Javeria Rana: When Learning Becomes Too Easy - Why Future-Ready Classrooms Need Productive Friction

The Paradox of Easier Learning

Education has spent decades trying to remove barriers to learning, and rightly so. Clear explanations, accessible materials, responsive feedback, thoughtful scaffolding, assistive technologies, and well-designed instructional resources can make learning more equitable and effective. Artificial intelligence now extends this trajectory dramatically. Students can obtain summaries before reading a difficult text, receive hints before completing a problem, generate examples on demand, translate unfamiliar language instantly, request explanations at different levels of complexity, and obtain feedback without waiting for a teacher. Used well, these capabilities can reduce unnecessary frustration and expand access to knowledge. Yet they also introduce a difficult question: what happens when the tools designed to support learning begin removing the intellectual work through which learning develops?

The assumption that easier performance indicates better learning is surprisingly fragile. A student who completes a task quickly may have learned deeply, but speed can also result from heavy prompting, worked solutions, automation, or access to an answer before sufficient thinking has occurred. Conversely, learning experiences that feel more difficult in the short term can sometimes produce stronger retention, transfer, and independence later. Robert and Elizabeth Bjork's research on desirable difficulties has been influential in demonstrating this distinction between immediate performance and durable learning. Conditions such as retrieval, spacing, interleaving, and variation can make practice feel less fluent while strengthening the learner's ability to retain and use knowledge over time.

The educational implication is important: friction is not always evidence that instruction has failed.

This does not mean that difficulty is inherently beneficial. Cognitive load theory, associated particularly with John Sweller and subsequent researchers, reminds educators that working memory is limited and that poorly designed tasks can overwhelm learners without producing meaningful understanding. Confusing instructions, inaccessible language, unnecessary complexity, poorly sequenced content, inadequate background knowledge, and technological obstacles are not intellectually productive simply because students struggle with them. Education should continue removing such barriers. The challenge is to distinguish unproductive difficulty, which consumes cognitive resources without advancing learning, from productive friction, which requires learners to retrieve, interpret, compare, explain, test, revise, or transfer understanding.

This distinction becomes especially important in AI-enabled classrooms because generative systems can remove friction with extraordinary efficiency. Consider a student facing a difficult paragraph. One form of support might clarify two unfamiliar terms so that the student can continue interpreting the argument. Another might immediately summarize the paragraph, identify its central claim, explain the evidence, and generate the analysis the student was expected to construct. Both interventions make the task easier, but they do not have the same educational consequence. The first removes a barrier that obstructs the intended thinking. The second may remove the intended thinking itself.

The same issue arises across disciplines. In mathematics, an automated system can reveal the next step before a learner has attempted to identify the structure of the problem. In science, AI can generate an explanation before students have wrestled with competing hypotheses. In literature, it can provide an interpretation before learners have examined ambiguity and textual evidence. In writing, predictive and generative tools can improve fluency while reducing opportunities to struggle with organization, precision, argument, and revision. In each case, technology can produce a stronger immediate product while leaving open the question of whether the learner has developed a stronger intellectual capability.

Research on productive failure, particularly the work of Manu Kapur, complicates the traditional assumption that students should always receive direct explanation before attempting difficult problems. Under carefully designed conditions, learners who initially struggle to generate solutions can become better prepared to understand subsequent instruction because their unsuccessful attempts activate prior knowledge, expose conceptual gaps, and create structures against which expert explanations can be understood. The point is not that failure should be manufactured or celebrated for its own sake. It is that well-supported struggle can perform an educational function that immediate success cannot always reproduce.

The emerging challenge for future-ready education is therefore not simply to make learning more efficient. It is to determine which forms of difficulty should be removed and which forms of intellectual effort should be preserved. Accessibility should reduce barriers that prevent learners from participating. Scaffolding should make complex thinking possible. Technology should extend what learners and teachers can accomplish. But support becomes educationally questionable when it consistently performs the reasoning, retrieval, interpretation, or decision-making that students need to develop themselves.

This article uses the term productive friction to describe those moments of deliberate intellectual resistance in which learners must do meaningful cognitive work rather than receive a completed pathway. Productive friction can arise when students retrieve knowledge rather than reread it, reconcile conflicting evidence, defend an interpretation, revise an unsuccessful strategy, encounter a competing perspective, transfer an idea to an unfamiliar situation, or remain with uncertainty long enough to formulate a better question. Its purpose is not to make education harder. Its purpose is to ensure that making education easier does not inadvertently make learning shallower.

The central paradox is therefore this: some of the effort educational technology can remove is exactly the effort through which expertise, judgment, and independence are formed. The future-ready classroom will need to become increasingly precise about the difference between friction that obstructs learning and friction that makes learning possible.

What Makes Friction Productive?

Not every difficult learning experience deserves to be preserved. The concept of productive friction is useful only if educators can distinguish challenge that strengthens learning from difficulty that merely exhausts, confuses, or excludes. The key question is not whether a task feels hard, but what kind of intellectual work the difficulty is requiring the learner to perform.

Research on desirable difficulties helps clarify this distinction. Bjork and Bjork have shown that conditions that reduce short-term fluency can sometimes improve long-term retention and transfer. Retrieval practice, spacing, interleaving, and variation often make performance feel less smooth because learners must reconstruct knowledge rather than encounter it in an immediately available form. The friction is productive because the effort is directed toward the target learning itself. A student trying to retrieve an idea from memory is practising access to that knowledge; a student struggling because the instructions are unclear is not.

Cognitive load theory provides an important boundary. Learners have limited working-memory resources, and those resources can be consumed by complexity unrelated to the intended learning. Productive friction should therefore preserve germane intellectual effort while reducing unnecessary burden. If the goal is to develop historical reasoning, difficulty should arise from interpreting evidence, reconciling conflicting accounts, or evaluating causation—not from an inaccessible interface or unnecessarily obscure instructions. If the goal is mathematical reasoning, students should struggle with relationships and structure rather than expend their attention deciphering poorly designed notation.

Kapur’s work on productive failure adds another condition: struggle becomes useful when it is designed and followed by consolidation. Asking students to attempt a complex problem before formal instruction can activate prior knowledge and expose gaps in understanding, but simply leaving learners in failure does not produce expertise. Productive friction requires a pedagogical sequence in which exploration is eventually connected to explanation, feedback, comparison, or explicit instruction. The difficulty prepares the learner to understand more deeply; it is not the final destination.

Vygotskian perspectives add the importance of supported challenge. Learning is most productive when students work beyond what they can accomplish effortlessly but not so far beyond current capability that meaningful participation collapses. Scaffolding should therefore adjust the conditions of the task without removing its intellectual core. A teacher may provide vocabulary, a worked example, a graphic organizer, a prompt, or peer support while still requiring the learner to construct the interpretation or solve the underlying problem.

These traditions suggest four tests for determining whether friction is educationally productive:

Purpose

Is the difficulty connected directly to the capability students are meant to develop?

Proportionality

Is the challenge demanding without overwhelming available knowledge and cognitive resources?

Support

Are learners given enough structure, feedback, and access to remain intellectually engaged?

Payoff

Does the effort strengthen retention, understanding, transfer, judgment, or independence beyond the immediate task?

This distinction matters increasingly in AI-supported learning. The appropriate question is not simply whether technology can make a task easier, but what exactly it is making easier. If AI removes transcription burden, clarifies inaccessible language, provides an alternative representation, or helps a learner overcome a barrier unrelated to the target skill, the support may increase access without weakening learning. If it supplies the inference, argument, explanation, or decision the learner was meant to construct, it may remove productive friction along with the difficulty.

The challenge for educators is therefore not to preserve struggle indiscriminately. It is to design learning environments in which students expend effort on the thinking that matters while receiving support for the barriers that do not.

Productive friction begins where unnecessary difficulty ends.

Four Forms of Productive Friction in Learning

Productive friction does not take a single form. The effort involved in remembering a concept is different from the effort required to defend a claim, negotiate disagreement, or apply knowledge in an unfamiliar setting. Distinguishing among these forms helps educators preserve the kind of challenge that serves the learning purpose rather than simply making tasks more demanding.

For analytical purposes, this article distinguishes four forms of productive friction that recur across established learning research: cognitive, epistemic, dialogic, and transfer friction. These categories are offered as a synthesis rather than as a fixed taxonomy; in practice, they frequently overlap. The four forms are not drawn from a single theorist or previously published model. Rather, they synthesize insights from several established traditions in learning theory and educational research.

Form of productive friction	What the learner must do	Principal theoretical foundations
Cognitive friction	Retrieve, reconstruct, compare, explain, infer, or revise rather than simply receive an answer	Bjork & Bjork’s desirable difficulties; retrieval practice; generative learning; cognitive load research
Epistemic friction	Evaluate competing claims, evidence, explanations, uncertainty, and justification	Deweyan inquiry; Deanna Kuhn’s argumentation and epistemic cognition; claims-and-evidence scholarship
Dialogic friction	Encounter, articulate, challenge, defend, and revise ideas through interaction with others	Socratic inquiry; Vygotskian sociocultural theory; Robin Alexander’s dialogic teaching
Transfer friction	Recognize underlying principles and adapt knowledge to unfamiliar situations	Transfer-of-learning research; deeper learning; application of knowledge across contexts

 Table 1. Four Forms of Productive Friction and Their Theoretical Foundations

Note. These four categories are an author-developed synthesis of established research traditions rather than a fixed taxonomy or previously published model. They may overlap within the same learning experience.

Cognitive Friction: Reconstructing Rather Than Receiving

Cognitive friction occurs when learners must retrieve, organize, compare, infer, explain, or revise rather than immediately access a completed answer. Research on retrieval practice, desirable difficulties, and generative learning supports the value of requiring students to reconstruct knowledge and relationships for themselves. A learner who must explain why a mathematical procedure works, recall a concept before consulting notes, or identify the central argument of a text is performing cognitive work that direct provision of the answer can bypass. The purpose is not to withhold necessary support, but to avoid supplying information at precisely the moment when productive retrieval or construction should occur.

Epistemic Friction: Deciding What Deserves to Be Believed

Learning also deepens when students encounter uncertainty, competing explanations, incomplete evidence, or claims of unequal credibility. This epistemic friction requires learners to examine how knowledge is justified. It connects strongly with Deweyan inquiry, argumentation research, and Deanna Kuhn’s work on coordinating claims and evidence. In science, students might compare explanations against data; in history, they may reconcile conflicting sources; in contemporary media, they may evaluate whether a confident claim is adequately supported. AI makes this form of friction particularly important because generated responses can appear coherent before their assumptions, evidence, or limitations have been examined.

Dialogic Friction: Allowing Ideas to Encounter Other Ideas

Learning should not always be intellectually comfortable. Dialogic traditions associated with Socratic inquiry, Vygotskian social learning, and Robin Alexander’s work on dialogic teaching demonstrate the value of having ideas questioned, challenged, clarified, and revised through interaction. Dialogic friction occurs when learners must respond to perspectives that do not simply confirm their own. Productive disagreement can reveal assumptions and expose weaknesses in reasoning that individual reflection may leave untouched. This does not mean creating adversarial classrooms. The quality of dialogic friction depends upon psychological safety, norms of evidence, and respect for the distinction between challenging an idea and diminishing the person expressing it.

Transfer Friction: Using Knowledge When the Context Changes

A final form appears when learners must use understanding beyond the conditions in which it was originally acquired. Transfer friction arises because new situations rarely reproduce the original example exactly. Students must recognize what principle remains relevant, determine what has changed, and adapt their knowledge accordingly. Research on deeper learning and transfer has repeatedly shown that this is difficult; successful performance in a familiar task does not guarantee that learners will recognize when or how to use the same understanding elsewhere. Carefully designed unfamiliar problems, interdisciplinary applications, real-world cases, and novel scenarios therefore provide valuable friction because they test whether knowledge has become flexible rather than merely rehearsed.

These forms can overlap. A rich inquiry may require students to retrieve prior knowledge, confront conflicting evidence, defend interpretations through dialogue, and apply the resulting understanding to another context. The categories are therefore not stages and should not become another instructional checklist. Their value lies in helping teachers ask a more precise question when designing support:

What intellectual work must remain with the learner if this experience is to produce the capability we intend?

That question becomes increasingly consequential as AI makes it possible to remove cognitive, epistemic, dialogic, and transfer demands almost instantly. The task of future-ready teaching is not to preserve difficulty for tradition’s sake. It is to recognize which forms of friction are carrying the learning—and to ensure that assistance does not remove them before they have done their educational work.

Figure 1. Four Forms of Productive Friction in Learning. An author-developed conceptual synthesis by Javeria Rana, illustrating four overlapping forms of intellectual effort: cognitive, epistemic, dialogic, and transfer friction. The categories synthesize established traditions in learning theory and educational research and are intended as analytical distinctions rather than a fixed taxonomy or sequential model.

When Support Becomes Substitution

The central challenge in AI-supported learning is not whether students should receive help. Effective teaching has always involved scaffolding, modelling, prompting, feedback, and adaptation. The more important question is whether the support preserves the intellectual work the learner is meant to perform or quietly performs that work on the learner’s behalf.

This distinction can be difficult to see because both forms of support may improve immediate performance. A student who receives a targeted hint may complete a problem more successfully; a student who receives the full solution may also complete it successfully. Yet the learning consequences are different. The hint preserves the need to reason through the structure of the problem, while the full solution can remove that requirement altogether. In this sense, support becomes substitution when assistance no longer enables thinking but replaces the thinking that the task was designed to develop.

The same distinction appears across disciplines. In writing, AI can help a learner identify unclear sentences or generate alternative vocabulary while leaving the argument and organization to the student. It can also generate the thesis, structure, evidence, and prose, leaving the learner primarily responsible for selection and editing. In science, a tool may provide data visualization that allows students to focus on interpretation, or it may supply the interpretation itself before students have examined the evidence. In history, AI may help learners locate competing perspectives, but if it synthesizes the significance of those perspectives before students evaluate the sources, it can remove epistemic friction. The educational value of AI therefore depends less on whether it is used and more on which intellectual responsibility remains with the learner after it is used.

This is also why the timing of support matters. Assistance provided too early can prevent students from encountering the productive uncertainty that prompts deeper reasoning. Assistance provided too late can allow confusion to become disengagement. Skilled teaching involves judging when learners need additional information, when they need a question rather than an answer, and when they need time to continue working through difficulty. AI can make support available instantly, but instant availability does not mean instant intervention is always pedagogically desirable.

A useful design principle is therefore to ask:

What is the learner supposed to become more capable of doing because of this task, and does the support leave that capability with the learner?

If the goal is interpretation, the learner must still interpret. If the goal is argumentation, the learner must still construct and justify the argument. If the goal is transfer, the learner must still decide how prior understanding should be adapted to the new context. Technology may reduce barriers around those activities, but it should not routinely remove the activity itself.

The issue is not one of purity. Learners will increasingly work with intelligent tools in academic, professional, and civic life, and education should prepare them to do so competently. The task is to design AI-supported learning without creating AI-dependent thinking. Support should expand access, increase feedback, and open possibilities while preserving enough cognitive, epistemic, dialogic, and transfer friction for learners to develop capacities they can eventually exercise without constant external direction.

The educational question is therefore not simply, “How can AI make this easier?” It is: “What should remain difficult enough for the learner to actually learn?”

Designing Productive Friction Intentionally

Productive friction should not be left to chance. If challenge is to strengthen learning rather than simply make tasks harder, educators need to design deliberately for where effort should occur, when support should enter, and what intellectual responsibility must remain with the learner. This requires a shift from asking whether a lesson is engaging or efficient to asking what forms of thinking the lesson actually requires students to perform.

One useful starting point is to identify the target capability before selecting the support. If students are learning to construct arguments, then some difficulty should remain around weighing evidence, anticipating counterclaims, and deciding how ideas should be organized. If the goal is conceptual understanding in mathematics, students may need to struggle with relationships and representations while receiving support with unfamiliar vocabulary or notation. If the goal is historical interpretation, learners should still have to evaluate sources and reconcile competing accounts even if technology helps them access those sources more efficiently. The location of friction should therefore follow the purpose of learning rather than the capabilities of the available technology.

Teachers must also think carefully about sequencing. Productive friction often requires learners to encounter a problem before receiving complete explanation, retrieve knowledge before consulting notes, commit to an interpretation before seeing an expert model, or attempt transfer before being shown exactly how the concept applies. These sequences create opportunities for students to reveal what they understand, identify what they do not yet understand, and become cognitively prepared for subsequent instruction. Support then becomes more meaningful because it responds to an intellectual need the learner has already encountered.

Choice can also strengthen productive friction when it increases responsibility rather than reducing challenge. Students might choose which evidence to use, which representation best communicates a solution, which strategy to test, or which question deserves further investigation. Such choices require judgment. By contrast, offering multiple preconstructed pathways that eliminate important decisions may create the appearance of agency without requiring learners to exercise it. The relevant question is not simply whether students have choice, but whether the choice asks them to think.

Feedback should operate similarly. Effective feedback does not always provide the correction. Sometimes the most productive response is a question, a counterexample, an indication of where reasoning breaks down, or a prompt to compare two possibilities. Feedback can therefore preserve friction by directing attention without resolving the intellectual problem prematurely. As students become more capable, support can be gradually withdrawn so that responsibility for monitoring, checking, and revising increasingly belongs to them.

AI makes these design decisions more important because the boundary between scaffold and solution can disappear quickly. Teachers may need to specify when students can use generative tools, what kinds of assistance are appropriate at different stages of a task, and what thinking must remain independently visible. For example, students might first develop their own interpretation, then use AI to generate a counterargument; attempt a solution before requesting hints; or evaluate an AI-generated explanation against disciplinary evidence. Such structures do not prohibit AI. They position it after, around, or against student thinking rather than automatically before it.

Designing productive friction therefore involves a form of pedagogical restraint. Just because an educator—or a technology—can remove a difficulty does not mean that it should. The strongest learning designs remove barriers that prevent participation while preserving the effort through which understanding, judgment, and independence are developed.

The aim is not maximum struggle and not maximum ease. It is the right resistance, at the right moment, for the right educational purpose.

Productive Friction Must Not Become Educational Exclusion

Any argument for preserving difficulty in learning must be accompanied by an equally strong commitment to accessibility and equity. Productive friction is not a justification for making students struggle through barriers that arise from disability, language, poverty, unfamiliar cultural assumptions, inaccessible technology, or insufficient prior opportunity to learn. A challenge is educationally productive only when it engages the capability being developed rather than testing whether learners can overcome obstacles unrelated to that capability.

This distinction is especially important in diverse classrooms. A multilingual learner may possess sophisticated reasoning but struggle to express it through the dominant language of instruction. A student with dyslexia may understand a concept deeply while expending disproportionate effort decoding dense text. Learners with limited access to technology may encounter difficulties that have nothing to do with the intellectual purpose of a digital task. In such cases, reducing the barrier does not necessarily reduce rigor. Translation support, accessible formats, additional processing time, assistive technologies, visual representation, or structured language scaffolds may actually allow students to encounter the intended intellectual challenge more directly.

Universal Design for Learning offers a useful principle here: access and challenge should not be treated as opposites. Learners may require different pathways into the same demanding intellectual work. Providing multiple representations of a concept, different means of engagement, or varied ways of demonstrating understanding can reduce unnecessary friction while preserving expectations for reasoning, interpretation, evidence, or transfer. The question is not whether every student encounters identical difficulty, but whether every student has meaningful access to the kind of thinking the curriculum intends to develop.

This also requires caution around the language of “productive struggle.” Teachers can sometimes interpret persistence as inherently virtuous and intervention as a sign that a task has been made too easy. Yet prolonged confusion can produce withdrawal rather than learning, particularly when students repeatedly encounter challenge without sufficient knowledge, feedback, or belonging.

Productive friction should create intellectual tension without communicating intellectual inadequacy. Students need to understand that difficulty is part of learning while also receiving credible evidence that progress is possible.

Artificial intelligence can play an important role in reducing inequitable barriers. It can translate language, offer alternative explanations, support accessibility, adjust presentation, provide additional examples, or help learners access background knowledge. These uses may increase rather than diminish productive friction because they allow students to direct more of their cognitive resources toward the intended intellectual task. The problem arises when personalization begins to lower the level of thinking expected from certain learners. An AI system that continually simplifies content based on previous performance may unintentionally restrict access to complex ideas rather than scaffold access toward them.

This creates an important distinction between differentiating support and differentiating intellectual entitlement. Students may need different amounts or forms of assistance, but support should remain oriented toward expanding their capacity to participate in increasingly demanding learning. Equity is not achieved by removing productive friction from the experience of learners perceived as less capable. It is achieved by removing unnecessary barriers so that those learners can engage meaningfully with it.

The principle is therefore simple but demanding: remove the barrier, not the thinking.

A future-ready classroom should make learning more accessible without assuming that accessibility requires intellectual simplification. The strongest educational design gives learners the support they need while preserving their right to encounter complexity, uncertainty, reasoning, and challenge.

Assessing Learning Without Removing the Friction

Assessment can either preserve productive friction or quietly dismantle it. When students know that every difficult task will eventually be reduced to a model answer, a rubric checklist, or a sequence of teacher-approved steps, they may learn to wait for certainty rather than develop the capacity to work through ambiguity. Conversely, assessment that values reasoning, revision, transfer, and justified decision-making can make intellectual effort visible and consequential.

This requires a shift from assessing only the product of learning to also examining the process through which understanding was constructed. A correct answer may conceal heavy reliance on hints, automated generation, or imitation. An imperfect answer may reveal stronger reasoning, productive revision, or a more sophisticated grasp of the underlying concept. Assessment should therefore capture not only whether students arrived at an acceptable response, but how they interpreted the problem, what evidence they used, how they responded to difficulty, and whether they could adapt their understanding when conditions changed.

Formative assessment is especially useful here because it can identify when friction is helping and when it has become counterproductive. A teacher can look for signs that students are making progress despite difficulty: explanations becoming more precise, strategies becoming more flexible, questions becoming more sophisticated, or errors becoming more diagnostic. These signals suggest that struggle is doing useful cognitive work. By contrast, repeated confusion without conceptual movement, disengagement, or dependence on increasingly explicit prompts may indicate that the task requires additional scaffolding.

Assessment in AI-supported environments also needs to make intellectual ownership visible. Rather than asking only whether students used AI, educators can ask what work remained theirs. Did they generate the initial interpretation? Did they verify the evidence? Did they revise the AI output? Could they explain why they accepted one suggestion and rejected another? Could they transfer the resulting understanding to a new context without relying on the same support? These questions are more educationally meaningful than treating AI use as a simple binary between permitted and prohibited.

This does not require every task to become a lengthy reflection exercise. Small design choices can preserve evidence of thinking: asking students to justify a decision, compare an initial and revised response, annotate where AI assistance changed their thinking, defend an interpretation orally, or complete a transfer task under reduced support. Such practices help distinguish fluent production from durable learning.

The broader principle is that assessment should reward the capacities education claims to value. If schools say they want learners who can reason, persist, revise, judge evidence, and transfer knowledge, those capacities must become visible in consequential assessment. Otherwise, students will rationally prioritize the easier path to the score.

Productive friction therefore has implications not only for instruction but for what counts as evidence of learning. The goal is not to penalize students for using support. It is to ensure that support does not make it impossible to see whether the learner has actually developed the capability the task was meant to build.

A Future-Ready Classroom Needs Deliberate Resistance

The central design challenge for future-ready education is not to choose between efficiency and difficulty. It is to determine where efficiency serves learning and where it begins to hollow it out. Technology should reduce administrative burden, improve access, widen feedback, and help learners reach knowledge that might otherwise remain inaccessible. But future-ready classrooms also need deliberate spaces in which students must retrieve, interpret, justify, revise, negotiate, and transfer rather than immediately receive.

This is especially important because artificial intelligence changes the default conditions of learning. Students increasingly have access to completed explanations, polished prose, worked solutions, summaries, translations, and recommendations before they have engaged deeply with the underlying task. The educational response should not be prohibition for its own sake. It should be sequencing. There are moments when AI should enter early to remove an access barrier, and other moments when it should enter later so that the learner first has an opportunity to think independently.

A useful example is writing. If the goal is to develop argumentation, students might first formulate a claim, select evidence, and establish their reasoning before asking AI to generate counterarguments or identify weaknesses. In science, students might propose explanations before comparing them with an AI-generated account. In mathematics, they might attempt a solution and explain their strategy before requesting alternative methods. In each case, technology is integrated into the learning process, but it does not automatically occupy the intellectual space where the target capability is meant to develop.

This requires teachers to become increasingly explicit about the distribution of cognitive responsibility. Instead of asking only what students are allowed to do with AI, educators can ask which parts of a task should be completed by the learner, which can be shared with technology, and which can reasonably be delegated because they are peripheral to the learning goal. The answer will vary by age, subject, task, and purpose. What matters is that delegation is pedagogically intentional rather than simply convenient.

Future-ready schools should also protect conditions in which uncertainty can remain unresolved for a while. Constant access to immediate answers can make not knowing feel like a problem that should be eliminated quickly. Yet uncertainty is often the beginning of inquiry, creativity, and judgment. Students need opportunities to remain with a difficult question, compare explanations, encounter disagreement, and revise an initial position before closure arrives. Productive friction gives uncertainty an educational function rather than treating it merely as an inconvenience.

The aim is therefore not to recreate a pre-digital classroom or to romanticize difficulty. It is to preserve the forms of intellectual resistance through which learners become less dependent on external direction. A future-ready classroom should use increasingly powerful tools while remaining clear about one principle: technology may assist the work of learning, but it should not routinely absorb the work through which learners become capable thinkers.

Conclusion: Preserve the Work That Builds the Learner

The promise of educational technology has always been tied to greater access, efficiency, personalization, and support. Artificial intelligence intensifies that promise by making assistance immediate and highly adaptable. Yet the educational question cannot be reduced to how much difficulty technology can remove. Learning depends partly on effort, uncertainty, reconstruction, revision, and transfer. If every obstacle is eliminated before the learner has engaged with the intellectual work beneath it, support can become a form of substitution.

The argument for productive friction is therefore not an argument for harder schooling. It is an argument for better-designed challenge. Desirable difficulties, productive failure, cognitive load theory, dialogic learning, and transfer research all point toward the same broad principle: difficulty contributes to learning only when it is connected to a meaningful capability, proportionate to the learner’s readiness, appropriately supported, and followed by opportunities for consolidation and reflection. Friction becomes educational when it strengthens the learner rather than merely delaying completion.

This distinction is especially important in AI-rich environments because fluent performance can increasingly be produced without equivalent depth of understanding. Students may submit stronger-looking work while exercising less judgment, retrieval, interpretation, or authorship. The solution is not to ban intelligent tools or to return to artificially inefficient forms of learning. It is to become more precise about the distribution of intellectual responsibility. Teachers need to decide what technology should remove, what it should support, and what must remain with the learner because that effort is itself part of the learning objective.

The four forms of productive friction discussed in this article—cognitive, epistemic, dialogic, and transfer friction—offer one way of identifying where that responsibility may lie. They are not a rigid taxonomy or a sequence to be followed mechanically. They are a synthesis of established learning traditions that helps educators ask whether students are still doing the intellectual work of retrieving, evaluating, challenging, adapting, and extending knowledge.

Equity remains central to this argument. Preserving productive friction must never become a justification for inaccessible materials, unnecessary cognitive overload, linguistic exclusion, or lowered support for learners who need it. The principle is not to preserve every difficulty. It is to remove barriers while protecting meaningful thought. Different learners may require different forms of support, but all learners remain entitled to substantive intellectual challenge.

In the end, future-ready education will not be defined by how seamlessly students move through learning tasks. It will be defined by whether those tasks leave them increasingly capable of thinking, judging, questioning, and acting without constant external direction.

The goal of educational technology should not be to make every task effortless. It should be to ensure that the effort learners still make is the effort that helps them become more capable.

That is the distinction future-ready classrooms will need to protect.

Further Reading 

The following works provide useful foundations for educators who want to explore the relationship among productive struggle, cognitive effort, transfer, instructional support, and learning in greater depth.

Robert A. Bjork and Elizabeth L. Bjork — Desirable Difficulties in Learning

Bjork and Bjork’s work is foundational for understanding why conditions that make learning feel more effortful in the short term can sometimes strengthen retention and transfer. Their distinction between immediate performance and durable learning is central to the argument that ease should not automatically be treated as evidence of effective learning.

John Sweller and colleagues — Cognitive Load Theory

Cognitive load theory provides the necessary boundary around productive friction. Because working memory is limited, challenge becomes counterproductive when learners must devote excessive attention to irrelevant complexity, poor design, or information they are not yet equipped to process. The theory helps distinguish meaningful intellectual effort from unnecessary cognitive burden.

Manu Kapur — Productive Failure

Kapur’s research demonstrates how carefully designed attempts at solving complex problems before direct instruction can prepare learners for deeper subsequent understanding. Productive failure is particularly useful for educators seeking to understand why unsuccessful first attempts can sometimes contribute more to learning than immediate access to the correct solution.

Henry L. Roediger III and Jeffrey D. Karpicke — Retrieval Practice

Research on retrieval practice demonstrates that attempting to recall knowledge can strengthen long-term learning more effectively than repeatedly reviewing the same information. This work provides an important foundation for understanding cognitive friction as purposeful reconstruction rather than unnecessary struggle.

Robin Alexander — Dialogic Teaching

Alexander’s scholarship demonstrates how classroom dialogue can become a vehicle for reasoning, challenge, explanation, and collective knowledge construction. His work is especially relevant to the concept of dialogic friction, in which learners encounter ideas that require them to clarify, defend, reconsider, or revise their own thinking.

CAST — Universal Design for Learning

Universal Design for Learning provides an important equity perspective. Its emphasis on reducing barriers while maintaining meaningful learning goals reinforces a central distinction in this article: accessibility should remove obstacles that interfere with learning without automatically removing the intellectual challenge through which important capabilities develop.

Taken together, these traditions suggest that effective learning design cannot be reduced to either making learning difficult or making it effortless. The more productive question is which forms of effort advance the intended learning, which barriers interfere with it, and what support allows learners to remain intellectually responsible for the work that matters.

Javeria will be presenting her Ripple Question Model to teach SDG-integrated lessons at the GPS Global Schools Symposium today and tomorrow, August 28th & 29th, with 3 tracks focusing on Building Sustainable School Communities, Inspiring Student Changemakers, and Innovation for Impact, from whole-school approaches and partnerships to student leadership, creativity, and technology.

Future-Ready Schools is an exclusive feature by Javeria Rana on The Worthy Educator. Check back regularly for new insights on education transformed!