Society & Ethics

AI Reaches Latin American Classrooms, but Preparedness Is Unclear

A TALIS 2024 finding shows 55% of surveyed Chilean lower secondary teachers reported AI use, while regional systems still face training, privacy, access and evidence gaps.

AI Is Entering Latin American Classrooms: Are We Prepared?
AI Is Entering Latin American Classrooms: Are We Prepared?AI-generated
By James Calloway6 min read

Why it matters

  • TALIS 2024 results cited by Fundación Santillana found 55% of surveyed Chilean lower secondary teachers had used AI in their work.
  • In TALIS 2024, Chilean teachers cited lesson planning and creating student-feedback texts as uses of AI.
  • Colombia's IDEP has provided AI-related teacher training since 2023.
  • Escola Estadual Santo Afonso participated in the Jemas pilot during 2026.
  • OEI's Tamara Diaz identified two learning risks: the convenience trap and the illusion of competence.

By James Calloway

According to TALIS 2024 results cited by Fundación Santillana, 55% of Chilean lower secondary education teachers surveyed said they had used artificial intelligence (AI) in their work. Their reported uses included lesson planning and creating feedback texts for students. The figure puts AI inside professional teaching work, but it does not show whether those uses improved learning.

“We are experiencing a time of great bewilderment; not only in Latin America, but in the whole world,” said Tamara Diaz, Director General of Education at the Organization of Ibero-American Organizations (OEI), in an interview with Unite AI. She said education systems are still trying to understand how to incorporate the technology.

Diaz said schools already use AI to prepare classes, create educational materials and support students with learning difficulties, although implementation remains limited. She considers clear public policies, evaluation of results, and teacher training and support necessary to capture the technology’s potential. Her framing makes teacher preparation, evaluation and public rules central to any rollout.

What does the 55% figure establish?

The TALIS 2024 finding measures reported professional use among Chilean lower secondary teachers. The Fundación Santillana citation does not disclose the survey sample size, frequency of use, named tools, or classroom outcomes. Those omissions leave adoption measurable but effectiveness unproven.

Lesson planning and feedback text show where AI enters the instructional workflow. The cited material gives no error rate for generated content, no measure of teacher review, and no comparison with classrooms that do not use AI. A high adoption share therefore cannot stand in for evidence of learning gains.

Chile provides evidence of use, not proof of benefit. Takeaway: The 55% figure establishes substantial teacher adoption, while its educational value remains unmeasured in the material cited.

How are schools putting AI to work?

Fundación Santillana de Chile shared with Unite AI the case of Deina Sepúlveda, an English teacher at Liceo Agrícola El Carmen de San Fernando, a technical high school. Sepúlveda used AI to develop a web page for English pronunciation practice. The page lets students hear words and texts and adjust audio playback speed; Educarchile later published her experience.

Ricardo F. Alvares, CEO of IA Instituto, told Unite.AI that AI can support personalized teaching, educational inclusion and teacher preparation. At Escola Estadual Santo Afonso in Belo Horizonte, the school joined the 2026 Jemas pilot, or Jornadas Educativas en el Mundo del Alumno. Teachers’ knowledge and AI tools were combined to prepare classes adapted to students’ interests, experiences and contexts.

Alvares’s account names no AI systems or publishes student-performance measures. At EMEF do CEU Pêra Marmelo, a São Paulo public school, eighth-graders participated in AI-supported mathematics activities. The University of São Paulo’s Institute of Mathematics and Statistics helped cover algebra and problem-solving, with the goal of strengthening mathematics and fostering a critical attitude toward AI use.

These projects use AI to organize preparation or support practice while teachers remain part of the design. Takeaway: Latin American schools are testing practical classroom workflows, but the cited examples are demonstrations, not evaluated interventions.

Are teachers and systems ready?

The OEI says institutions and specialists across several Latin American countries are exploring AI’s opportunities and risks. It frames the main questions as teacher preparation, student privacy, educational inequality and critical thinking, meaning the ability to evaluate claims and reasoning.

Diaz says the principal obstacle is attaching tools to clear pedagogical objectives, meaning specific teaching goals, rather than novelty. She again places teacher training and students’ critical thinking at the center of any implementation strategy.

Colombia’s Institute for Educational Research and Pedagogical Development (IDEP) has provided AI-related teacher training since 2023. A spokesperson told Unite AI that teachers show interest in practical uses and concern about ethical implementation. IDEP says it still lacks enough systematic evidence to determine how teachers apply that training in classrooms.

In Costa Rica, the Omar Dengo Foundation issued a written statement to Unite.AI warning against exclusive dependence on AI. The foundation said schools must protect how students develop their own abilities, strengthen critical thinking, and teach students to evaluate AI answers and understand its limitations. It presented the challenge as learning with AI rather than outsourcing learning to it.

These accounts point to three infrastructure gaps: teacher preparation, protection of student data, and unequal access to technology. The OEI also calls for guidelines and educational strategies that account for different realities across schools. None of the cited programs supplies regional measurements of privacy failures or access gaps.

Teacher interest and isolated pilots have not resolved those gaps. Takeaway: The region has experiments and training programs, but not yet the evidence base to judge responsible adoption across different schools.

What can students mistake for learning?

Diaz identifies two ways AI can interfere with learning. The “convenience trap” describes students seeking immediate answers without investing the intellectual effort required to learn. Unite AI provides no prevalence estimate for that behavior, so the risk is conceptual rather than quantified.

The “illusion of competence” occurs when students believe an explanation from a tool means they have acquired the underlying knowledge. Diaz says both risks confuse access to information with actual learning.

Her warning does not reject classroom AI. It makes independent reasoning and student autonomy, meaning the capacity to direct one’s own learning, central conditions of responsible use.

The Costa Rica foundation’s formulation offers a practical test: students should evaluate answers and understand limitations, not simply accept output. The source provides no classroom data on whether students retain knowledge without AI or verify its answers. It also does not measure the privacy or equity effects of the cited programs.

AI can therefore expand access to explanations while obscuring who did the thinking. Takeaway: The central educational danger is not an answer appearing on screen, but students mistaking that answer for knowledge they have earned.

What determines the next phase?

Unite AI’s regional assessment is uneven. Some institutions experiment with tools that create new learning methods, while others still face technology-access and teacher-preparation difficulties. That split complicates any regional claim about readiness.

The OEI’s forward-looking condition is specific: access to AI matters, but integration must not sideline critical thinking, student autonomy or teachers’ central role. Diaz’s recommendations remain clear public policies, evaluation of results, and sustained training and support. The unresolved question is whether education systems can turn those principles into documented classroom practice.

For now, the most credible signal is adoption rather than achievement. Takeaway: The next test is whether schools can document classroom application, protect student data and preserve independent learning while expanding teacher access to useful tools.

Source: unite.ai

Share this article:

More from James Calloway

James Calloway

Show full bio

News editor covering people, launches and day-to-day practice in artificial intelligence and machine learning at AI In Context.

722 articles

Related articles

  1. 79% of Students Use AI for Homework, Only 47% Get Guidance: OpenAI Flags Education 'Guidance Gap'
  2. OpenAI Partners With CodeAI to Train the First AI-Native Students
  3. 81% of Australian students use AI weekly as universities respond
  4. A Former Professor Estimates 90% of His Online Students Used AI for Coursework
  5. Google's Gemini Tutor Boosted Sierra Leone Math Scores 0.258 SD

« Previous articleNext article »