Brazil has officially entered a decade in which education and technology are no longer an “optional cross-cutting theme” but are now guiding decisions regarding curriculum and administration. This shift is already taking shape in regulatory frameworks such as CNE/CEB Resolution No. 2/2025, which establishes National Operational Guidelines for the use of digital devices in school settings and for the curricular integration of digital and media education throughout Basic Education.

Meanwhile, as part of the international discussion, UNESCO has published the AI Competency Framework for Students, which supports the integration of AI learning objectives into official curricula and defines 12 competencies across four dimensions (human-centered mindset, AI ethics, techniques and applications, and AI system design) with progression across three levels (Understand, Apply, and Create).

Here, this move comes as no surprise; it is merely confirmation. Because when we talk about the National Common Core Curriculum (BNCC) for Computing, we know that the three pillars (Computational Thinking, the Digital World, and Digital Culture) form the foundation. But we also know that Brazilian schools need to take it a step further: to treat Artificial Intelligence (AI) as a fourth pillar—one that is explicit, intentional, and structured.

On Start, we understand that this movement is not just about “including content,” but about taking a close look at a structural change that requires us to rethink how these pillars and new content are taught.

According to the new National Operational Guidelines (CNE/CEB Resolution No. 2/2025), the integration of digital and media education must be intentional, balancing the use of technology with a focus on the teaching-and-learning process.

As the Computacional.com.br website emphasizes, Artificial Intelligence is not just a tool, but a component that requires teachers to master new digital knowledge and develop specific skills—such as understanding algorithms and machine training—so that teachers can move beyond being mere users and become critical mediators.

By elevating AI to the status of a fourth pillar, we emphasize that educators must reposition themselves in light of a curriculum that now prioritizes student autonomy and ethical problem-solving in a data-driven world.

What has changed in terms of general behavior and usage by everyone: AI is no longer just a “tool” but has become an “agent”

The National Common Core Curriculum (BNCC) for Computer Science provides us with a roadmap. However, AI has changed the landscape. What used to mean “using software” now means working with systems capable of taking the initiative, carrying out actions, and producing complex outputs with human supervision.

David Heinemeier Hansson describes this shift well: agents moved beyond simply “responding” to acting autonomously, using tools (searching for documentation, running tests, operating services) and delivering “production-grade” contributions within a supervised collaboration model.

With AI, the game has changed in computing: learning has become a partnership facilitated by technology. That’s why, in addition to asking good questions, we need to train students to learn with AI and learn about AI—that is, to know how to use it, but also to understand how it works, where it goes wrong, and what risks it poses.

In practice, this means developing skills in coordination and accountability: setting goals, providing context, verifying evidence, reviewing results, and taking ownership. It is precisely because it cuts across Computational Thinking, the Digital World, and Digital Culture that AI stands as an integrated (rather than “add-on”) pillar within the BNCC for Computing.

The educational risk is real: AI can improve performance but hinder learning

If schools don't teach students how to use AI, students learn on their own—and they usually learn in the most dangerous way: by taking shortcuts, simply looking for the final answer without recognizing what's happening or understanding the process that led to the solution.

The most important point here is the difference between task performance and actual learning. The OECD Digital Education Outlook 2026 emphasizes that using GenAI without a pedagogical purpose can create an “illusion of learning”: it improves immediate results but does not solidify the skill.

And the numbers help drive this point home: in studies summarized in the report, there are scenarios in which AI boosts performance during use (e.g., +127% in a tutoring format), and when access is withdrawn, there is a decline (-17%) compared to the control group—the so-called “crutch effect.”

The turning point for Basic Education is this:

Teaching AI isn’t about teaching students to “answer faster.” It’s about teaching them to think better, even when the answer is just a click away. Teaching with AI must be intentional. In other words, AI is often not the END, but rather the MEANS to improve learning.

Its use must be evaluated and justified. A smart tutor can provide analogies and examples in different contexts and explain a mistake. In contrast, the student receives a ready-made answer without the critical thinking skills needed to understand how it was derived.

Why AI Deserves Its Own Pillar in the BNCC for Computer Science

AI deserves its own pillar because it:

• It enhances computational thinking (patterns, data, algorithms) but can also discourage critical thinking if used as a crutch. Therefore, the pedagogical focus needs to shift from “using AI to arrive at the answer” to learning with AI and learning about AI. As a core principle, Start has always understood that AI cuts across what the BNCC calls Computational Thinking: it does not replace this skill; rather, it demands even greater clarity of reasoning, argumentation, and validation.
   
• It reshapes the digital world (models, data, privacy, automation, biases) because students go beyond simply “consuming information” and begin to interact with systems that generate, summarize, and make decisions based on data. This changes the concept of authorship and trust: students must learn to cite sources, recognize limitations and biases, protect personal data, and validate results. In other words, the core competency becomes using AI judiciously and responsibly, not just quickly.
   
• Critically analyze whether AI reproduces social patterns and prejudices that are being or have been overcome. AI learns from large volumes of data produced by people (texts, images, and records) that reflect worldviews from different eras and contexts. As a result, many responses may reflect patterns and opinions from the past, as well as biases present in that data.

In practice, treating AI as a pillar means recognizing that it is already part of the “air” we breathe in the digital world and that schools need to teach criteria, not just tools.

From “AI as a solution” to “AI as an agent”: the digital literacy schools need

If agents are already able to perform tasks independently using tools (with humans reviewing and guiding them), AI literacy must include three core competencies:

• Smart Specification (briefing and criteria): This is what makes AI shine when it comes to organizing ideas or creating an initial draft. Define the objective, constraints, target audience, acceptable evidence, evaluation criteria, and “what’s off-limits.” Smart Specification (briefing and criteria) — define the objective, constraints, target audience, acceptable evidence, evaluation criteria, and “what’s off-limits.”
   
• Validation and auditing: this involves checking sources, testing and comparing answers to identify misinformation (media literacy), recognizing deepfakes, identifying biases, and correcting hallucinations.
   
• Authorship and accountability: documenting decisions, citing references, explaining the reasoning, and taking responsibility for the final result.

This transforms the student from a “prompt user” into a critical orchestrator—exactly the logic of “supervised collaboration” described by DHH: autonomy with human governance.

How This Is Put into Practice in Basic Education (With an Emphasis on Ethics from an Early Age)

When Start proposes AI as the “fourth pillar,” it doesn’t mean starting with technical terms. It means starting with a functional understanding:

• In the early years: explore pattern recognition (the foundation of computational thinking and how AI works) and discuss “how computers learn from examples.”
   
• From elementary through high school, the goal is to move students away from the role of passive consumers and develop them into critical and ethical creators. This involves understanding data, biases, and privacy, as well as practicing fact-checking, comparing answers, and responsible authorship.
   
• Always with integrity: responsible use, transparency in the process, and the judgment to decide when not to use AI.

For Effective Scaling: The Teacher at the Heart of Transformation

The PNE also emphasizes the importance of teacher training, and this is the key to its success. Start is already operating on a large scale and with a proven approach: 3 million students, 14,000 teachers, and 7,000 schools, combining ongoing training and monitoring.

Here, diagnostic tools and professional development pathways help educators move beyond generic approaches and toward practices that make a real difference. The CIEB—Center for Innovation in Brazilian Education—for example, maintains a matrix and a self-assessment tool for teachers’ digital competencies to support school networks in advancing the pedagogical use of technology.

This is how AI becomes a “pillar”: not because of the tool itself, but because of an ecosystem that positions itself at the forefront of this need. For us, it’s an educational choice. What we’re saying by advocating for AI as the fourth pillar is simple, yet ambitious:

• Schools cannot outsource thinking to technology.
• AI can be a catalyst for learning or a crutch that weakens skills.
• AI literacy is now intentionally included as part of basic digital literacy.

The PNE sets the direction: critical, reflective, and ethical digital education. The Start is committed to helping school districts and schools achieve this goal, proposing that Artificial Intelligence be treated not as an “extra” subject, but as a foundational pillar of an education that prepares young people to use tools and, above all, to think for themselves.

*Content is the sole responsibility of the advertiser.
**This text does not necessarily reflect the opinion of Bett Brasil.

Sources:

News: New Education Plan Sets Goals for Brazilian Education Through 2034 - News - Chamber of Deputies Website - https://www.camara.leg.br/noticias/1077593-novo-plano-de-educacao-institui-metas-para-a-educacao-brasileira-ate-2034 (Accessed February 20, 2026)

BNCC Guide to Computing Without Secrets - 2025 - Published by Start By Alura

Article: OECD Digital Education - Outlook 2026 - EXPLORING EFFECTIVE USES OF GENERATIVE AI IN EDUCATION - https://www.oecd.org/en/publications/oecd-digital-education-outlook-2026_062a7394-en.html (Accessed January 29, 2026)

Report: Promoting AI Agents - https://world.hey.com/dhh/promoting-ai-agents-3ee04945 (Accessed February 24, 2026) / AI Competency Framework for Students - https://www.unesco.org/pt/articles/marco-referencial-de-competencias-em-ia-para-estudantes (Accessed February 25, 2026)

CNE/CEB RESOLUTION No. 2, OF MARCH 21, 2025 - Establishes the National Operational Guidelines on the use of digital devices in school settings and the integration of digital and media education into the curriculum - https://www.gov.br/mec/pt-br/cne/2025/marco/rceb002_25.pdf Articles and content consulted: www.computacional.com.br

Share on social media: