AI improves – New research tracking nearly 80 million student interactions with Pearson eTextbooks suggests the difference between helpful and harmful AI in education comes down to whether it pulls students into active reading—testing understanding, revisiting difficult con
By the time students open a digital course, the temptation is already there: ask a tool for the answer, skim what’s in front of you, move on.
Generative AI has arrived in education quickly and broadly. surrounding learners with prompts. explanations. and shortcuts that promise efficiency at every turn. But learning has never been about convenience alone. The real question now is harder: is AI being built to support how humans actually learn—and to improve outcomes—or is it optimizing for speed while thinking slips away?.
A new analysis of millions of real student interactions in higher education points toward one decisive lever: active reading.
Active reading is not a slogan from pedagogy textbooks. It describes the concrete behaviors effective readers use with text—testing their understanding. highlighting key ideas. asking questions. taking notes. and revisiting concepts that don’t come easily. Learning science ties those behaviors to better comprehension, stronger retention, and higher academic performance. Reading, in other words, is cognitive work, not passive consumption.
The research examined nearly 80 million student interactions across Pearson eTextbooks aligned to college courses over two semesters. The findings were stark. Students who used AI study tools in their eTextbooks were three times more likely than non-users to be classified as active readers.
The gap widened further in a specific setup. When students used AI tools built into instructor-led digital platforms—platforms that included assessment features and other interactive tools—the students were over 20 times more likely to be classified as active readers compared to non-users.
That matters because reading remains one of the strongest predictors of college success. and the need for stronger reading engagement is growing. National data from the National Assessment of Educational Progress shows that fewer than two-thirds of incoming students are prepared for college-level reading. At the same time, faculty report growing struggles with close reading and analysis. The challenge is not lack of access to content. It is whether students engage with it deeply enough to learn.
What distinguishes effective learning AI from the broad wave of general-purpose apps is intent—what the technology is designed to do when students reach for it. In this study’s framing, AI-enabled reading features are meant to do more than generate immediate answers to homework questions.
Instead. they support struggling readers with short. accessible summaries to aid comprehension. provide clarification on confusing concepts. and give students opportunities to practice retrieving information from memory—an approach associated with long-term retention. The goal is not to replace learning, but to augment it.
Consumer AI tools, by contrast, can drift toward a different kind of learning experience: skimming and outsourced thinking. Without trusted content or pedagogical design, speed can become the objective and depth can get pushed aside.
Responsible AI in education, in this description, looks different. It is transparent. It is trained and evaluated against expert-vetted content. And it is embedded within high-quality, trusted material that instructors and institutions already rely on. When AI is designed that way, the technology can pull students back into reading rather than away from it.
For Pearson Higher Education and Virtual Learning president Tom ap Simon. the takeaway is clear: the future of education will be shaped by AI. but the most important innovation is using technology to help students do the hard. essential work of learning more effectively. If AI can turn passive consumption into active engagement. it can change what students understand—and ultimately. the outcomes educators and students are chasing.
AI in education learning outcomes active reading generative AI digital learning platforms Pearson eTextbooks student interactions higher education learning science