From Stage Sage to Learning Builder
Picture a university student—call her Sara—sitting at her laptop at 11 p.m., three days before mid-term exams. He watched every lecture, downloaded every slide deck, and highlighted his notes until the pages were more yellow than white. You understand what’s important, more or less, the way you understand a city you’ve seen on a map. When the test comes, the map will not be enough.
Sara’s situation is unusual. It is the leading form of learning across higher education and many online platforms. There is a lot of content. True insight—the kind that survives three weeks and moves on to a new crisis—is extremely rare. Completion rates for online courses hover below 15%, according to MIT and Harvard researchers who study MOOCs. Students sign up with real intent, then run away. Content has never been a problem. The design was like that.
Generative AI has introduced something that the traditional classroom has never fully realized: a learning environment that is personalized, available at any hour, and able to meet the student where they are. The question is no longer whether AI is for education. Whether the platforms that deploy it understand enough how people learn to use it intelligently.
What AI Platforms Can Do That Classrooms Struggle With
Think about what a good private tutor does. They notice if you hesitate before you answer. They remember that two weeks ago, you confused two related ideas and quietly went back to check if the confusion was resolved. They adjust, in real time, to a certain state of your understanding.
A teacher in charge of thirty students cannot realistically do all of that—not because teachers are unskilled, but because the structural equation does not allow for it. A well-designed AI system can. It tracks which students need more retrieval practice, have persistent misconceptions, and are disengaged—simultaneously, across the entire cohort.
The performance record here is reasonable. A 2016 meta-analysis by Kulik and Fletcher in Review of Educational Research examined 50 controlled studies of smart tutoring programs and found mean effect sizes of 0.66 standard deviations above control conditions. Benjamin Bloom’s seminal 1984 study of the “two sigma problem” showed that individualized instruction outperformed traditional classroom instruction by two standard deviations—a gap that had not been measured economically until now. AI teaching is not a perfect match for a great human teacher, but it moves the needle in an access gap that education systems have spent four decades unable to close.
This is also where AI platforms become important, not as a new phenomenon but as a structural response. By allowing students to self-create lessons with their own materials and providing contextual AI support for personalized content, these platforms are shifting the transition from passive consumption to active construction—the kind of engagement learning science often associates with deep retention.
Why Motivation Is The Wrong Goal—And The Habit Is Right
This is where most of the EdTech industry has made a consequential mistake. The dominant design philosophy in consumer-facing platforms has been engagement optimization: streaks, badges, leaderboards, timed notifications to pull you back. The idea is that motivated students continue to learn. It is an assumption that flatters the product and fails the person.
Motivation is not a stable resource. It fluctuates with mood swings, stress, and situations. The person fired up to study on a Sunday afternoon is usually not the same person who can summon that energy on a Wednesday evening after a hard day. Designing a learning program about motivational peaks is designing an invisible version of the student that is reliable.
Self-determination theory, developed by Deci and Ryan, makes the problem more precise: extrinsic motivation—driven by rewards and social pressure—tends to crowd out intrinsic motivation when the extrinsic trigger is removed. A student who has studied every day to maintain a streak may find, when the streak is broken, that they have no internal reason to return.
A long-term goal is a habit. Research by Wendy Wood and her colleagues on the automaticity of behavior shows that habits—habits that result from contextual cues rather than deliberate motivation—are the most stable predictors of ongoing behavior. A student who has developed a consistent study habit does not need a motivational environment to begin. The cue starts the routine. The habit becomes independent.
This is the design philosophy that AI platforms should be built around. Rather than competing for motivational engagement, their design should guide the creation of sustainable learning habits—behavior that continues regardless of whether the student feels particularly motivated on a particular day.
A usability study conducted by Kampster with students enrolled at the London School of Economics in 2025 showed that students clearly distinguish between short-term engagement tools and systems designed for long-term learning. Therefore, a standard of practice the EdTech field urgently needs: building on cognitive science first, then stress-testing design decisions through systematic research with rigorous, analytically trained users.
Bjork and Bjork’s work on “desirable complexity” reinforces why this is important. Situations that feel easy—relearning passive, content below current skills—produce poor long-term retention. Attractive retrieval and spaced repetition produce strong learning because they feel strong. A platform optimized for satisfaction results delivers the former. A platform designed for storage prefers the latter, even if it is the less immediately rewarding option.
The New Role of the Teacher
None of this makes a teacher obsolete. It changes what a teacher’s best hours are spent doing.
If AI handles retrieval planning, adaptive feedback, and first-pass conceptual explanation, the teacher’s irreplaceable contribution changes to something that is difficult to automate: the relative dimension of learning, teaching that connects academic content with the student’s sense of ownership, the ability to recognize that the silent student is not giving up but struggling. These are nowhere near education. In most cases, they are the point.
OECD Report 2023 Teachers as Designers of Learning Environments puts it well: teachers are increasingly working as creators of learning, designing experiences rather than delivering content. It’s a very difficult role, not a small one—and it requires institutions to invest in teacher development rather than treat AI as a cost-cutting tool.
The conclusion
Return to Sara on her laptop. What he needed wasn’t more satisfaction. He needed a routine that helped him find, place, and struggle productively with things in the past weeks—doing a poor job of building real retention, not just the illusion of familiarity.
That system is now technically possible to build at scale. The cognitive science behind it is not new. What has evolved is the ability to work on it in a more accessible, accessible, and personable way than the average student imagines. Platforms that take this seriously—designing practice over motivation, retention over engagement—are working on the right problem. Likewise, teachers learn to work with them.
References:
- Ho, AD, et al. 2014. “HarvardX and MITx: The first year of open online courses.” HarvardX and MITx Working Paper No. 1.
- Kulik, JA, & Fletcher, JD 2016. “Effectiveness of intelligent tutoring programs: A meta-analytic review.” Review of Educational Research, 86(1), 42–78.
- Bloom, BS 1984. “The 2 sigma problem: The search for methods of group instruction as effective as individual instruction.“Educational Researcher, 13(6), 4–16.
- Deci, EL, & Ryan, RM 1985. Intrinsic motivation and self-determination in human behavior. Plenum Press.
- Deci, EL, Koestner, R., & Ryan, RM 1999. “A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation.” Psychological Bulletin, 125(6), 627–668.
- Wood, W., & Neal, DT 2007. “A new look at the practice and goal interface.” Psychological Review, 114(4), 843–863.
- Wood, W., Quinn, JM, & Kashy, DA 2002. “Habits in everyday life: Thought, emotion, and action.” Journal of Personality and Psychiatry, 83(6), 1281–1297.
- Bjork, EL, & Bjork, RA 2011. “Making things difficult for yourself, but in a good way: Creating desirable difficulty to improve learning.” In MA Gernsbacher et al. (Eds.), Psychology and the real world: Essays on important contributions to society (pages 56-64). What Publishers Should.
- Ebbinghaus, H. 1885. Über das Gedächtnis [Memory: A contribution to experimental psychology]. Duncker & Humblot.
- The OECD. 2023. OECD education at a glance 2023. OECD Publishing.
- The OECD. 2023. Teachers as designers of learning environments: The importance of innovative pedagogy. OECD Publishing.
- UNESCO. 2023. A guide to productive AI in education and research. UNESCO Publishing.
