Declining Literacy Is Now a Core AI Governance Problem

Education’s foundational skills crisis has become its AI readiness crisis and the data is now impossible to ignore.

KEY TAKEAWAYS For Superintendents & District Leaders

• Your literacy and assessment practices may be misaligned with evidence-based benchmarks, creating downstream AI deployment risk.

• January 2026 compliance requirements require an execution layer most districts do not yet have.

For Ministers & System-Level Policymakers

• Three global reports reveal the same pattern: strong visions, weak implementation capacity.

• National indicators (credentials, attainment) mask declining adult skills that AI will expose at scale.

For Board Members & Governance Leaders

• AI systems already operating in schools require governance for GDPR, FERPA, COPPA, and EU AI Act obligations.

• Boards face fiduciary risk if 2026 compliance requirements are not reflected in budgets and oversight plans.

The Central Finding Education systems that cannot reliably implement evidence-based literacy instruction are now deploying AI systems that amplify both instructional failures and compliance gaps. This is not a technology issue; it is an execution-infrastructure crisis.

EU AI Act enforcement begins January 2, 2026. California’s rules hit the same week. Colorado follows in June.

PART 1 — THE FOUNDATION IS FAILING

Seventy percent of students in low- and middle-income countries cannot read a simple text after years of schooling. That figure comes from the World Bank’s October 30 literacy report (one of the largest syntheses of reading research ever produced). But the more revealing finding is this: after three years of instruction, 90 percent of students still cannot decode basic words or identify letter sounds.

The problem is not a lack of knowledge about effective teaching. The World Bank’s six-skill model is already globally validated. The gap is execution, not pedagogy.

Education leaders are not to blame for this. These are inherited structural constraints—systems built for a different era, now facing demands they were never designed to meet.

This matters for AI governance because reading is not "basic." Reading is the cognitive infrastructure for comprehension, reasoning, ethical judgment, and human-AI collaboration. If a system cannot reliably implement evidence-based reading instruction, it cannot safely deploy AI systems that depend on that same instructional capacity.

Simply put, you cannot teach ethical AI use if you cannot teach reading with fidelity.

PART 2 — THE MIDDLE IS COLLAPSING

Three weeks before the World Bank report, the OECD released Education at a Glance 2025. The data is equally stark: tertiary attainment is rising, adult literacy and numeracy skills are stagnating or declining, only 43 percent of bachelor’s entrants finish on time, and adult skills fell or flatlined between 2012 and 2023 despite record credentialing.

In advising ministries on workforce strategy, this is a "systemic mismatch signal"—the moment when official metrics diverge from actual capability and derail planning.

We built an education system designed for credentialing, stability, and standardized pathways. But the AI age requires comprehension, adaptability, and continuous reasoning.

The OECD data shows a widening credential–competence gap: we are reinforcing the top of the ladder while the base is eroding.

Why this matters for leaders

• For superintendents: Staffing, curriculum pacing, and PD cycles were built for a period when adult skill levels were stable. They are no longer stable.

• For ministers: National benchmarks are now decoupled from actual capability levels—a risk in international competitiveness.

• For boards: Fiduciary oversight cannot rely on attainment metrics that don’t measure the competencies AI will expose in real time.

When AI enters the system without governance infrastructure, it doesn’t fix the gap—AI amplifies execution failures at machine speed.

PART 3 — THE VISION OUTRUNS THE INFRASTRUCTURE

On November 7, the OECD released Education for Human Flourishing. The report calls on education systems to strengthen five human capabilities: adaptive problem solving, ethical and civic decision making, social intelligence and collaboration, meta-intelligence (evaluating AI outputs), and human agency in an AI-saturated environment.

These are worthy aims. The structural challenge, as I've seen in system-level planning across jurisdictions, is that systems optimized for credentialing cannot suddenly deliver capabilities they were never designed to produce.

The Human Flourishing model sits at the top of a capability ladder. The first two reports show that systems are struggling at the earliest rungs. The OECD adds another layer through the idea of a Moral Space—the zone where education must protect human security, dignity, and planetary limits. These collide with a simple reality: systems cannot implement basic literacy frameworks, cannot maintain adult skills, and are still expected to deliver flourishing pedagogy and deploy AI safely.

This is not a failure of leadership. These are inherited structural constraints within systems built for a different era. School systems now operate in an environment shaped by AI and global capability demands, and they lack the operational design to perform.

Without a governance layer, AI becomes an amplifier of every underlying execution gap.

PART 4 — THE LITERACY-TO-FLOURISHING CONTINUUM

In my work with districts, boards, and companies, I keep seeing the same pattern: vision documents exist in abundance; implementation capacity does not. These three reports make sense only when placed on a single pathway.

1. Systematic Decoding (World Bank) This is where the foundation breaks. Evidence-based reading instruction exists, but systems do not implement it consistently or with fidelity.

2. Deep Comprehension (OECD Education at a Glance) This is where the credential-to-competence gap becomes visible. Adults hold more degrees, yet literacy and numeracy skills are stagnant or declining.

3. Knowledge Building When systems focus on attainment rather than mastery, students move forward without the conceptual scaffolding that allows for deeper reasoning.

4. Ethical Reasoning Ethical instruction depends on comprehension and reasoning. You cannot teach digital ethics, civic judgment, or AI-related decision-making when foundational skills are weak.

5. Human-AI Partnership (OECD Human Flourishing) This is the destination described in the OECD report. It is not attainable if the earlier stages are not stable.

You cannot skip steps. You cannot move toward flourishing if the foundation of literacy is inconsistent.

When districts introduce AI systems that recommend learning pathways, shape assignments, or assess student work, these systems rely on human capability that may not exist yet. Without a governance framework, AI increases the speed and scale of existing failure patterns.

What this means for governance

• For superintendents: You need practical execution architecture, not new programs.

• For ministers: You need national indicators that focus on competence rather than credentials.

• For boards: You need oversight mechanisms aligned with the January 2026 AI compliance rules.

School systems do not fail because leaders lack vision. They fail because they lack execution infrastructure. The Continuum provides a way for leaders to explain the problem and a pathway for building the capacity to solve it.

A diagnostic for your system

Based on the Continuum, I've developed a one-page diagnostic framework that helps systems self-assess their execution capacity across all five stages. It is designed to be used before districts finalize their 2026 AI governance plans, and I make it freely available to education leaders. I’ll post it here on Linkedin or you can send me a message directly.

CLOSING THOUGHT

Education does not need more vision documents. It needs execution infrastructure that can connect evidence-based literacy, real human capability, and safe AI deployment.

 The three reports from the World Bank and OECD are not separate commentaries. They are diagnostic instruments that reveal the same structural weakness. The question for every system is simple: can your governance architecture deliver what your vision documents promise?

Compliance is not optional in January 2026. It is already on the doorstep.

ABOUT THE AUTHOR

Ryan James Purdy works with education systems on building the execution infrastructure needed for safe and equitable AI use. He is the founder of Purdy House Publishing and Consulting and the creator of the Stop-Gap AI Policy Guide Series, a practical set of governance frameworks designed for use by K to 12 systems and higher education institutions in Canada, the Global South, the United States, and the European Union. His work supports ministries, districts, and universities in developing AI governance plans, aligning compliance with GDPR, FERPA, COPPA, and PIPEDA, and all other existing laws – as well as improving system readiness using existing staff and resources. He writes about AI governance, literacy, and the execution gap in global education.

REFERENCES

 World Bank (October 30, 2025). Effective Reading Instruction in Low- and Middle-Income Countries: What the Evidence Shows. Global Education Evidence Advisory Panel. Co-hosted by the UK Foreign, Commonwealth and Development Office, UNICEF, and the World Bank.https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099448110272527300

OECD (September 9, 2025). Education at a Glance 2025: OECD Indicators. https://www.oecd.org/en/publications/2025/09/education-at-a-glance-2025_c58fc9ae.html

OECD (November 7, 2025). Education for Human Flourishing: A Conceptual Framework. https://www.oecd.org/en/publications/education-for-human-flourishing_73d7cb96-en.html

UNESCO (2025). Survey data indicating that 79 percent of educators report unclear or incomplete AI policies in their institutions. UNESCO AI in Education Initiative, Office of the Director for Technology and AI in Education. https://www.unesco.org/en/artificial-intelligence/education

Regulatory Frameworks Referenced

GDPR, FERPA, COPPA, PIPEDA, and the EU AI Act enforcement schedule for January 2026; California AI transparency requirements for January 2026; Colorado AI governance requirements for June 2026.

Methodological Note

This analysis synthesizes findings from three major international reports released between September and November 2025. The Literacy to Flourishing Continuum is the author’s original framework for mapping the relationship between foundational literacy capacity and advanced AI-age competencies.