmic responsibility to evaluate authorship (P. #1).

In Peru, identity has shifted toward epistemic vigilance and ethical guidance. The teacher is a “content tester” (P. #6) and a “captain of the ship” (P. #5), whose value lies in data validation and raising awareness about proper technology use (P. #4). This approach revalues academic integrity (Moorhouse et al., 2023).

In Mexico, El Salvador, and Venezuela, faculty emerge as experience designers. In Venezuela, technology acts as an engine of curiosity (P. #1), where teachers use design-based pedagogy (Weng et al., 2024) to create learning situations where students interact critically with AI.

1. Humanizing Efficiency and the Democratization of Knowledge

Operational efficiency is the primary advantage. It is perceived as a device that frees time for high-value pedagogical tasks, allowing the teacher to focus on content rather than form (P. #3). By delegating “educational bureaucracy” (Mollick & Mollick, 2023), teachers reclaim their role as mentors (P. #8, P. #14). This enables mass personalization and democratic access to cutting-edge information (P. #3, P. #28), validating the “precision learning” paradigm (Ouyang & Jiao, 2021).

2. Cognitive Atrophy, Dependency, and the Illusion of Learning

The most critical challenge is the risk of mistaking efficiency for real learning. Faculty warn against “facilismo” (the path of least effort) (P. #5, P. #26), where delegating writing to AI can lead to the atrophy of critical skills (Dwivedi et al., 2023). P. #28 defines this as a risk of “excessive dependency” that compromises autonomous thought. If students skip the process of structuring ideas, the university risks becoming a “vender of empty credentials” (P. #36).

3. Dehumanization, Teacher Anxiety, and the Algorithmic Divide

Integration generates ethical tensions, such as the fear of “artificializing” teaching (P. #7) and losing the human bond (Selwyn, 2024). Additionally, “cognitive stratification” (Williamson, 2024) emerges, as unequal access to paid software creates a student elite with algorithmic advantages over the majority (P. #38, P. #41, P. #25).

4. The Assessment Crisis and Institutional Reinvention

Traditional grading methods are becoming obsolete. P. #20 argues that the written essay has lost its validity, forcing a migration toward “AI-resistant” assessments like applied projects and oral defenses. This transition toward “dialogic assessment” (Bearman et al., 2024) requires institutional backing, which is currently lacking (P. #22, P. #25), leaving teachers in a zone of “normative vulnerability” (Selwyn, 2024).

5. Impact Management: Regional Comparative Analysis

In Chile, management focuses on “reflexive depth” against dependency. GenAI is valued for quality time (P. #14), but there is a constant alert regarding student passivity (P. #36).

In Colombia, the impact is pragmatic and research-oriented. The advantage is measured by breaking linguistic barriers (P. #19, P. #20), while the challenge is the assessment crisis and the need for intellectual authorship validation (Crompton & Burke, 2023).

In Peru, didactic agility predominates alongside the challenge of integrity. AI is seen as an ally for information organization (P. #40, P. #41), but there is a founded fear of data falsification and plagiarism (P. #6).

In Venezuela, El Salvador, and Mexico, impact management is rooted in “creative resilience,” where GenAI acts as a partner against teacher burnout (P. #2, P. #17). The structural challenge remains the digital divide (Williamson, 2024).

1. Adaptive Learning, Leveling, and Democratizing Success

The most significant impact is the capacity to break classroom standardization via adaptive learning (Ouyang & Jiao, 2021). P. #28 notes that each student receives an education adapted to individual needs. In Latin America, this is critical for leveling students with secondary school gaps (P. #24). This multimodal personalization democratizes academic success (Baidoo-Anu & Ansah, 2023).

2. Autonomy, Self-regulation, and Dialogic Interrogation

Student agency is shifting toward “augmented autonomy.” Professors #42 and #18 report that students achieve better autonomous learning through automated Socratic tutoring. However, this depends on self-regulation (Lodge et al., 2023); students must become “critical prompters” (P. #3), fostering active participation (Crompton & Burke, 2023).

3. Cognitive Offloading and the Writing Crisis

The counterpart is “cognitive offloading” (Dwivedi et al., 2023). Professors #5 and #26 warn that GenAI can lead to superficial learning. There is an imminent risk of a rupture between writing and thinking; by delegating synthesis, intellectual authorship and the “desirable difficulty” of learning are lost (Bearman et al., 2024).

4. Information Management and Higher Cognitive Functions.

P. #40 highlights that GenAI allows students to handle large volumes of data, enabling them to focus on reasoning and critique. By optimizing bibliographic searches (P. #38) and independent work hours (P. #42), there is a promise of greater intellectual depth, provided technology serves to elevate rather than atrophy cognitive functions.

1. The Authorship Crisis and Next-Generation Plagiarism.

P. #14 warns about the substitution of personal work. P. #5 emphasizes that students perceive these tools as a way to complete activities without effort. This “Authorship Crisis 2.0” (Moorhouse et al., 2023) renders current detection tools insufficient, requiring a shift toward ethical education over technical surveillance.

2. Hallucinations, Veracity, and Algorithmic Bias.

P. #10 stresses the need to verify data, while P. #33 highlights the teacher’s role in exposing AI biases. In Latin America, this is particularly sensitive due to “cultural bias”; P. #43 alerts to unfair decisions based on models trained under Global North worldviews (Katz & Shifman, 2024).

3. Privacy, Data Governance, and Dehumanization.

P. #8 expresses concern over student data privacy and the lack of transparency in automated processes. There is a latent fear of mass data collection (P. #35) and a loss of human interaction; P. #38 values face-to-face encounters as irreplaceable, fearing the “transactionalization” of education (Selwyn, 2024).

4. Management Strategies: From Prohibition to Critical Literacy.

P. #11 argues for reflexive and ethical integration. P. #45 summarizes the ideal stance: instilling ethics and responsibility, transforming risk into an opportunity to discuss intellectual property and critical thinking. Teachers must transition from “plagiarism police” to “architects of integrity” (UNESCO, 2024).

5. Regional Comparative Analysis: Ethical Dimension

In Chile and Colombia, the approach is proactive. Colombian (P. #1) and Chilean (P. #3) educators link ethics to researcher responsibility, fearing cognitive dependency and the substitution of personal work (P. #14).

In Peru and El Salvador, the focus is on “veracity vigilance.” Peruvian educators (P. #10, P. #6) are vocal about system hallucinations. In El Salvador, the lack of teacher ethical training is seen as a critical risk (P. #17).

In Mexico and Venezuela, social justice and the human bond are paramount. Mexican participants (P. #25, P. #43) alert to algorithmic bias perpetuating racial or cultural discrimination. In Venezuela, the greatest danger is dehumanization and the loss of interpersonal contact (P. #38).