AI in EdTech: The preparation gap no one talks about
Every major EdTech conference now has at least three sessions with the words “AI-powered” somewhere in the title. Understood. The energy is real, and to be fair, the numbers back it up. But after more than 13 years of developing software and apps, I’ve found that the most interesting questions are rarely about the technology itself. What matters is that the ground underneath it is well supported. And right now, the EdTech landscape is incredibly volatile for most presentations.
Numbers are not hype
To be clear, the increased adoption of AI across education is not a manufactured trend. A 2025 study by the Institute for Higher Education Policy found that 92% of students now use AI tools for learning, up from 66% the previous year. It’s not a gradual change. This means that the behavior of the entire learner changes over 12 months.
In terms of market, the global EdTech market is estimated to be $187 billion in 2025 and is expected to grow at a CAGR of 10.8% to reach $437 billion by 2033, with cloud-based adoption expected to grow at the fastest pace of 15.9% annually. [1]. These are not numbers for a niche segment. EdTech is now a significant part of the global technology economy, with AI fueling its core. Yes, growth is real. But here’s something that tends to be hidden.
The readiness gap that no one talks about out loud
According to a Rand Corporation study published in 2025, more than half of students and teachers currently report using AI in their schools, but teacher professional development, student training on responsible AI use, and school-level policies all lag significantly behind adoption rates. [2]. The part that really sticks with me is: 76% of education leaders believe their users have received adequate AI training, but 45% of educators and 52% of students report not receiving any training. This is not a small rounding error. It is a systemic blind spot that sits at the heart of what should be a story of transformation.
This readiness gap is not unique to education. This happens in almost every field where the adoption of technology outpaces the readiness of the organizations to support it. But education is especially high-stakes, as the impact extends not only to workflows but also to learners. From a development perspective, this is a common problem. This reflects what happens when you ship a feature without testing the real user journey. This tool works in a controlled environment. It doesn’t work in the field. Technology is not the cause of failure. It’s all about the assumptions about how the technology will actually be used.
Where EdTech platforms really fall short
There are certain patterns that I continue to see in how AI is integrated into educational software. The team builds a sophisticated personalization engine. Benchmark results look good. But the platform has since been rolled out to school districts, where 40% of students access the platform on lower-tier Android devices with unstable connectivity. Recommendations are delayed. Adaptive evaluation times out. Accessing the analytics dashboard for teachers requires 3 clicks and an additional 2 minutes to load. AI wasn’t the problem. The underlying infrastructure was.
Developing great educational software starts with actual usage conditions, not ideal conditions. It is designed with degraded conditions as the primary scenario, rather than edge cases. Especially in many emerging markets, where mobile learning is already the primary, and sometimes only, access point for learners, mobile-first is treated as a hard requirement rather than a design priority.
This is more important than ever before, as EdTech now targets a wider range of learners than at any point in history. Platforms that work well for college students in connected cities but not for professional learners in small towns are not solving the readiness gap problem. Optimized for the simplest version.
Three notable shifts
Beyond the headline adoption numbers, there are some structural changes that appear to be durable rather than cyclical. First, financial institutions are becoming more demanding of buyers. They are no longer impressed by the AI in the demo room. They want evidence of learning outcomes. Research shows that students who use well-designed AI tutors can achieve significantly greater performance gains compared to those who use basic AI chatbots. This suggests that the quality of implementation, not just the presence of AI capabilities, is very important.
Second, data governance is moving from a legal issue to a strategic issue. With UNESCO data showing that only 10% of schools and universities have established AI usage policies, and regulators across multiple regions tightening requirements, platforms built with privacy as an architectural principle will outlast those that treat compliance as a checkbox.
The third is interoperability. We are increasingly seeing long-term contracts for platforms that cleanly connect to student information systems, HR tools in corporate learning environments, and third-party content libraries. The days of standalone point solutions in EdTech are rapidly narrowing.
Why AI really works in education
Teams making real advances in AI in EdTech have a few common habits. They invest heavily in understanding the user’s context before specifying functionality. It not only generates graphs that appear on dashboards, but also builds feedback loops tight enough to influence teacher decisions in real time. And personalization algorithms are only as good as the behavioral data fed to them, so treat the quality of the underlying data pipeline as a top priority.
There is something to be said for restraining yourself. Not all parts of a learning platform will benefit from AI integration. Institutions and vendors that are thoughtful about where AI actually adds value and where it adds complexity without providing meaningful benefit tend to build more stable and reliable products over time.
Our student acceptance rate of 92% is impressive. But a more meaningful number might be one that tracks how many students feel that the AI they’re using actually helps them learn, not just help them complete assignments faster. That’s a more difficult measurement. That’s also the right thing to do.
References:
[1] Educational Technology Market Overview
[2] Use of AI in schools is rapidly increasing, but instruction is lagging behind
