From sage on stage to architect of learning
Imagine a college student (we’ll call her Sarah) sitting at her laptop at 11 p.m., three days before her midterm exam. She watched every lecture, downloaded every slide deck, and highlighted her notes until the pages were yellow instead of white. She understands it more or less the same way she understands a city she has only seen on a map. When exams are coming up, a map alone is not enough.
Sarah’s situation is not unique. This is the primary mode of learning across higher education and most online platforms. It’s rich in content. Genuine understandings, the kind that survive three weeks and move on to new problems, are much rarer. According to researchers at MIT and Harvard University who study MOOCs, completion rates for online courses remain below 15%. Students enroll seriously, but then leave. Content was never the issue. The design was so.
Generative AI has introduced something that traditional classrooms cannot fully achieve: a learning environment that adapts to each individual, is always available, and can meet learners exactly where they are. The question is no longer whether AI belongs in education. It’s whether the platforms that deploy it have a good enough understanding of how humans actually learn to use it wisely.
What can AI platforms do for classrooms that are struggling?
Consider what a skilled tutor actually does. They will notice if you hesitate before answering. They remember that two weeks ago you confused two related ideas and quietly came back to test whether the confusion was resolved. These are adjusted in real time to your specific form of understanding.
It is not realistically possible for a teacher managing 30 students to do all of this. Not because the teacher lacks the skill, but because the structural arithmetic does not allow it. Well-designed AI systems can do that. Across your entire cohort, simultaneously track which learners need more retrieval practice, which learners have persistent misconceptions, and which learners are disengaged.
The validity record here is meaningful. A 2016 meta-analysis by Kulik and Fletcher published in the Review of Educational Research examined 50 controlled studies of intelligent tutoring systems and found an average effect size of 0.66 standard deviations above the control condition. Benjamin Bloom’s 1984 basic research on the “Two Sigma Problem” showed that one-on-one tutoring outperformed traditional classroom instruction by two standard deviations. This difference has hitherto been economically unscalable. AI tutoring won’t perfectly replicate a good human tutor, but it will revolutionize an access gap that has taken the education system 40 years to close.
This is also where AI platforms become relevant, not as a novelty but as a structural response. These platforms allow learners to generate courses from their own materials and provide contextual AI assistance tailored to personal content, shifting the dynamic from passive consumption to active construction. This is the kind of approach that learning science consistently links to deeper memory retention.
Why motivation is the wrong goal and habit is the right goal
This is where most of the EdTech industry makes a critical mistake. The main design philosophy for consumer-facing platforms is engagement optimization. That means streaks, badges, leaderboards, and timed notifications to bring users back. It is assumed that motivated learners will continue learning. It’s an assumption that flatters the product and disappoints others.
Motivation is not a stable resource. It fluctuates depending on mood, stress, and situation. The person who gets fired up to study on a Sunday afternoon is often not the same person who can summon that energy on a Wednesday evening after a difficult day. Designing a learning system for peak motivation means designing for a version of the learner that will certainly not emerge.
Self-determination theory developed by Deci and Ryan makes this issue more precise. Extrinsic motivation, driven by rewards and social pressure, tends to crowd out intrinsic motivation when external factors are removed. Learners who study every day to maintain a streak may find that when their streak breaks, they see no reason to go back.
A more permanent goal is a habit. Research on behavioral automaticity by Wendy Wood and colleagues shows that habits (routines triggered by situational cues rather than intentional motivation) are more stable predictors of persistent behavior. Learners who have developed consistent study habits don’t need a motivational state to get started. A cue triggers a routine. The routine becomes independent.
This is the design philosophy on which AI platforms should be built. Rather than competing motivational efforts, the architecture should aim to form sustainable study habits, behaviors that persist regardless of whether the learner is feeling particularly energized that day.
A usability study conducted by Kampster with students admitted to the London School of Economics in 2025 showed that learners clearly differentiate between short-term engagement mechanisms and systems designed for lasting learning. Therefore, the methodological standard that the EdTech sector urgently needs is to first build on cognitive science and then pressure test design decisions through structured research with demanding, analytically trained users.
Björk and Björk’s work on ‘desirable difficulties’ confirms why this is important. Situations that feel easy (passive rereading, content that is below your current ability) weaken your long-term memory. Effortful retrieval and spaced repetition create sustained learning precisely because it feels difficult. Platforms optimized for satisfaction scores provide the former. Platforms designed around retention will choose the latter, even if it’s not an immediately available option.
The new role of the educator
This does not make teachers obsolete. That changes what teachers best spend their time on.
When AI handles retrieval schedules, adaptive feedback, and initial concept explanations, educators’ invaluable contributions shift toward things that are difficult to automate: the relational aspects of learning, instruction that connects academic content to students’ sense of identity, and the ability to notice when quiet students are struggling rather than motivated. These are not incidental to education. That’s often the point.
The OECD’s 2023 report “Teachers as designers of learning environments” sums this up exactly. Educators increasingly function as learning architects, designing experiences rather than delivering content. This is a tougher role, not a tougher one, and requires institutions to invest in teacher development rather than treating AI as a cost-cutting measure.
conclusion
Return to Sarah’s laptop. What she needed was no more content. She needed a system to help her search for material, take up space, and grapple productively over the past few weeks. This was to perform the unglamorous task of building real memory, not just a superficial impression of familiarity.
The system is now technically possible to build on a large scale. The cognitive science behind it is not new. What has changed is the ability to approach learning in a way that is personalized, accessible and affordable, rather than the imagined average learner. Platforms that take this seriously and design for habits over motivation and retention over engagement are addressing the right problem. Educators are also learning 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. “The Effectiveness of Intelligent Tutoring Systems: A Meta-Analytic Review.” Review of Educational Research, 86(1), 42–78. Bloom, B.S. 1984. “The Two Sigma Problem: Exploring Group Instruction Methods That Are as Effective as One-on-One Individual Instruction.” Educational Researcher, 13(6), 4–16. Deci, E. L., and Ryan, R. M. 1985. Intrinsic motivation and self-determination in human behavior. Plenum Press. Deci, E. L., Koestner, R., and Ryan, R. M. 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 habits and the habit-goal interface. Psychological Review, 114(4), 843–863. Wood, W., Quinn, JM, and Kashy, DA 2002. Habits of Daily Life: Thoughts, Feelings, and Actions. Journal of Personality and Social Psychology, 83(6), 1281–1297. EL Björk, RA Björk, 2011. “Make things tough on yourself, but in a good way: Create the desired difficulty to enhance learning.” In MA, Gernsbacher et al. (Eds.), Psychology and the real world: Essays that illustrate its fundamental contributions to society (pp. 56–64). A worthy publisher. Ebbinghaus, H. 1885. Über das Gedächtnis [Memory: A contribution to experimental psychology]. Dunker and Hambrot. OECD. 2023. OECD Education 2023 at a glance. OECD Publishing. OECD. 2023. Teachers as designers of learning environments: The importance of innovative pedagogy. OECD Publishing. UNESCO. 2023. Guidance on generative AI in education and research. UNESCO Publishing.
Source link
