AI in Education: What El Salvador’s School Pilot Can Teach Us

Artificial intelligence in education is often presented as a choice between two extremes.

Embrace AI and transform the classroom.

Or keep AI away from students before it replaces thinking, reading and teaching.

But that may be the wrong question.

A recent education pilot in El Salvador, together with new findings from PISA 2025, points toward a more useful conversation.

The question is not simply whether schools should use AI.

It is:

How can AI help teachers create better learning opportunities for students?

In September 2026, President Nayib Bukele drew international attention to results from 171 public schools participating in an education modernization program that includes AI supported tutoring.

The participating students performed above El Salvador's national average in reading, mathematics and science, with results comparable to averages reported for countries including Germany and Sweden.

Those results do not prove that AI caused the difference.

But they do make the program worth looking at.

Because the most interesting part may not be one particular technology.

It is the combination of better tools, better access to materials, teacher support and more opportunities for students to learn.

What happened in El Salvador?

The results came from a PISA for Schools assessment involving 1,198 students across 171 public schools participating in El Salvador's education modernization program.

The program includes AI tutoring, but AI is only one part of a broader effort.

It also includes teacher and school leader support, educational materials, monitoring, infrastructure, technology and connectivity.

Bukele said that the program has since expanded to more than 1,000 public schools, with plans to expand it further.

The results attracted attention because students in the participating schools performed strongly in reading, mathematics and science.

There is an important limitation.

The 171 schools were participating in the program and were not selected as a statistically representative sample of every public school in El Salvador.

That means the results should not be interpreted as proof that every school in the country suddenly performs at the same level.

They also cannot tell us which individual part of the program made the biggest difference.

But that does not make the results less relevant.

It shows why educational technology works best when it is part of a broader instructional approach.

Technology alone is rarely the whole story.

Teachers, materials, practice, feedback and school support still matter.

How AI can support better learning

AI can make learning more accessible, more relevant and easier to adapt to individual students.

For teachers, it can make it possible to create learning materials that would otherwise take much more time to prepare.

A teacher might use AI to:

• create a reading passage around a class topic
• adapt a text to a different reading level
• generate additional comprehension practice
• create vocabulary activities
• translate or adapt material for multilingual learners
• prepare different versions of the same lesson
• create extra practice for a student who needs it
• connect reading practice to a student's interests

That can make differentiation much more practical.

Instead of every student needing to work from exactly the same text, teachers can create multiple entry points into the same topic.

A student who needs more support can receive a more accessible version.

Another student can work with more complex vocabulary or deeper questions.

The learning goal stays the same.

The route toward it can change.

What PISA 2025 adds to the discussion

PISA 2025 also offers useful insight into how students are already using AI.

Students may use AI to research a topic, ask questions, get feedback, summarize information, explain difficult concepts or help with assignments.

These activities are not all the same.

The important question is what the student is actually doing during the learning process.

PISA findings suggest that students who used AI to generate text for writing assignments tended to perform less well on average.

At the same time, frequent AI users performed better when they had learned how to evaluate the quality of AI generated information and used it as a learning tool.

That distinction is important.

AI can support learning when it helps students investigate, practice, receive feedback, compare information or explore ideas.

It can also help teachers provide more relevant practice and support.

In other words:

The educational value of AI depends on what students are asked to do with it.

Good learning still requires active participation

Whether a student works on paper or on a screen, learning still requires active participation.

Students need opportunities to:

• read
• recall information
• make connections
• infer
• compare
• question
• explain
• discuss
• revise their thinking
• practice again

AI can help create more of those opportunities.

A teacher might generate a passage at several reading levels and let students work independently.

A student might complete an online comprehension quiz and receive immediate feedback.

A small group might discuss the same text together.

Another student might practice vocabulary through a game.

The medium can differ.

The learning goal remains central.

Digital learning can still be active learning

Screen time is often discussed as if every digital activity were the same.

It is not.

Reading a passage and answering comprehension questions is different from passively scrolling through unrelated content.

Playing a vocabulary game is different from watching videos without an instructional purpose.

Comparing information from different sources is different from copying an answer.

A well designed digital activity can require sustained attention, retrieval, reasoning and decision making.

And a printed worksheet can also be ineffective if the activity itself does not require meaningful thinking.

The format alone does not determine the quality of the lesson.

The instructional design does. That gives teachers more flexibility.

A teacher can print a passage for close reading. Another class can read the same passage online.

Students can complete a quiz, work in pairs or practice vocabulary through a game.

Teachers can choose the format that best fits the learning goal, the student and the moment.

AI can make differentiation more practical

One of the strongest opportunities for AI in education is differentiation.

Teachers already know that students do not all read at the same level, learn at the same speed or need the same amount of practice.

The challenge has always been time.

Creating several versions of the same activity manually can take hours.

AI can dramatically reduce that preparation time.

A teacher can start with one topic and create:

• an accessible reading version
• a more challenging version
• additional vocabulary practice
• comprehension questions
• a translated version
• an online quiz
• a printable activity

That makes it easier to give students work that is closer to what they actually need.

AI does not replace the teacher's judgment.

It helps make that judgment easier to act on.

What could responsible AI use look like in a classroom?

For reading instruction, a practical workflow might look like this.

1. Start with the learning goal

Decide what students need to practice.

Maybe it is identifying the main idea.

Maybe it is making an inference.

Maybe students need to build background knowledge before starting a new science or history unit.

Maybe a student needs extra vocabulary support. The learning goal comes first. The technology follows.

2. Create or adapt the material

AI can help create a text around a relevant topic or adapt existing material to a different level.

This can be especially useful when several students need to work on the same topic at different levels.

It can also make it easier to connect reading practice to topics students already care about.

3. Review the output

AI generated educational content still needs professional judgment.

Check the facts. Check the reading level. Check the questions. Check whether the activity actually practices the skill you intended.

The teacher remains the instructional decision maker.

4. Choose the right format

Some lessons work best on paper. Others work well online. Some students benefit from immediate quiz feedback.

Others need a printed passage they can annotate. Some activities work best independently. Others work better in a small group.

There is no single correct format.

The teacher chooses the format that supports the goal.

5. Keep students actively involved

The most important question is not whether AI was used somewhere in the process.

It is whether students are actively learning. Are they reading? Thinking? Explaining?

If the answer is yes, technology can become a powerful part of the learning environment.

AI can also become a topic students investigate

AI does not only have to be a tool.

It can also become something students read and think about.

For example, PicoBuddy's 7th Grade report The Digital Tutor: How Artificial Intelligence Is Transforming Modern Classrooms explores personalized learning, accessibility, teacher support, AI literacy, bias and the changing role of technology in schools.

It comes with comprehension practice and a printable version.

Teachers can also adapt the passage to another reading level or create a version in English or Spanish.

That opens up several possibilities. Students can read the passage online.

A teacher can print it for close reading. A class can discuss the benefits and challenges of AI.

Students can evaluate different arguments. They can compare viewpoints. They can decide what they think.

The technology helps create the opportunity. The learning comes from what students do with it.

The question is not whether schools will use AI

Bukele has argued that the AI age is already here.

On that basic point, it is difficult to disagree.

AI is already part of students' lives, teachers' workflows and the wider world children are preparing to enter.

The more useful question is how schools choose to use it.

AI can help teachers create more relevant materials. It can make differentiation easier. It can give students more opportunities to practice.

It can support feedback, accessibility and multilingual learning. It can help turn one topic into several different learning experiences.

The strongest use of AI in education is not about replacing teachers or replacing learning.

It is about expanding what teachers and students are able to do.

Use AI to create better opportunities to read, practice, think and learn.