Looking back, 2025 will likely be remembered as the year artificial intelligence fully entered the mainstream.

AI moved quickly from experimentation to everyday use. It began supporting lesson planning, tutoring, grading, feedback, and administrative workflows at scale. Adoption by educators, institutions, and learners quickly outpaced the policies, infrastructure, and norms needed to govern it.

But 2026 will be different.

It will be the year the education system must either adapt or risk leaving learners unprepared for the future they are entering. The acceleration of AI is no longer theoretical. Expectations are rising faster than institutions can respond, and the tension between innovation, trust, and readiness is becoming impossible to ignore.

From ETS’s vantage point at the intersection of assessment, skills, readiness, and responsible AI, three forces will shape what comes next. Recent global research reinforces just how quickly this inflection point is arriving. In the forthcoming 2026 ETS Human Progress Report, workers estimate that nearly one-third of their work already involves directing AI tools, and they expect that share to exceed half within the next two years. At the same time, anxiety about readiness is rising alongside adoption. More than half of workers report concern about becoming obsolete, and the gap between how important AI skills are perceived to be and how proficient workers feel remains wide.

 Together, these signals underscore that the challenge ahead is not simply AI adoption, but preparedness, trust, and clarity at scale.

Over the past few years, AI in education has largely existed as a collection of standalone, separate tools. Writing assistants, tutors, graders, and analytics platforms offered value, but often in isolation.

That era is coming to an end.

AI will stop being something educators add on and instead become embedded across the entire learning journey.

Early signals of this shift emerged in 2024 and 2025, as education-focused initiatives from major AI developers—including OpenAI and Google—acknowledged that learning environments demand different design principles than general productivity tools. Meanwhile, institutional frustration with fragmented, non-integrated AI systems accelerated the push toward AI embedded across the full learning journey. As a result, integration will accelerate.

AI copilots embedded directly into learning management systems such as Canvas, Blackboard, Moodle, and Google Classroom will become the default experience rather than optional enhancements. Institutional demand will move away from individual AI tools and toward AI infrastructure that is secure, interoperable, and deeply embedded across courseware, student information systems, and assessment platforms.

For ETS, this shift reinforces a fundamental truth. The value of AI in education depends on trusted measurement and high-quality data. Assessment, learning analytics, and feedback cannot sit outside the ecosystem. They must be embedded, interoperable, and designed to support meaningful outcomes rather than surface-level efficiency.

In the next phase of AI adoption, trust will come from integration done well.

As AI matures, one reality is becoming clear: the most valuable asset in AI-powered education is not the model itself. It is the learning data behind it.

Education-specific AI systems depend on high-quality, learning-relevant data to train, validate, and improve. General-purpose models can only go so far without insight into how students learn, where they struggle, and how progress to desired outcomes unfolds over time.

That data largely resides within learning management systems and institutional platforms, which hold the most comprehensive records of student interaction, engagement, and performance (i.e., not just the grades themselves). As a result, competitive dynamics in education AI are about to shift.

We expect a first major wave of consolidation and strategic alignment around educational data. This is likely to take the form of large-scale acquisitions or multi-year partnerships between leading AI developers, such as OpenAI, Google, Microsoft, and Anthropic, and major LMS providers including Canvas, Blackboard, Moodle, and D2L.

These moves will reshape the market quickly. They will also push data portability, interoperability, and governance to the center of policy debates, intensifying questions around data ownership, responsible use, transparency, and who ultimately sets standards for fairness and accountability.

For ETS, this moment presents both opportunity and responsibility. There is growing demand for partnerships built on validated, ethically sourced data that protects learners and institutions  to provide both transparency and explainability. At the same time, the need for transparent data governance and responsible scaling has never been greater.

As competition over data accelerates, trust will distinguish the organizations that endure.

This year, the pace of AI adoption will collide with institutional reality.

Many institutions already hold two competing beliefs. AI is moving too quickly for educators to implement thoughtfully, and even faster than policies can evolve to provide the safeguards and protections required. At the same time, it is not moving fast enough to address labor shortages, equity gaps, and rising student needs.

That tension will reach a breaking point.

Governments will push for AI-enabled efficiencies to control costs and address workforce shortages. Universities will prioritize risk management, academic integrity, and ensuring students are prepared for what comes next. Students will increasingly expect AI-native learning experiences that are personalized, relevant, and practical; and they will be expected to enter the workforce with foundational AI skills. Meanwhile, educators will face growing pressure as they navigate tool fatigue, inconsistent guidance, and evolving expectations.

The result will be a moment of reckoning.

The sector will be forced to make a choice. One path is incremental change, treating AI much like calculators are treated today, permitted, bounded, and layered onto existing models of teaching and assessment. The other is true transformation, rethinking what is taught, how learning happens, and where education fits in a rapidly changing world.

Most institutions are likely to choose incremental steps for now. But avoiding the decision is not neutral. Choosing to change slowly, or not at all, is still a decision, and one that risks leaving learners unprepared for the future.

This year, we expect to see the first wave of AI accreditation or quality frameworks emerge. New professional development standards for AI-ready educators will take shape. Clearer global guidance on academic integrity will follow. Outcomes-based evaluation will gain momentum, grounded in trustworthy assessment. This will require rethinking not only which outcomes matter, but also the processes and methods used to measure and reach them.

For ETS, this is a defining moment. As AI becomes more accessible, there is a real risk that off-the-shelf tools will be seen as “good enough” by educators under pressure, even when issues of fairness, reliability, and validity remain hidden beneath the surface. When AI content appears plausible, critical questions about bias, consistency, and appropriate use can easily fall away.

That is precisely why trusted measurement, reliable assessment, and transparent AI design will become non-negotiable. The role ETS has long played as a steward of fairness, validity, and quality becomes even more central as AI reshapes how learning is evaluated and recognized. In an AI-enabled education system, rigor cannot be optional, and models designed specifically around educational data and outcomes will matter far more than general-purpose solutions built for convenience.

By the end of the year, the AI education landscape will look fundamentally different.

The ecosystem will consolidate. Data will become a strategic battlefield. The gap between AI acceleration and human capacity will demand resolution.

At ETS, we take responsibility to help shape a future where AI enhances learning in ways that are fair, safe, and meaningful. The next era of AI in education will not be defined by hype or speed alone.

It will be defined by trust.