UNESCO Roundtable: AI in Education is Public Choice

This position underscores the belief that education is a human right, and AI's role within it must align with that foundational principle.

The ADG for Education highlighted that this collective decision reflects a unified global perspective on the ethical governance of AI in learning environments. The discussions focused on how to navigate the complexities of AI to ensure it enhances, rather than detracts from, educational equity and quality.

These priorities aim to ensure that AI technologies serve the broader societal good and educational objectives.

To achieve this, UNESCO's 2021 Recommendation on the Ethics of Artificial Intelligence serves as a foundational guideline, with its Readiness Assessment Methodology (RAM) now being implemented in 80 countries. This methodology helps nations evaluate their preparedness for ethical AI adoption. Further, the organization launched new publications, 'Governing AI in Education' and 'The Algorithm in the Room,' to provide practical guidance for policymakers and educators.

The International Telecommunication Union (ITU) issued a stark warning about the potential for an AI divide, emphasizing that the current connectivity gap could exacerbate inequalities. With 2.2 billion people globally still lacking internet access, the ITU stresses that today's digital exclusion must not become tomorrow's AI exclusion, hindering access to educational opportunities.

The ITU's strategy for progress rests on three pillars: connectivity, capacity, and trust. Efforts include the 'Giga' initiative, which has already mapped 2.3 million schools to identify and address connectivity needs, aiming to ensure equitable access to digital resources and AI education. This integrated approach seeks to build a robust foundation for inclusive AI adoption.

Mandated through the Global Digital Compact, this panel aims to provide comprehensive, expert-driven insights to inform global AI governance.

The panel comprises 40 experts from diverse disciplines, reflecting the multifaceted nature of AI. Its first report underscored the rapid pace of AI development, noting the difficulty even for specialists to keep abreast of changes. This highlights the critical need for continuous, multidisciplinary research to guide policymakers effectively.

This platform ensures equal representation for all UN member states, promoting inclusive discussions on critical AI policy issues.

The dialogue aims to prevent AI governance from being shaped by a limited number of actors, ensuring a global, equitable approach. Participants are encouraged to attend to contribute to shaping international norms and standards for AI, fostering a future where AI benefits all nations.

Finland implemented a national policy from August 2025 restricting the use of mobile devices, including phones and other digital tools, in classrooms. This measure aims to improve student concentration and reduce distractions during lessons. Schools retain autonomy to manage device use during breaks, balancing policy with practical needs.

Teachers also maintain authority over the pedagogical use of devices, allowing for flexibility when technology genuinely enhances learning. This nuanced approach ensures that restrictions support, rather than hinder, effective educational practices.

Students demonstrated better focus and concentration during lessons, indicating the policy's effectiveness in mitigating digital distractions. The initiative also saw increased peer-to-peer social interaction during breaks as students engaged in play and communication rather than screen time.

Parents expressed strong support for the policy, recognizing the benefits for their children's learning and social development. These initial results suggest that carefully implemented restrictions can foster a more conducive learning environment and healthier social dynamics among students.

This approach prioritizes human interaction and critical thinking over technological dependence, ensuring AI acts as a supplementary tool rather than a replacement for human educators.

In early stages, the focus is on introducing concepts like patterns and distinguishing between machine and human capabilities. As students advance, the curriculum delves into more complex topics such as data ethics and algorithm design, preparing them to be responsible digital citizens. A pilot study conducted in core subjects revealed improved student engagement following a reduction in device usage, reinforcing the human-centric philosophy.

This strategy emphasizes that AI should enhance learning outcomes by supporting educators and students, rather than automating the entire educational process. The UAE's framework seeks to cultivate a balanced environment where technology complements, but does not overshadow, human pedagogical elements.

AI skills refer to the ability to utilize and develop AI technologies, while AI literacy pertains to understanding AI's concepts, implications, and ethical considerations. This comprehensive approach aims to equip individuals with the necessary competencies to engage with AI responsibly.

The Finnish government has demonstrated its commitment by investing in strategic research on AI usage from 2020 to 2026. This long-term investment supports the integration of AI into the educational system and the broader society, ensuring that its citizens are well-prepared for an AI-driven future.

The UAE emphasized a clear distinction between scaling AI and achieving genuine educational impact, stating that novelty must never be mistaken for impact. The nation's vision is that automation is not an end in itself; true success is measured by demonstrable improvements in learning outcomes. This perspective guides their iterative implementation process, which is continually informed by feedback from teachers and learners.

To ensure true impact, the UAE distinguishes between 'gain' and 'evidence.' The key metric for evaluation is whether learning gaps are narrowing and if students are reasoning more deeply. This rigorous approach prevents superficial adoption of AI and prioritizes measurable benefits for student development, ensuring that technology serves pedagogical goals.

Students frequently use AI tools to complete tasks without genuinely engaging in the learning process, merely fulfilling requirements rather than acquiring knowledge. A language teacher observed that students utilized AI to write entire essays, bypassing the crucial brainstorming and drafting stages meant to develop their writing skills. This phenomenon highlights a critical challenge in AI integration.

Educators must critically assess whether AI tools foster independent thinking or inadvertently lead to greater student dependency. The concern is that while the tool may complete the task, it does not necessarily support the learning journey. A strong call was made for involving teachers in the testing and evaluation of AI before its widespread adoption, ensuring these tools are pedagogically sound and supportive of genuine learning outcomes.

For ministries of education to govern AI effectively, they require both deep insight into current developments and robust safety protocols. Understanding what is happening in the rapidly evolving AI landscape is crucial for informed decision-making. This includes comprehending the technological advancements, their applications, and their potential impacts on educational systems.

Equally important is ensuring the safety of students and educational data. Decisions must be evidence-informed to guarantee that AI serves the public interest, protecting against potential harms while maximizing benefits. A clear vision of where AI integration is headed is essential to proactively shape its trajectory rather than reactively respond to challenges.

Shingai Manangua likened the current state of artificial intelligence to the 'exposed copper wire phase' of electricity's early adoption. This analogy highlights that, similar to how early electrical systems lacked safety standards and led to numerous accidents, AI is currently in a chaotic, unregulated phase.

Historically, powerful and dangerous technologies, from electricity to aviation, eventually developed established norms and safety protocols that allowed society to live with them safely. The goal for AI is to transition from its current nascent state, characterized by significant risks, to a mature stage with comprehensive governance and safety mechanisms.

The choice of which AI tool to use is a strategic decision, fundamentally tied to national sovereignty. As AI becomes critical infrastructure in sectors like healthcare, education, and finance, the power to decide its deployment and management becomes paramount. The example of Anthropic's 'Fable 5' being withdrawn due to U.S. export controls illustrates how external dependencies can compromise national systems.

This incident underscores the vulnerability of countries relying on foreign-controlled AI, raising questions about who dictates when such critical tools are pulled or modified. There is a pressing need for developing local and regional technical expertise to 'decolonize' and fine-tune AI models. This autonomy ensures that AI infrastructure aligns with national interests and ethical standards, preventing external forces from dictating its availability or functionality.

When an AI-related incident occurs, robust governance should define clear lines of responsibility, identify the responsible system, and enable its swift offline removal. This level of control is crucial for managing powerful and potentially dangerous technologies.

The analogy to aviation safety highlights the need for comprehensive investigations, binding findings, and the courage to 'ground a fleet' when necessary, as demonstrated by the Canadian incident involving AI interaction leading to tragic loss of life. This framework ensures that AI tools are systematically identified, halted, and regulated to prevent harm, prioritizing public safety above all.

AI-powered search engines and algorithms often prioritize visibility, surfacing results based on complex computations rather than historical accuracy or provenance. These systems determine what information rises to the top, potentially overlooking crucial historical truths and diverse perspectives. This algorithmic selection can lead to homogeneous or biased information landscapes, raising concerns about the integrity of knowledge dissemination.

Unlike traditional repositories of history, current AI data sources may not be curated for authenticity and provenance. This deficiency highlights a critical need for scrutiny regarding how AI systems source and present information. Ensuring the authenticity and reliable origin of data is essential to counteract algorithmic bias and uphold the veracity of historical and contemporary information, particularly as AI becomes an increasingly influential source of knowledge.

The prevailing narrative that AI adoption is universal, inevitable, and requires rapid implementation is beginning to shift. This perspective, once widely accepted, is now being questioned by public authorities, technologists, scientists, and emerging evidence. This skepticism reflects a growing recognition that AI deployment is not a predetermined path but rather a conscious decision with far-reaching implications.

In this period of questioning, an older idea is resurfacing: that education is a common good, and AI is a choice, not an inevitability. This emphasizes that societies have the agency to shape how AI integrates into education and other sectors. The shift underscores the importance of public accountability in AI deployment, moving away from a technology-driven imperative toward a human-centric approach that prioritizes societal values and collective well-being.

A fundamental recognition emerging from various case studies, including those from the Netherlands, Brazil, Africa, and Argentina, is that access to AI is not synonymous with control over it. For an education system, this distinction marks the difference between genuine choice in technology adoption and a structural dependence on external providers. True AI sovereignty encompasses multiple dimensions.

These dimensions include jurisdictional control over data and governance, technological independence in developing and maintaining AI, fiscal autonomy in procurement, cultural alignment of AI with local values, and pedagogical suitability for educational contexts. Without robust control across these areas, education systems risk becoming reliant on AI solutions that may not align with their unique needs or values, potentially compromising their autonomy and long-term educational goals.

Procurement, often seen as a one-off and mundane event, is being re-evaluated as a critical, yet overlooked, lever for AI governance in education. Unlike traditional textbooks, AI tools are dynamic and change over time, necessitating continuous oversight throughout their entire lifecycle. This shift requires a new perspective on how schools acquire and manage technology.

Effective governance must extend beyond initial acquisition to include ongoing monitoring and adaptation as AI tools evolve. Integrating public interest conditions into procurement processes ensures that AI solutions align with educational goals and ethical standards. This continuous oversight transforms procurement into a powerful mechanism for shaping how AI integrates into schools, moving beyond a transactional event to a strategic instrument for long-term governance.

Educators and institutions cannot simply attribute mistakes or inappropriate grades to an AI that 'ran away' or acted independently. Clear accountability for all AI-driven decisions must be maintained, as the ultimate responsibility rests with the human entities deploying these tools.

This principle is crucial given the inherent power imbalance between AI providers and students. The right to explanation, detailing how AI decisions are made, becomes a critical entitlement for students and their families. Transparent accountability frameworks ensure that educational institutions remain answerable for the integrity and fairness of AI-integrated systems, preventing the erosion of trust and safeguarding student welfare.

Developmental science indicates that children's cognitive and emotional development varies widely, making rigid age-based milestones unreliable indicators of AI appropriateness.

Children globally are already using AI unsupervised, challenging the practicality and effectiveness of arbitrary age cutoffs. This reality suggests that policy should focus more on adaptive guidelines and education about responsible AI use rather than blanket restrictions. A nuanced approach, sensitive to individual developmental stages and existing usage patterns, may be more effective in guiding safe and beneficial AI integration for younger users.

Schools face significant commercial risks due to the unannounced integration of AI into their management systems, often without full transparency about data practices. A key concern is how schools can ascertain if children's data is being harvested to improve commercial products when students engage with AI tutors or utilize automated content summaries. This lack of transparency can lead to ethical and privacy breaches.

Many current educational AI solutions are primarily designed as commercial products rather than solely pedagogical tools. This commercial orientation means that improving the product, which often involves data collection, may supersede educational objectives. Schools must therefore be vigilant about the terms of service and data privacy policies associated with AI tools, ensuring that student data is protected and that technology serves educational goals first.

There is currently no definitive evidence demonstrating that AI is unequivocally superior to traditional human instruction in all educational circumstances. While AI may offer supportive roles, its comprehensive impact on a child's holistic development and the overall school environment remains largely unproven through rigorous, human-centric studies. This absence of broad evidence necessitates a cautious approach to AI integration.

Policymakers and educators are urged to consider the potential hidden costs of curriculum disruption and the erosion of human interaction. The call is to prioritize trusting teachers and their expertise while more comprehensive, context-specific evidence is gathered. This perspective advocates for evidence-based decision-making that thoroughly evaluates AI's benefits against its potential drawbacks, always keeping the child's well-being and the educational ecosystem at the forefront.