Dewey, in his convoluted prose style, welcomed in schools of interpretation that distorted or perverted the ideas he attempted to convey. This distortion happened to such a degree that Dewey was compelled to write Experience and Education, a more succinct explanation of his educational ideas as an answer to critics and followers alike who ran away with concepts to form their own personal Dewey. While I would certainly recommend picking up Experience and Education (it’s only about 91 pages), it’s only a starting point and there is much more to Dewey that can lead to some truly profound positions.

We will take this task in a classic serialized format. We’ll engage with a chapter a week. You should feel free to read along and turn this into a kind of book club. I am reading the free Standard Ebooks Edition on my Remarkable tablet with medium margins, and this will be reflected in my cited page numbers. You can access that text here.

I. Education as a Necessity of Life

1. A living thing differs from an inert one in that it persists by renewal.
   
   1.1. Whereas a stone merely resists the forces acting on it, an organism turns those forces to its own continued existence.
   
   1.2. “Life is a self-renewing process through action upon the environment” (p. 6).
   
   1.3. “With the renewal of physical existence goes, in the case of human beings, the recreation of beliefs, ideals, hopes, happiness, misery, and practices” (p. 7).
   
   1.4. Education is the means of this social continuity of life.
   
   1.5. “The primary ineluctable facts of the birth and death of each one of the constituent members in a social group determine the necessity of education” (p. 7).
   
   1.6. “[T]here is the necessity that these immature members [of a social group] be not merely physically preserved in adequate numbers, but that they be initiated into the interests, purposes, information, skill, and practices of the mature members: otherwise the group will cease its characteristic life” (p. 7-8).

   
   1.6.1. “Mere physical growing up, mere mastery of the bare necessities of subsistence will not suffice to reproduce the life of the group” (p. 8).
   
   1.6.2. Humans are born unaware and indifferent to the aims and habits of the social group or community they are born into.
   
   1.6.3. Education spans the gap.
   
   

2. “Society exists through a process of transmission” (p. 8).
   
   2.1. Transmission occurs through communication.
   
   2.1.1. “Without this communication of ideals, hopes, expectations, standards, opinions, from those members of society who are passing out of the group life to those who are coming into it, social life could not survive” (p. 8).
   
   2.1.1a. I am reminded here of a particular passage from Augustine’s Confessions, “If babies are innocent, it is not for lack of will to do harm, but for lack of strength” (Confessions I.7).
   
   2.1.1b. Augustine’s child is malicious. Dewey’s child is unaware and indifferent to the group’s aims. Dewey views this as dangerous, for if the ignorance goes unaddressed, the result is societal collapse due to a lack of shared understanding and common beliefs. We must question the degree to which an ignorant state of egocentrism results in the maliciousness Augustine describes. Or, whether cruelty born of ignorance and unknowing self-concern is something altogether different from malignant intent and the will to be cruel.
   
   2.1.2. “So [poor] in original efficiency [are] the young...that even the powers needed for physical sustentation have to be acquired under tuition. How much more, then, is the case with respect to all the technological, artistic, scientific, and moral achievements of humanity!” (p. 8-9).
   
   

3. Schools are a means of the transmission of societal knowledge and dispositions, but they are only one means among many.
   
   

4. “Society not only continues to exist by transmission, by communication, but it may fairly be said to exist in transmission, in communication” (p. 9).
   
   4.1. We live in a community by virtue of what we hold in common.
   
   4.2. Communication is the means by which we come to hold things in common.
   
   4.2.1. Things in common may include “aims, beliefs, aspirations, knowledge—a common understanding” (p. 9).
   
   4.2.2. Physical proximity does not make a community. Social influence may extend across miles and years.
   
   4.2.3. “Consensus demands communication” (p. 10).
   
   

5. “[A]ll communication is educative” (p. 10).
   
   5.1. “To be a recipient of a communication is to have an enlarged and changed experience.
   
   5.1.1. Nor is the one who communicated left unaffected.
   
   5.1.2. The experience has to be formulated in order to be communicated.
   
   5.1.3. To formulate requires getting outside of it, seeing it as another would see it, considering what points of contact it has with the life of another so that it may be got into such form that he can appreciate its meaning” (p. 10).
   
   

6. Education occurs in two modes: the incidental and the deliberate.
   
   6.1. Incidental education is what is absorbed by living and communicating with others. It is natural, and it is important, but it is not the express purpose of human associations.
   
   6.1.1. “[T]he measure of the worth of any social institution, economic, domestic, political, legal, religious, is its effect in enlarging and improving experience; yet this effect is not a part of its original motive, which is limited and more immediately practical” (p. 11).
   
   6.1.2. This effect of enlargement and improvement—this educative effect—is most evident in the association of the older with the younger members of the institution.
   
   6.2. Deliberate education is the intentional educating of the young. Here, the educative effect can no longer be left to accident.
   
   6.2.1. Our chief business with the young is to enable them to share in a common life, so we cannot help asking whether we are forming in them the powers that will secure that ability.
   
   6.3. In simple societies, the incidental mode largely suffices. The young acquire the dispositions of the group by sharing directly in what the adults do, and little formal teaching exists.
   
   

7. “[A]s civilization advances, the gap between the capacities of the young and the concerns of adults widens” (p. 13).
   
   7.1. “Much of what adults do is so remote in space and in meaning that playful imitation is less and less adequate to reproduce its spirit. Ability to share effectively in adult activities thus depends upon a prior training...Intentional agencies—schools—and explicit material—studies—are devised” (p. 13).
   
   7.1.1. “Without such formal education, it is not possible to transmit all the resources and achievements of a complex society” (p. 13).
   
   7.2. “But there are conspicuous dangers attendant upon the transition from indirect to formal education. Sharing in actual pursuit, whether directly or vicariously in play, is at least personal and vital.
   

   7.2.1. Formal instruction, on the contrary, easily becomes remote and dead—abstract and bookish...What accumulated knowledge exists in low-grade societies is at least put into practice; it is transmuted into character; it exists with the depth of meaning that attaches to its coming within urgent daily interests” (p. 13).
   
   7.2.2. “There is the standing danger that the material of formal instruction will be merely the subject matter of the schools, isolated from the subject matter of life experience” (p. 13).
   
   

8. “To avoid a split between what [we] consciously know because [we] are aware of having learned it by a specific job of learning, and what [we] unconsciously know because [we] have absorbed it in the formation of [our] characters by intercourse with others, becomes an increasingly delicate task with every development of special schooling” (p. 14).

In Summary

Living things differ from inert ones in that they persist by renewal, by acting on their environment to sustain themselves. A social group is alive in the same sense: its members die, but the group’s characteristic life continues only if each new generation is initiated into the aims, knowledge, and practices of the old. Education is the name for such social renewal, and its medium is communication. A society does not merely persist by transmission, it exists in transmission and in communication. For if we are to hold aims, beliefs, and knowledge in common (what makes a community at all) we must engage in communication for this is how we come to hold things in common. Because all communication enlarges and reshapes the experience of everyone it touches, social life is already educative as a by-product of itself. In simpler societies, such incidental education suffices, since the young learn by sharing directly in what adults do. As a civilization grows complex, however, the distance between a child’s capacities and adult concerns widens until direct participation can no longer suffice, and deliberate institutions—schools—with explicit materials—studies—become necessary. This formal mode is indispensable, yet it tends to drift from lived experience and become abstract and inert. The standing problem of education is therefore one of balance: to keep the formal and the informal joined, so that what we are taught and what we absorb by living in a community do not split apart.

Works Cited

Augustine, St. (1961). Confessions. (R. S. Pine-Coffin, Trans.). Penguin.

Dewey, J. (1916). Democracy and Education (Standard Ebooks ed.). Standard Ebooks. https://standardebooks.org/ebooks/john-dewey/democracy-and-education

In writing this, I consider the 100 hours I’ve spent in the game. I consider what I have accomplished—the discovery and exploration of numerous planets, as well as the categorization of many species of flora and fauna. I built a trade network that spans solar systems through a series of settlements, mining operations, and trade vessels that I’ve constructed, pirated, or negotiated for along my seemingly Sisyphean journey to the center of the universe. I wonder what my choices in this universe reveal about me and my preconceived notions regarding power and empire. However, the journey has also been a familiar one. It is built upon tried-and-true game mechanics à la Minecraft—mining to refine, refining to construct, only to mine and refine and construct more efficiently. At base, and more familiar still, the game seems to align with Koster’s (2013) assertion that “classifying, collating, and exercising power over the contents of a space is one of the fundamental lessons of all kinds of gameplay.” This concept rings true in No Man’s Sky, as the player literally discovers and classifies various species for in-game currency. However, despite its size and sci-fi trappings, the game is about something even more profoundly human—survival. Moreover, I argue that the ways in which we choose to survive in No Man’s Sky as we continue our journey to the center of the universe say much about ourselves and the very purpose of play.

Game Overview

At base, No Man’s Sky is a game of discovery, expression, fellowship, and fantasy to borrow Hunicke, LeBlanc, and Zubek’s aesthetic taxonomy (2004, p. 2). The in-game aesthetics stem from how players interact with the game’s core mechanics—time, tools, minerals, and planets (specifically, each planet’s climate, geography, culture, and resource distribution). Scarcity is the immediate challenge players face. Every tool you own and every technology you engineer runs on some kind of fuel or requires a specific part to continue functioning. The challenge of resource management coupled with the breakdown of necessary tools makes for a satisfying gameplay loop that is again similar to Minecraft, but different in that the in-game threat comes not from a horde of night creatures, but rather from more passive sources like atmospheric toxicity, intense radiation, extreme temperatures, and the occasional aggressive inhabitants, pirates, or parasitic alien organisms (beware dilapidated frigates).

In order to look closely at No Man’s Sky, we must first determine where it finds itself among the many definitions of games. Juul’s (2003) configuration of what constitutes games, borderline cases, and “not games” serves as a useful starting place. No Man’s Sky features all of the core mechanics of a game as described in the figure below.

However, while No Man’s Sky has fixed rules (e.g., if your life support system fails, you die), and while the player may be attached to the game’s outcome as they invest countless hours to further their journey to the center of the universe, the win state is such that the game eschews valorizing outcomes and spurns the concept of negotiable consequences. The game takes cues from open-ended simulations and free-form play where the journey, more than the destination, is what truly matters.

No Man’s Sky, with all of its procedurally generated planets, has shown us through its years of frequent updates and DLCs that it consistently has something new to teach us. It is through this constant tinkering, updating, and expansion of the game that the developers skirt around Koster’s (2013) definition of a good game, “one that teaches everything it has to offer before the player stops playing,” by maintaining a game in such a way that it may never stop teaching us for as long as the developers add new, free, and interesting mechanics that alter the play experience. The numerous patches, DLCs, and updates are not quite a reinvention of the initial game as much as they are an expansion of, or a fulfillment of, the promise that the game made upon its initial release. The game is ‘good’ or ‘fun’ insofar as its size, scope, vision, and new, rewarding challenges have players revisiting it time and again to learn more about this boundless universe from its inception to its present-day iteration. Moreover, this trend does not seem to be slowing. One only has to look at reporting from both game and general tech publications for evidence of the claim. Countless articles and video essays taking a stab at “revisiting” or “re-evaluating” the game and its burgeoning legacy. That said, my thoughts regarding the game extend beyond Koster’s definition of a good game or the newfound admiration of reporters and critics. McGonigal (2011) points out that “[i]n a good...game you’re always playing on the very edge of your skill level, always on the brink of falling off. When you do fall off, you feel the urge to climb back on. That’s because there is virtually nothing as engaging as this state of working at the very limits of our ability” (p. 24). You might call this flow. To this point, No Man’s Sky also succeeds.

Gameplay Analysis

I wake up stranded on a planet with no recollection of how I got there. The atmosphere is highly toxic. I do not know where my ship is, and my life support systems are failing. In other words, the game tosses players into a dire situation. However, it is also here that I first learn from the game. It teaches me the fundamentals of mining and life support systems. I quickly learned that to survive, I would need to mine sodium to recharge my hazard protection systems. The space suit’s handy AI helps players figure out how they may go about finding sodium. I pressed the right stick, and my suit’s visor ran a scan of the surrounding environment, marking a nearby sodium deposit. Once I reached it, I used my mining beam to collect the mineral, which recharged my life support. In other words, the game’s mechanics (a toxic atmosphere) interacted with my failing life support system, which caused my health to decrease (dynamics), and worried me enough (aesthetics) to gather a surplus of sodium.

It is in the early game that I would like to focus my consideration on moments of in-game learning. In such an expansive universe with so many things that can go wrong, the game risks overloading a player with too many choices, too many problems, and too many opportunities, too early on. The developers solve this problem by introducing you to the most basic dangers and needs and then scaling up from there. Yes, while starting with a starship, stable climate, and a set of awesome technologies would be great to get the player exploring the universe quickly, it would likely be damaging in the long run as the player would not understand the fundamentals that the game is built upon, nor how all of the different in-game systems build upon one another. Being given everything from the start would not equip the player to adapt the mechanics and technologies to novel situations and problems. At best, the novice player would have a shallow understanding of the things, but not of their functions and constituent elements. By learning the foundations and recalling them frequently within the gameplay loop, we develop the ability to plan and reason with the mechanics. This is where the game shares some similarities with Daniel Willingham’s ideas about critical thinking. The player cannot explore the universe or answer its many problems in a novel, self-directed, and effective way without first possessing the necessary background knowledge along with some meaningful practice in applying it (2016). Chief among this necessary background is how to mine and repair gear, which is where the game begins.

Once the player mines the necessary minerals and recharges their life support systems, they are prompted to locate their ship by heading to an SOS beacon. Each new challenge in the early game allows the player to practice a specific skill that builds their knowledge base, leading to mastery. However, just as the player masters, say, how to mine sodium, the game presents a new challenge—repairing your launch thrusters and pulse engine. This introduces a new game mechanic: starship repair. Raw materials like sodium and ferrite dust may not be useful in such an endeavor, and now the player must find a way to refine those minerals and create metal plating for the ship’s haul, or fuel for the launch thrusters. The game taught me the basics of survival in its universe; now it teaches me the basics of crafting. I mine the necessary materials, build a small portable refinery, and create the necessary objects to get my ship operational.

It is only when the player has mastered mining, crafting, repair, and survival that the game allows them to explore. James Paul Gee (2013) is instructive here. The information presented by the game is both just-in-time and on-demand. What those last two phrases mean in the context of J.P. Gee and No Man’s Sky is that the game gives the player a small piece of information at first, and the player must apply that information just as they learned it (e.g., your life support system is failing, you need sodium, go mine sodium). Moreover, as the player meets new and unanticipated challenges (e.g., their multi-tool fails), they must seek out more information on how to solve the challenge (i.e., use the menu to discover that you need carbon nanotubes, which require carbon that you can then craft into the necessary tubes).

As I described earlier in the paper, the game initially disregards much of its complexity and subsequently builds up to it. Gee (2013) would describe this as the “fish tank principle,” a simplified micro-world that keeps the core variables and relationships visible while temporarily removing distracting complexity. As learners grasp how the essentials interact, you gradually reintroduce additional elements, rules, and noise until they can handle the full ecosystem. However, as the game progresses and the player gets their ship off the ground, they escape the metaphorical fish tank and enter the wider sandbox, where they can explore, test, and try new things. As the game expands, so do the various issues and goals, which lead to more practice of key skills needed later in the game as you come across more complex situations that require you to use your skills in new, self-directed ways, like space combat, managing a settlement, or running trade routes.

Conclusions & Classroom Connections

No Man’s Sky, as I hope to have conveyed, has taught me a great deal about its underlying systems and best practices for succeeding on my journey. However, I am left to wonder whether the game’s mechanics resulted in anything other than in-game proficiency. Were there more implicit lessons with broader implications that apply to, say, a social studies classroom? An overarching concept from the game is that of systems thinking. I encountered a complex game world and was confronted with how my actions and other variables led to unforeseen consequences. I considered how ethical it was to strip planets of their resources to fuel an individual quest, with no regard for other species, outside of classifying them to make money from the discovery. The game taught me valuable lessons regarding scaffolding and the application of the cycle of expertise. Still, the game itself is likely not something I could use in a classroom without significant tinkering, design, and ultimately limiting what the students’ play experience would be. That said, the game allowed me to carefully consider the relationship between resources, trade, power, and ecosystem diversity. In sum, an endless universe with a vague objective, such as the one found in No Man’s Sky, is indeed a game. The developers scaffold the player’s experience in such a way that we are not dropped into a boundless world with no idea how to get what we need and meet our goals, but always with just enough information and just enough practice so that we are playing to the very edge of what we know. And that is a good game.

References

Baron, S. M. (2012, March 22). Cognitive Flow: The psychology of great game design. Retrieved from: https://gamasutra.com/view/feature/166972/cognitive_flow_the_psychology_of_.php

Egenfeldt-Nielsen, S., Smith, J. H., & Tosca, S. P. (2013). Understanding Video Games: The Essential Introduction (pp. 31-59). New York, NY: Routledge.

Gee, J. P. (2013, November 13). Jim Gee principles on gaming [Video]. YouTube.

Gee, J. P. (2013). What video games have to teach us about learning and literacy. New St. Martin’s Press.

Hunicke, R., LeBlanc, M., & Zubej, R. (2004, July). MDA: A formal approach to game design and game research. In Proceedings of the AAAI Workshop on Challenges in Game AI, 4(1), 1-5.

Klein, D. (2021, August 22). No Man’s Sky 5 Years Later. Gamespot. Retrieved from https://www.gamespot.com/videos/no-mans-sky-5-years-later/2300-6455970/.

Koster, R. (2013). A theory of fun for game design. O’Reilly Media, Inc.

McGonigal, J. (2011). Reality is broken: Why games make us better and how they can change the world. Penguin Press (p. 11).

Stuart, K. (2021, August 10). No Man’s Sky: give years of meteors, mining, and metaphysics.

The Guardian. Retrieved from https://www.theguardian.com/games/2021/aug/10/.

Tassi, P. (2021, July 30). ‘Cyberpunk 2077’ Is Way Off Schedule For A ‘No Man’s Sky’ Style

Comeback. Forbes. Retrieved from https://www.forbes.com/sites/paultassi/2021/07/30/

Willingham, Daniel T. “Knowledge and Practice: The Real Keys to Critical Thinking.” Knowledge Matters, no. 1, Mar. 2016, pp. 1-7.

Q1. Teacher burnout is at an all-time high with grading, planning, and paperwork. In your view, how can AI realistically reduce teacher workload without lowering instructional quality?”

• The first piece worth considering here is that what a teacher does most of the time is teach. If you look at the schedules of most teachers in the US today, active teaching hours are the majority of the schedule, and that in itself isn’t typically perceived as the area of lost time. The feelings of burnout and perceived slights of disrespect that contribute to teacher turnover are myriad, and we’ll discuss a few. Yet, in general, they all contribute in some way to the teacher feeling underprepared and unsupported for working with their students. It is these two feelings—lack of support and inadequate preparation—that, over time, are demoralizing, making one cynical, jaded, or burned out.

  It’s a snowball effect of bureaucratic red tape, administrative duties, paperwork, incessant meetings, emails, hall or lunch duties, and valuable parent-teacher communications. All this happens while the teacher is on their planning, during their lunch, or before or after school. Two of those four options are unpaid. One option, lunch, I believe, should be reserved for the teacher to take a break—being on for 8-9 hours a day is taxing. So, that leaves planning time as really the only paid time for a teacher to address all those previously mentioned pulls on their time and resources. On top of that, they have to plan for 4-8 classes, perhaps two or more subjects, and who knows how many unique learner needs are represented in those classes. This is what most educational AI tools are designed for—the non-instructional time in the teacher’s day. The question is how to make that limited time maximally productive so the teacher enters the classroom prepared, such that they can make that time productive.

  Quality, in this context, comes through in the degree to which the system used contributes to the creation of high-quality instructional materials, a clear idea of how to implement those materials, strategies to address misconceptions and issues, planned differentiation for those students who have shown a need for it, extensions, elaborated analogies, and opportunities for practice. In other words, does the use of the tool meaningfully contribute to the teacher coming prepared to be a good teacher? If you provide that, if you enable that to happen, if you give professionals the tools and support they need to be effective, our thesis is that you reduce teacher burnout. It will not eliminate burnout, there are too many mounting systemic pressures that no mere piece of technology can remedy, but perhaps it serves as a pressure valve.

  This links to the broader principle of sustainability we’ve talked about before: the profession cannot endure if it continually extracts unpaid labor from teachers. You can’t build excellence in a field that struggles to retain talent. AI, as a means of reducing the translation cost between evidence and practice, can redistribute the cognitive and temporal load, making teaching once again a sustainable career.

Q2. Teachers often spend hours creating differentiated lesson plans for diverse learners. How can AI help make planning more efficient while still honoring the individuality of each student?”

• AI, if used well, can make planning more efficient, precisely by honoring student individuality. This does not mean merely pandering to surface-level engagement. Dropping in references to Fortnite or TikTok to catch attention only goes so far and is hardly a durable lever to pull for student engagement over time. Rather, it means embedding general heuristics teachers hold about their craft into instructional design—spacing practice, interleaving concepts, retrieval through low-stakes quizzes, modeling with worked examples, connecting to prior knowledge, scaffolding access and then fading supports, providing corrective feedback, giving examples and non-examples. Each of these practices has robust evidence supporting them, but implementing them consistently as the needs arise is demanding on the teacher’s already strained resources. Here we touch the research-to-practice gap. Teachers often know, in principle, that retrieval practice or spaced learning works. But operationalizing them across 5–8 classes, dozens of students, and varied needs is overwhelming. As Carl Hendrick put it, AI has the potential to assist deeply in the orchestration of instruction through the time-honored holy trinity of educational practice: Retrieval, Spacing, and Interleaving.

Q3. AI tools like Erasmus can help to support teachers looking for ideas for differentiation. How do you see AI ensuring equity—so advanced learners, struggling students, and everyone in between all get meaningful support?”

• Well, the first part of the answer is in the question. AI tools, such as our agent Erasmus, assist teachers who are looking to differentiate instruction for their students. This means that the technology is an amplification of the teachers' intentions. In short, equity in education has never been a feature of technology alone, but has always depended on the intention with which the technology is used.

  Could an AI tutor today adapt to the needs of all students for meaningful support, which I would define as effective support insofar as it contributes to positive student learning outcomes, perhaps not possible without the aid of the tool? No, they most certainly do not. AI requires intention and direction. In their current state, they ensure equity only if ensuring equity is the goal of the user and the sandbox for creation is large enough to accommodate that intention. In the case of Eduaide, this might involve calling our Differentiated Instruction Assistant, specifying your need, objectives, standards, and content, and then exploring potential options and instructional interventions that you may or may not use. This should not be viewed as shirking responsibility from the designers and builders of the technology. You can certainly obstruct or pervert the user's intentions, and that should weigh heavily on the minds of everyone involved in the production of educational technologies.

Q4. Many new teachers struggle with a lack of onboarding and support. Could AI serve as a mentor or coach for new educators? What might that look like? How does Eduaide help?

• The transition into teaching is notoriously difficult. Novices often lack the tacit knowledge that experienced teachers deploy: when to fade a scaffold, how to anticipate misconceptions, or what a good worksheet looks like. Could AI serve as a mentor or coach in this regard? Not in the human sense, but perhaps in a way that makes effective practices visible and accessible to both novices and seasoned teachers. Imagine a tool that not only generates problems but also provides implementation guidance: “Start with a fully worked example. Then try a partially worked problem. Then an independent one. Along the way, ask goal-free questions that prompt students to self-explain their steps.” In this way, AI can function as a practice space for pedagogical thinking. It expands the toolbox of methods and techniques and gives novices a framework for using them. In doing so, you can address one of the chronic causes of early-career attrition: lack of support.

Q5. Some educators are hesitant about AI, fearing it will either replace teachers or add another layer of tech to learn. What would you say to build trust and encourage thoughtful adoption?”

• First, there is no panacea, no cure-all. Everything in life is a trade-off. AI will certainly add another layer of tech to learn, and this incurs an opportunity cost on the educator. The question is whether there is utility bought at the cost. That’s one reason why we started Eduaide: to provide an on-ramp for teachers to think about AI in pedagogical terms. However, there is no easy answer here. I would worry less about the replacement of the teacher by AI. The tech simply doesn’t have the capabilities to engage in the most economically valuable tasks. I think we’re seeing this with many of the recent studies on the effects of GenAI in various pilots. That said, in education, the technology will assuredly amplify existing problems, as well as presenting interesting emergent properties from the technology interacting with teachers and students. Like any technology, there are unique affordances and boundaries worth considering.

Q6. Looking ahead, do you believe AI will become as essential in classrooms? What’s your vision for how AI could transform the teacher’s role in the next 5–10 years?

• All I can say for certain is that AI will have a meaningful effect on teaching and learning since the technology so closely deals with human cognition. What these effects will be and to what degree the effects will be felt is a difficult thing to predict. I do see paths forward where there can be meaningful tools made for teachers and students to assist in the work of teaching and learning, but I also know that if learning science has taught us anything, it’s that this is all a game of probabilities, not prescription or standard solutions. The hope is that AI helps us break dogma—that it allows teachers to experiment, adapt, and refine practice in light of evidence. The fear is that AI could harden dogma, offering prefabricated routines that lull us into comfort, shallow fluency, fragile understanding, and all the while, meaningful learning declines.

For essays and expanded show notes, subscribe to the ⁠⁠⁠Learning Stack on Substack⁠⁠⁠.

Episode Chapters

• 00:00 Introduction

• 01:06 Defining Big I and Small i Innovations

• 04:25 Tracing AI's effects in Education

• 06:15 Evaluating Innovations in Education

• 08:05 On scaling Small i Innovations Responsibly

• 12:08 Guardrails for Evaluating New Tools

• 15:43 On Assessments in Higher Education

• 19:58 Building Trust and Relationships in Education

• 23:54 The Role of AI in Student Assessment

• 27:51 Concerns About AI Detectors and Predictive Analytics

• 31:13 Minimum Viable AI Policy Architecture

• 34:44 Defining Success for AI Policies in Education

• 37:08 Implementing AI Policies in Education

• 38:22 Crowdsourcing AI Policies: Insights and Trends

• 40:15 Norms in AI Syllabi: Verification and Attribution

• 41:26 The Serials Crisis in Academic Publishing

• 42:30 The Rise of For-Profit Publishers

• 44:44 The Impact of Subscription Models on Access

• 46:43 The Ethics of Academic Piracy

• 49:29 The Role of Academic Pirate Networks

• 50:37 Rethinking Copyright in the Age of AI

• 52:40 The Future of Open Access and AI

• 59:33 Access Disparities in Global Academia

• 01:01:50 The Interplay of Open Licensing and AI

• 01:06:19 Lance Eaton's Vision for the Future

Resources and Links

⁠Lance's Blog⁠

⁠Lance's Substack AI + Education = Simplified⁠

⁠Lance's Syllabi Policies for Generative AI Repository ⁠

Peter Suber, Open Access (⁠Free Access⁠)

• Matt Beane, The Skills Code

• Sci-Hub, Library Genesis, Anna’s Archive (discussed in context of academic access)

***

As always, I’ll put out a Too Long Didn’t Listen Summary in the next post. Stay tuned!

The Learning Stack Podcast is produced by the team at ⁠⁠Eduaide.Ai⁠

The real pleasure of design is in the process of awakening gradually to the multitude of accumulated insights interwoven with that environment.

Kenya Hara, Designing Japan

To design a tool is to imagine a form of life.

We can call this form of life an occupation (and for some folks, a preoccupation). Each and every occupation (whether work or recreational) comes with its own jargon, concepts, operations, and procedures codified into practice. We can refer to the codified practices as a kind of grammar. This grammar, mastered through repetition and accumulated experience in our occupation, shapes our perceptions, assumptions, heuristics, metaphors, and biases about the world. The shorthand used by a carpenter differs from that of a doctor or a tabletop gamer. The metaphors used by a chef to think about the world are different than those of a teacher.

The tool you design must fit within the form of life it is meant to accompany. More than fit, however, the tool should subtly elevate and express that occupation to its peak potential. The tool should not be ostentatious in its elevation unless, of course, the occupation calls for ostentatious displays (a double-neck Gibson SG comes to mind).

Designing a workspace for teachers is not the same as designing a tech “platform.” Any piece of educational technology is attempting to cohabitate with a grammar of practice that long predates the technology: routines, constraints, expectations, procedures, and mental models—hundreds of micro-decisions compressed into 45-minute class periods. In making a tool that fits within this context, the simple claim “we save teachers’ time” is almost offensive in its banality. Time is not an inert substance to be hoarded. Time saved is meaningless unless it points toward something—toward reflection, toward better methods of instruction, toward more humane classrooms. Time is not saved, only reinvested. If your design doesn't clarify what kind of reinvestment it imagines, then it is failing its users for lack of imagination. However, bad incentives can also influence design and lead a tool astray.

Design that begins with metrics like engagement, session length, or LTV quickly forgets the form of life it claims to serve. When metrics become targets, they cease to be good metrics and devour the meaning of the practice itself. This is how tools become extractive, and I believe this fallacy of extractive engagement goes a long way to explaining the "enshittification" of so much consumer software.

The educational technology tool, in its best form, clarifies and concentrates the work of teaching and learning. It makes the tacit knowledge of the field explicit, it isolates our focus on the essential, and it gives contours to those things previously felt, but unnamed (here I am thinking of the heuristics we have, like “space your practice,” or “do a review.” Things which may be effective, but are imprecise in execution). A good tool, like a good sentence, achieves inevitability. It could not be otherwise.

Design is the discipline of attention to use. The aim of design in educational technology is not to disrupt or stylize, but to serve. A tool designed with care reflects a belief that even the most ordinary tasks deserve extraordinary respect. The height of design is to make a completely ordinary tool that fits perfectly into the user’s daily activity.

Let’s state the virtues plainly:

Simplicity

Not the startup virtue of “minimalism” as a euphemism for underbuilding, but simplicity as the elegance of necessity and the stark beauty of utility. This means boiling down your tool to the point at which nothing can be removed without consequence.

Precision

A fittingness between form and function. Precision is the sense that the tool respects its own purpose and is neither looser nor tighter than what is called for.

Thoroughness

Completeness without spectacle. This necessarily means a respect for edge cases and a refusal to outsource difficult challenges. A thorough tool anticipates how the user’s needs shift over time and builds around that arc.

Delicacy

This is the rarest virtue in software. It is an understanding that the user is not merely an input vector, but a human situated in a unique form of life. Delicacy means restraint in service of dignity, a kind of respect for the end user.

I owe a significant debt here to Kenya Hara, the Japanese Graphic designer, whose books have been a great source of inspiration. Especially his text, Designing Japan

Everything we see, hear, and think about is critically dependent on and influenced by our long-term memory.

Kirschner, Sweller, & Clark (2006)

When is it appropriate for a student to use an LLM? It’s a hotly debated topic, and not one I expect to resolve here. However, it forces open quite a few issues and assumptions. We’ll take one of them today. AI as a tool for cognitive offloading—a means of displacing thinking tasks onto external systems—reveals something about our assumptions regarding the role of memory in education.

For decades, dominant pedagogical models in American public education have downplayed the role of memorization. In particular, Constructivist approaches1, despite their descriptive accuracy of learning as a process, do not lead to effective pedagogical strategies because they ignore human cognitive architecture (Kirschner et al., 2006; Kirschner et al., 2004). Constructivist models presume that novices can learn like experts through inquiry and discovery, yet this overlooks the limited capacity of working memory and the essential role of prior knowledge (this is the expertise reversal effect: unguided methods benefit only the knowledgeable.) Yet, the aims were always well-intentioned. We only de-emphasized factual knowledge in order to elevate skills (even though they are two sides of the same coin). We had lofty goals to teach students to “learn how to learn,” to “think critically,” and become researchers of their own understanding. The teacher in these models was, to use an old adage, a guide on the side, rather than a sage on the stage.

The common refrain was as follows: why commit something to memory when a search bar, chatbot, or AI tutor can retrieve it in seconds? The logic is seductive: free up working memory for “higher-order” thinking. Yet, seductive as it may be, the logic is built on a misunderstanding of how the brain works and how humans learn.

Let’s state it up front. You may not agree with what follows. In fact, I hope some of you don’t. Even still, I’ll build the argument piece by piece, and along the way, I invite you to note where you diverge, and I'm sure that can open up an interesting conversation.

Here’s my thesis: Declarative knowledge is the substrate for higher-order thinking. You cannot evaluate what you do not understand. You cannot synthesize ideas that were never explained. Moreover, we cannot effortlessly assimilate declarative academic knowledge. Knowledge and competence do not spread via osmosis. We require deliberate practice and repeated retrieval.

Constantly looking things up instead of internalizing them results in shallow fluency and fragile understanding2. This implies that, despite an abundance of external data and information—constantly available on Google, Wikipedia, and ChatGPT—insight and wisdom still emerge from internal knowledge. Underuse of the brain's declarative3 and procedural4 memory systems undermines reasoning, impedes learning, and diminishes productivity in novel problem-solving. When learners chronically offload information—definitions, examples, methods—they fail to form the durable memory traces that underlie comprehension and transfer. The issue, to be clear, is not offloading per se, but premature offloading before memory traces are formed. Domain expertise necessarily means a command of some set of facts core to that discipline, which, when committed to long-term memory, supports problem-solving within that domain as working memory capacity is freed up (Deans for Impact, 2015). Moreover, looking something up is not a free lunch. It incurs a cognitive cost, involves task switching, and requires reconciling what we just looked up with the argument or task at hand. Knowledge committed to memory is accompanied by a level of contextual awareness, not provided by a Google Search or zero-shot prompt.

I am, therefore, not advocating for taking a bunch of disconnected facts and banging them back. I understand the need for simple frameworks and mental models as scaffolding to hang information and knowledge on. Yet it is through the very processes of retrieval, spacing, and interleaving that we build the factual bedrock necessary for the construction of mental models. We can’t simply short-circuit the effortful process of learning and jump to high-order thinking. Experiences leave traces that shape perception, guide attention, and influence judgment. The twinge we feel when we see an incorrect instance of something we know to be true tells us as much. We do not always know we know something, but our ability to think well often depends on knowledge that has quietly become automatic. This is a type of transfer from declarative to procedural knowledge through practice, retrieval, and application.

When language models serve as permanent proxies for memory, they erode the very capacities they aim to support. The student who never memorizes anything becomes fluent only in searching. But search fluency is not mastery. It’s a simulation of knowing, not knowing itself. In this sense, internal knowledge enables us to filter out noise, detect patterns, ask meaningful questions, and identify contradictions. Without it, there is no ground on which to build higher thought.

Offloading, again, is not the enemy. Used judiciously, it extends our capabilities. But used indiscriminately, it displaces them. The paradox is not that we use tools to help us think. It’s that we begin to mistake the tools for the thinking itself.

Even if you can retrieve a fact, that retrieval is not the same as integration. Knowing that 12 * 5 = 60 is different from seeing that fact as part of a larger schema of multiplication, number sense, and problem-solving ability. A language model might provide an elegant answer. However, the human learner must still possess the cognitive structure to make sense of it, remember it, and apply it.

That said, not all uses of AI displace memory. There are important and growing exceptions. When used to generate low-stakes quizzes, summarize key ideas for later review, or support retrieval practice, LLMs can enhance memory formation rather than hinder it. For example, a student using GPT to self-test on key terms from a biology unit—then correcting and explaining their errors—is engaging in exactly the kind of elaborative retrieval that supports durable learning5.

Likewise, students with learning disabilities or executive functioning challenges may benefit from structured offloading. AI tools can reduce extraneous cognitive load by organizing information, chunking tasks, or modeling step-by-step thinking.

Even expert thinkers offload routinely. Mathematicians use symbolic tools like WolframAlpha not because they lack understanding, but because they’ve built the internal schemas necessary to interpret, verify, and apply results. LLMs can serve similar purposes to augment rather than substitute human reasoning if the foundational knowledge is already in place.

The key distinction, then, is not whether we offload, but when and why we do so. Offloading must follow, not precede, internalization. It must support reflection, not shortcut it. In this sense, responsible use of AI in education is not about refusing offloading altogether, but about building the internal infrastructure that makes offloading useful in the first place.

References

Deans for Impact. (2015). The science of learning. Austin, TX: Deans for Impact.

Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work: An analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based teaching. Educational Psychologist, 41(2), 75–86. https://doi.org/10.1207/s15326985ep4102_1

Kirschner, P. A., Martens, R. L., & Strijbos, J.-W. (2004). CSCL in higher education? A framework for designing multiple collaborative environments. In P. Dillenbourg (Series Ed.) & J.-W. Strijbos, P. A. Kirschner, & R. L. Martens (Vol. Eds.), Computer-supported collaborative learning: Vol. 3. What we know about CSCL … and implementing it in higher education (pp. 3–30). Boston, MA: Kluwer Academic.

In this conversation, Michael joins Thomas to demystify what it actually takes to run a public school system. He explains why the superintendent’s job is more “orchestra director” than instructional leader, what most people get wrong about school governance, and why political noise is one of the least glamorous but most consequential parts of the job. He also shares candid reflections on the future of teacher retention, the importance of school culture over salary, and why professional development is the most underrated line item in a school budget.

Along the way, they discuss the importance of student voice, what metrics really define educational success, how school systems can evolve beyond outdated age-based structures, and why collaboration across school types—public, private, charter, and micro—is essential. If you want to understand how schools really function and what it means to lead with humility, purpose, and clarity, this episode offers valuable insights from a true practitioner.

“Public education is the great equalizer. That doesn’t mean it’s perfect. It means it’s necessary.”

Read an enhanced transcript with helpful links at ⁠Eduaide.Ai⁠.

Recorded 18 July 2025

1.1. The language of a discipline is not arbitrary. It is the reflection of its form of life, a codified set of tools, terms, principles, and mental models shaped and reshaped through use across generations.

1.1.1. Teaching students to write an argumentative essay about a historical claim, then, is not just teaching them how to express ideas about history, but how to think more clearly about history by inhabiting the language of that discipline.

1.2. If we agree that “language provides the social architecture of our thoughts,” then it not only shapes thought but also delimits it, thus scaffolding what can be known, said, and thus even conceived in a social context.

1.2.1. Writing as a process of synthesizing, grappling, and reformulating concepts and claims is why it was such a useful proxy for assessing student learning (which again is essentially invisible to us, so we take all kinds of proxies to measure it).

1.2.1a The above italicized was is a consequence of the challenge presented when we can outsource the production of language to a party outside of ourselves, because while we may then produce coherent language on an assignment, or in a blog, the language is devoid of the student or “writer’s” ideas, and thus devoid of that synthesizing, grappling, and reformulating that made writing assignments a useful proxy for thinking and learning in the first place.

1.3. The product (the finished essay or report) is less indicative of understanding than the cognitive work undertaken to produce it.

2. Writing is thinking

2.1. Simply because there is technology that can mimic the form of writing so well does not mean we throw writing out entirely. Verification is what’s needed (and not verification of the sort an AI detector promises, which is the shallow and punitive version).

2.1.1. Daniel Willingham put out a recent piece on the power of conferencing with students in regard to their writing.

2.1.2. Conferencing in his conception takes on the dimensions of Elaborative Interrogation + Feedback where the teacher elicits explanations and justifications (elaborative interrogation), offers tailored, high-quality feedback based on the student and their justifications, and reveal misconceptions and correct them in real time, while also assessing not only the correctness of ideas but the structure of the student’s mental models—how they synthesize, grapple with, and reformulate ideas.

2.1.2a Objection: Are one-on-ones scalable to all contexts where students are writing? Also, this approach favors verbal capacity while penalizing those students who benefit from extended time, research, and careful preparation.

2.1.2b. Rebuttal: The students who benefit from extended time, research, and preparation have access to those benefits in the writing process. The conference is no different than a presentation they would prepare for.

2.1.2c. Objection: Conferencing would not stop students from using AI tools.

2.1.2d. Rebuttal: True, but is there something to be said for the ability to explain and defend the choices made in human-AI collaboration? If everything about the assignment was AI-generated, down to the citations, would the student be able to defend it or speak on it with authority? Is this a better form of ensuring academic engagement and integrity than dystopian futures of screen monitoring? I’d argue that conferencing opens a window into authenticity: Did the student actually think this, do they understand the reasoning on display in the paper, or did they outsource their cognition to a machine?

3. When students use AI to generate fluent but shallow responses, they’re simulating intelligence, not necessarily exhibiting it.

3.1. In this context, intelligence is not what was written, but the ability to reflect on, explain, revise, and argue for what was written.

A picture held us captive. And we could not get outside it, for it lay in our language and language seemed to repeat it to us inexorably (PI, §115).

The 2024 paper “Large Language Models for Education: A Survey and Outlook” proposes a comprehensive taxonomy to capture the breadth of LLM use in education. Like all taxonomies of such and such domain of education, it forms a particular picture of what teaching and learning are. This picture centers on performance as the benchmark for quality. Teaching and learning in this scheme consist of discrete, automatable units, including “question solving,” “error correction,” “material creation,” and so on. This captures neither the full grammar of educational activity nor the lived context in which those activities gain meaning. I do not say this to critique the work of the researchers involved, they state clearly that they are developing a technology-centric taxonomy. Rather, I mean to surface a deeper problem: that in articulating this particular taxonomy, we see a field emerging that structures our perception of educational work in ways that are shallow and epistemically flattening. The danger, in other words, is not that the taxonomy is inaccurate within its own frame, but that its frame is becoming a more dominant lens through which technology in education is interpreted.

Let’s take each domain and subdomain of the taxonomy in kind before launching into a critique.

Domain 1: Study Assisting

Subdomain 1.1: Question Solving (QS)

In this subdomain, LLMs are positioned as advanced problem solvers across various fields, including math, law, medicine, and programming. The authors also survey the techniques used to enhance LLMs’ abilities in question-solving, including Chain-of-Thought prompting, in-context learning and few-shot prompting, and multi-agent collaboration. It is in this question-solving affordance that LLMs can provide “high-quality answers to [students] blocking questions in a timely manner” (p. 2). What is the picture of study implied here? The learner seems to be reduced to a consumer of “high-quality” solutions. Does it follow that learning occurs via exposure to correct answers?

I should clarify that if we have any aspirations toward AI tutors, we should, of course, test whether they can solve the questions that the student would be working on. However, it does not follow that the necessary end state is that the AI simply provides answers. More vital, and harder to gauge, is the ability to guide a student to correct methods of interrogating and solving the problem at hand. These methods used by students for interrogating and solving problems seem to come from deliberate practice and experience gained through spaced repetition and the interleaving of new and related concepts. An interesting area of study would be how an AI system can adapt and respond to changes in a learner’s cognitive states, allowing us to optimize the spacing of practice intervals and the types of questions we interleave.

Subdomain 1.2: Error Correction (EC)

This section of the paper discusses how Large Language Models (LLMs) are being applied to correct student errors in grammar and programming. The authors state that such error correction is particularly important during the early stages of learning when timely correction is most impactful. Models like Codex and GrammarGPT are shown to outperform prior systems in error detection. Effectiveness in error correction is defined by fix-rate, patch size, and improved outputs. So, what is the implicit assumption of what learning is by LLMs as error correctors? Learning is cast as the elimination of error, and the model is the corrector. This is of vital importance as I would hazard a guess that in the provision of feedback and correction, a teacher would not view themselves primarily as a corrector, but perhaps as a coach or guide. These two roles carry with them a motivational domain not accounted for in the paper. Just how much of the role of the teacher is to perform a motivational role? That question is open to debate, and for a more thorough handling of the question, I refer you to this recent piece featured on ACX.

In short, this subdomain is primarily couched in terms of performance metrics—the number of errors fixed, patch size, and alignment with human corrections. However, from an instructional perspective, the value of error correction lies not just in correctness, but also in productive struggle (the student taking steps to correct errors), and in how we direct attention (the actionable steps for improvement). Without this pedagogical framing, there's a risk that LLM-based correction becomes mere repair, and not instruction.

Subdomain 1.3: Confusion Helper (CH)

Unlike QS and EC, this subdomain is explicitly pedagogical in nature. CH avoids solving the problem for the student and instead generates hints, guiding questions, or explanations to facilitate learning. Tellingly, this section reports mixed results:

• LLM-generated explanations are often less effective than those provided by human tutors.

• Synthetic hints may be too general or use vocabulary that is inaccessible to students.

• Adaptive explanations to different student subgroups show some promise.

This subdomain is explicitly less performance-centered and recognizes that learning happens in the process of resolving confusion, not just in getting the right answer. More fittingly, the evaluative emphasis shifts toward instructional value, student comprehension, and even pedagogical fit.

Domain 2: Teach Assisting

This section surveys how LLMs support educators by automating routine tasks such as generating questions, grading assignments, and producing instructional materials. Across these domains, LLMs are framed as productivity tools that unburden teachers from repetitive work and streamline the preparation process. The underlying claim is that LLMs enable teachers to focus on the “human” aspects of instruction by outsourcing rote or clerical tasks.

Subdomain 2.1: Question Generation

The primary preoccupation of this subdomain is with LLM-generated questions. This makes sense, as questions are among the most powerful tools in a teacher’s toolbox, and having a diverse array of them at hand is generally useful. The authors survey fine-tuning and prompt-engineering methods to calibrate outputs to match textbook content, curricular goals, or instructional taxonomies. For example, Doughty et al. (2024) examine GPT-4' ’s ability to generate high-quality multiple-choice questions aligned to Python programming objectives, while Lee et al. (2024) map LLM-generated prompting questions in reading comprehension tasks.

The LLM is praised for its ability to produce balanced questions of “clear language” and “high-quality distractors,” but nowhere is there inquiry into the sequence, function, or cognitive demand of those questions within a learner’s trajectory through a course. Here again, the underlying theory of instruction appears underdeveloped. The questions discussed exist as discrete objects, detached from the lessons or learners they are meant to serve. They exist at the level of fluency, question structure, and alignment to curricular goals and objectives.

Questions for elaborative interrogation? Those are more difficult to pin down.

Subdomain 2.2: Automatic Grading

I don’t have much to say here. This subdomain addresses the use of LLMs to score student work, particularly open-ended writing and short-answer responses. Previous grading models relied heavily on similarity-based comparisons to “golden answers,” lacking sensitivity to student reasoning or alternative expressions of correctness. With the advent of LLMs, however, the models are incorporating rubrics, prompting for reasoning, and producing both scores and narrative feedback of “satisfactory” quality. Moreover, just as earlier domains treated learning as the accumulation of correct answers, this domain risks treating teaching as the accumulation of judgments. It abstracts away the relational, motivational, and developmental aspects of assessment, reducing the act to something transactional and post hoc.

Subdomain 2.3: Material Creation

In this final subdomain of Teach Assisting, LLMs are deployed to generate instructional content: asynchronous course modules, EFL materials, worked examples, and other learning resources. Several studies emphasize the importance of a “human-in-the-loop” process, combining zero-shot prompting with human review and revision. Among the subdomains, this one is the least reductive in spirit. The authors acknowledge the need for contextual relevance, adaptation to learner needs, and integration with the instructor’s existing materials and teaching style. Still, the primary value of LLM use is framed in terms of efficiency: streamlining content creation, reducing friction, and minimizing time investment by teachers.

Again, the constant appeals to efficiency remind me of the error in the study of Open Educational Resources (OER), which consistently made appeals to access and cost savings at the expense of understanding the instructional implications and outcomes of adopting OER materials. What is the instructional opportunity cost of using LLMs for material creation? Do teachers have reduced fluency and command over the materials they are teaching? You could imagine a study in which we have two groups of novice teachers: those using an AI system like SchoolAI or Magic School, and those planning with traditional materials. Then, after a planning session, we ask them to explain the lesson material to us. Who do you think would perform better at the explanation?

The hypothetical study raises a question worth considering. Is the process of materials creation an expression of teachers’ content and pedagogical knowledge and a form of practice? My hunch is that they are, and that despite the professional development offered in school, we often overlook the fact that the most effective forms of development for teachers are peer collaboration, mentorship, and the consistent practice of simply doing the job with timely feedback and actionable steps for improvement. Perhaps the administrative burdens in school are the problem, and that while what a teacher mostly does is teach, how much of the other side of the job is filled with what the anthropologist David Graber would describe as beurecratic bullshit? I leave that question for you to reflect on.

Domain 3: Adaptive Learning

Unlike previous domains, which focused primarily on task automation or material generation, Adaptive Learning enters more ambitious conceptual territory: diagnosing what students know and adapting instruction accordingly. Even here, however, the center of gravity remains squarely in the realm of performance optimization. Learning is treated as a kind of data-driven navigation. The goal is to measure, map, and match the “mastery” of “concepts” to content via a dynamic delivery system. If the picture of teaching in Domain 2 was that of a clerical worker made more efficient, the picture of learning in Domain 3 is that of a GPS unit routing a user from confusion to correctness.

Subdomain 3.1: Knowledge Tracing

Knowledge Tracing attempts to model a student’s mastery of content over time, typically by analyzing patterns of correct or incorrect responses to determine what they "know." Of the studies mentioned, Lee et al. (2024) warrants a further note in that they use LLMs to predict question difficulty based on question stems and associated concepts, allowing for better estimations of challenge levels even for unseen items. This is a clever enhancement, especially when considered alongside Sonkar and Baraniuk (2024) exploring whether LLMs can simulate incorrect student reasoning if given a learner’s “knowledge profile.” Despite the inversion and enhancement, we cannot absolve LLMs of the above critiques. Even here, we’re still tracking observable performance against a latent model of “mastery” inferred through proxy indicators.

Subdomain 3.2: Content Personalization

We finally arrive at the perennial unfulfilled promise of edtech—personalization. I have a few ideas about why personalization doesn’t seem to be a generalizable instructional intervention, but that’s for a different essay, as this one is already too long. For more on the failure and complications of “personalization,” I again refer you to the review of school on ACX for some more flavor on this conversation.

Also, as I type the phrase “more flavor,” Grammarly is telling me to change the phrasing to “additional context,” and now I’m lamenting how AI-generated text makes us more boring writers. This topic is dry enough as it is. If I want to add some flavor and taste, I should be able to without a pop-up bugging me to try the “beta” version of my sentence.

With that aside aside, let’s not dismiss the authors’ treatment of personalization out of hand. This subdomain presents the most aspirational vision in the paper: an intelligent system that not only diagnoses a learner's current level but also guides and motivates them using personalized content. But how far does this personalization go? Currently, most efforts appear to be limited to surface-level adaptations, such as swapping in keywords or redirecting learners based on concept mastery (or, to unmoor ourselves from jargon, how many correct answers the student got). The idea of personalization is reduced to tailoring content, not reshaping the path, process, or product of the learning exercise. An algebra problem mentioning LeBron James to a student who likes basketball is still an algebra problem. True personalization involves operationalizing heuristics such as spacing practice and interleaving concepts, tailored to the specific classroom context.

Critiques and Conclusions

The taxonomy also provides an overview of commercial tools, including chatbots, content creation tools, teaching aids, quiz generators, and collaboration tools. However, given the length of this post, I’ll address these in a separate post. Let’s center ourselves on what has been presented so far and consider it rationally.

The taxonomy treats LLMs as tools. Tools for solving problems, generating questions, and creating materials. However, tools gain meaning only within specific human activities. The primary difference between a griddle spatula and a paint spreader lies in their intended use. In descriptive terms, they are both flat pieces of metal affixed to a handle of some sort. The difference, again, is in their use in a particular form of life—cooking and painting. Similarly, error correction or hint generation are not simply functions; they are roles embedded in the form of life we call education. This is not to say the taxonomy is wrong on its own terms. It is technically accurate, even comprehensive, but it is grammatically confused. It takes the surface of tasks (inputs and outputs) and treats them as the essence of teaching and learning. It overlooks the deeper grammar of educational activity: why we ask questions, how we correct, when we guide, and what it means for something to be understood. Again, this is not the fault of the authors, who do a commendable job with their stated task. It is an issue of the enterprise of LLMs in Education. As such, the taxonomy risks perpetuating a picture of education as performance optimization—learning as the accumulation of correct answers, teaching as the efficient delivery of prompts and feedback. It is a picture that is hard to see beyond—a picture of correct answers without meaningful learning.

If LLMs are to become part of the educational form of life, the real challenge is not to automate what teachers do, but to preserve, and perhaps even deepen, the intentionality and human responsiveness that define teaching as a practice.

Thanks for reading. Talk to you soon!

I was motivated by a conviction that Education is meant to be the inheritance of every citizen and, as such, access to it should not be curtailed by the state, by another individual, or by any other public or private circumstance. Yet, too often remains, in practice, access to the best in Education remains gated by money, licensing, or sheer institutional inertia. Some barriers are visible on a college bill, others hide in daily practice, or in the reluctance of institutions to do things differently. I studied Open Educational Resources because they promised a clean break from all that. Make learning materials free, open, and adaptable, and you unshackle students from costs and teachers from rigid courseware. Freedom begets progress… At least, that was the faith.

The answer was more complicated than the slogans suggested. The basic logic of OER remains compelling. By removing price tags and copyright barriers, these resources could ease the financial burden on students and give teachers freedom to tailor materials to local needs. In theory, access expands and learning follows. In practice, however, the research base reveals persistent weaknesses. Most studies I reviewed neglected to control for the very conditions that shape educational outcomes: prior knowledge, teacher practices, instructional quality, and the difficulty of the materials themselves. As a result, claims that OER boosts learning often rest on shaky ground. When students have no book at all, giving them a free one does help. However, once that floor of access is raised, what matters more is whether the content supports sound instruction and whether the teacher can use it well. Move past the basic rescue provided by access, and the old truths reassert themselves: poor teaching, sloppy sequencing, and weak practice tasks do not become potent simply because they carry an open license.

This distinction between access and instructional rigor is easy to overlook. It is also the same blind spot that now threatens to repeat itself, more dramatically, under the allure of Generative AI. Large Language Models can produce an endless supply of quizzes, explanations, and lesson plans on command, often at zero marginal cost to the end user. Like OER, AI seemingly lowers the cost of producing and distributing knowledge artifacts.

Yet abundance, while seductive, is no guarantee of educational value. A thousand generated practice questions mean little if they misrepresent what makes practice effective. A summary dashed off by a Large Language Model may feel helpful, but for the purpose of instruction, it lacks coherence, clarity, or cognitive scaffolding, the whole project risks flattening student thought into the regurgitation of trivia rather than deepening understanding. Or as Carl Hendrick put it,

If students rely on AI to generate ideas, structure arguments, or retrieve facts, they run the risk of cosplaying domain knowledge while bypassing the effortful processes that lead to genuine learning. The result is a kind of intellectual outsourcing that short-circuits schema construction and leaves students with shallow fluency but fragile understanding. The point is not producing the thing, it’s the effort in producing the thing.

The lesson my dissertation keeps reminding me is that the removal of barriers—whether financial, legal, or technical—must be paired with a rigorous commitment to instructional quality. Open licensing gave us freedom to adapt and share, but without careful study of what works and for whom, that freedom often produced uneven results. We now see the same tension with AI. The real work lies not in churning out more content, but in designing, testing, and refining how that content interacts with how people learn. Good teaching, whether human or machine-assisted, demands a respect for how knowledge is constructed. And as far as we know, that pursuit is best made through clear sequencing, appropriate challenge, timely feedback, and opportunities to practice and apply new ideas.

Open Educational Resources did not transform learning simply because they were free. They helped when they were thoughtfully integrated, well-crafted, and grounded in sound pedagogy. Generative AI, for all its speed and novelty, will follow the same law: it will serve as a tool for expanding opportunity only if paired with the quiet, deliberate discipline of instructional design and careful research.

If we are serious about using AI to broaden educational horizons rather than flatten them, we must look beyond the allure of frictionless generation and focus instead on the sturdy, unglamorous, messy, uneven, and perhaps at times invisible craft of building knowledge.

Access is the beginning, not the end. What happens after we open the door determines whether learning truly takes root.

* Any guesses on the cover image’s aesthetic style? Those who have attended a scholastic book fair have an advantage.

I plan to write another post on an interesting thread that emerged in some of the discussions at the summit on Small Language Models, local deployment, distributed computing, and self-hosting. For now, I’d like to share a short talk on some of the constraints facing AI’s potential in closing the research-to-practice gap in Education.

What I am going to say here is not new. Rather, I owe a great debt to those many thinkers who came before me: Daniel Willingham, Carl Hendrick, John Dunlosky, Paul Kirschner, Robert Bjork, Patricia Alexander, William James, Jerome Bruner, Daniel Kahneman, Richard Mayer, the organization Dean’s for Impact, and many others.

The application of their ideas, with slight modification, is enough to address in some meaningful way many of the problems and challenges that come up at the intersection of AI, Teaching, and Learning.

That said, before we can begin discussing how AI may interact with the research-to-practice pipeline in Education, we must sketch the boundaries of the discussion.

For starters, these models are epistemically hollow. That is, they achieve a remarkable facsimile of knowledge—text outputs that are fluent, coherent, and even seemingly reasoned, but remain a facade. A Large Language Model (LLM) does not know anything. It is a collection of impressive statistical algorithms that process the patterns embedded in language, trained to guess the next word based on the trillions of words and word combinations it has seen. It reflects language, but it does not understand it.1

The philosopher Ludwig Wittgenstein cautioned that philosophical problems arise when language “goes on holiday.” In the discourse around AI, it seems that language has indeed packed its bags. We’re borrowing words from one language game2 to play another, and this semantic problem gives rise to philosophical ones which will haunt generations far beyond our own.

Take the term “hallucination.” It evokes a mental lapse as if the AI were an individual who momentarily took leave of its senses. An AI that generates a false statement isn’t deviating, it’s doing exactly what it was built to do: predict the statistically probable sequence of words.

A hallucination, then, is a statistically plausible response that also happens to be factually incorrect.

The error here lies in our expectations, not in the system’s “mind” (it has none). By calling these errors hallucinations, we smuggle in a metaphor of human perception that obscures more than it reveals. We start asking the wrong questions.

This line of thinking begins to reveal the three great threats of AI in Education: shallow learning, misinformation, and cognitive overload.

Education is no stranger to ideas that sound right yet ultimately prove false. In fact, the field is littered with convenient myths and intuitive misunderstandings. We shouldn't be too hard on ourselves about this either. Learning is complex, and to all the parties involved, invisible. We have many surrogates for learning that help us measure and think about it aspects of it, but the change in the memory of an individual is, by its very nature, a deeply internal process, and our instruments for measuring it—assessments, classroom observation checklists, and even brain imaging—are very blunt indeed.

What we call "Education" then arises from a panoply of effects. The rigid, dogmatic application of a single method, framework, or technique is misguided and almost always results in diminishing returns. If you try to use inquiry-based learning every day, you run risks. Students may not have the necessary prior knowledge to engage in the inquiry and can thus form misconceptions and misunderstandings that are difficult to get out of. If you try to use game-based learning all of the time, you are substituting a kind of intrinsic motivation for the extrinsic engagement of the game.

All this to say, every method and technique has its affordances and boundary conditions depending on contextual factors. Even still, teachers can, and do, wield a massive influence on student learning.

To provide an example of an intuitive misunderstanding, or stubborn myth in Education, let’s talk briefly about Learning Styles. For decades, teachers have been told that students learn best when taught in their preferred modality (visual, auditory, kinesthetic, etc.), leading to well-intentioned efforts to “personalize” lessons by style. It sounds wonderfully empathetic and intuitive. Yet rigorous reviews have found no evidence that matching instructional style to a learner’s supposed preference boosts learning outcomes.

Students have preferences, sure, but humans are not so neurologically pigeonholed that a “visual learner” can only learn through images. The concept of Learning Styles never needed AI to entrench itself, it spread on the power of a tidy, hopeful narrative, but one can easily imagine, and in fact test for themselves, an AI system offering “custom learning style lessons” if fed that narrative, and references to learning styles are certainlyabundant in the training data of all the foundation models. Without a grounding in research, technology ends up amplifying our fads and dampening our blind spots. And again, in Education, there are many of those.

Let's take one more. Bloom’s Taxonomy, originally a simple heuristic for thinking about the complexity of cognition and knowledge, morphed in practice to a rigid hierarchy. A Pyramid (or ladder or steps or arrow or ice berg, depending on the visualization) to scale from mere facts at the bottom to the exalted realm of creativity at the top3. Teachers were admonished to always push students toward analysis, evaluation, and synthesis, often at the expense of foundational knowledge. The result is an unintended denigration of factual knowledge, as though one could analyze or create in a vacuum. Yet, analysis and critical thinking, for that matter, are not generic skills standing alone in a vacuum. They’re an emergent property of deep domain knowledge.

Higher-order thinking requires a bedrock of factual knowledge. In the pragmatic words of a popular adage: “You can’t connect the dots if you don’t have any dots to connect.” Facts and concepts stored in long-term memory are the very dots that insight connects.

Carl Hendrick, in a recent Substack post, put this in stark language worth repeating here: "If students [and I would add teachers] rely on AI to generate ideas, structure arguments, or retrieve facts, they run the risk of cosplaying domain knowledge while bypassing the effortful processes that lead to genuine learning. The result is a kind of intellectual outsourcing that short-circuits schema construction and leaves students with shallow fluency but fragile understanding."

The question is, how do we avoid a future in which students and teachers become only prompters and collectors of AI outputs without the requisite disciplinary thinking to make sense of the AI outputs? In short, and I stress this is the nothing new part, AI tools should support retrieval practice, elaboration, and spaced repetition, to make actionable the conception of a mental latticework in which recall, explanation, application, and creation intertwine and overlap.

AI, as we've established, doesn’t think, but it does pattern-match with extraordinary scale and speed. This affordance makes it well-suited to generate varied practice problems, draft examples, surface analogies, give step-by-step instructions, and provide feedback on well-defined tasks. I again agree with and would like to echo Carl Hendrick when he writes, "AI's genuine educational potential lies not in mimicking human cognition but in amplifying distinctly human learning processes."

Heuristics and rules of thumb, like interleave concepts or space practice and knowledge retrieval opportunities, can be effective tools, but they’re also difficult in their vagueness. Interleave in what instructional context? Spaced practice for what content domain? Retrieval with what form of knowledge (declarative, procedural, conceptual)? When should I interleave X concept? What about in Y circumstances?

The effectiveness of these methods hinge on timing, sequencing, and individual adaptation. All tasks that are cognitively demanding and logistically difficult to manage at scale in real classrooms. An AI, however, may hold potential in its ability to dynamically orchestrate instruction. Not in terms of total automation and the removal of the teacher, but in very targeted terms, guided by the teacher. Small-scale interventions like adjusting spacing intervals, question sequencing, or concept mixing in response to each learner’s progress or the teacher’s need. In algebra, it may be the creation of varied problem sets employed in dynamic spaced intervals to help consolidate procedural fluency. In history, it might be elaborative interrogation questions to deepen causal reasoning.

The challenge ahead is one of scaffolding understanding. That work will not be led by engineers alone, nor by educators alone, but by a new kind of partnership that respects the affordances and limits of both human and machine.

To return to my use of the term language game at the beginning of this talk, if we think of Education itself as a kind of language game—shifting between instruction, exploration, feedback, and performance—then the key design challenge is ensuring that AI participates appropriately. It must follow the rules, even if it doesn’t know the game.

I’ve reproduced the whole of my brief remarks below in the hope that they capture something of the spirit of the event. The theme for the day was igniting innovation, and in that sweltering south Georgia heat, I felt I might have ignited before walking in the door. So to say we could all feel the theme in the air wouldn’t be much of a stretch.

The address was slated to be about 5 minutes, but I came in a little under time. Ringing in my ears was the wise counsel of Dr. Johnson on Milton’s Paradise Lost: “No one ever wished it longer.”

Opening Remarks on Innovation in Education

It is an honor to be here with you today and to celebrate your work, to learn from your ideas, and to witness the many ways you've used technology in the service of great teaching. I am especially excited for the presentations and the lesson artifacts—to see what they reveal about how applications and tools can be shaped by the hands of educators to make a real impact on student learning. And more than that: to create the conditions for learning to continue.

There is a mistake, however, in believing that innovation is found only in the adoption of new tools. True innovation in the classroom emerges from a deeper place, in the teacher’s intent, in their goals, in how they respond to the unique circumstances they find themselves in. Put another way, true innovation comes from your desire to improve the conditions for learning, to want to respond to the needs of the students, and to adapt in the face of complexity and change. A tool is nothing without your intent, without your goals, without you.

Long before AI, smartboards, video streaming, and adaptive software, teachers were innovating. Educators were designing their own scaffolds, coaching one-on-one, discussing with small groups, and improvising learning experiences to reach every student as best as they could with the resources they had. What we might call instructional innovation, then, is often the natural extension of a teacher’s professional judgment.

This judgment is not just instinct. It’s the result of training, of reflection, and of accumulated experience—the ability to synthesize content knowledge, student understanding, and pedagogy into something dynamic and responsive.

If we were to put it in a simple formula, it would go something like:

Professional Judgement + Responsiveness = Instructional Innovation.

So when we talk about technology in the classroom, and the innovative ways it’s used, what we really ought to be talking about is amplification. Meaning that the technology amplifies the teacher's voice, it doesn't replace it. It amplifies the teacher's intent and their skill, but still doesn’t replace it. When the tool is well designed, that is to say, aligned with learning sciences research and grounded in sound instructional practice, it can become a partner in your work. But only in the hands of a skilled teacher does it become a catalyst for transformation.

I think this showcase will be a testament to the kinds of things you can do when schools support teachers as designers, and not just deliverers of instruction. So thank you for your work, your ideas, and your willingness to share them today. I can’t wait to see what you’ve created and how it’s shaped your practice.

With that said, please join me in a round of applause for the phenomenal staff who organized this event, and for all of you, for the ingenuity and care you bring to your classrooms each day.

Originally published on the Eduaide.Ai Blog.

A recent article in The New York Times offered an unsettling but, the more I think about it, unsurprising headline: the latest generation of “reasoning” AI models are more error-prone, not less. Despite impressive advances in math and programming capabilities, factual grounding is deteriorating.

"The company found that o3 — its most powerful system — hallucinated 33 percent of the time when running its PersonQA benchmark test, which involves answering questions about public figures. That is more than twice the hallucination rate of OpenAI’s previous reasoning system, called o1. The new o4-mini hallucinated at an even higher rate: 48 percent.

When running another test called SimpleQA, which asks more general questions, the hallucination rates for o3 and o4-mini were 51 percent and 79 percent. The previous system, o1, hallucinated 44 percent of the time."

— Cade Metz & Karen Weise (5 May 2025)

A curious paradox thus emerges: as models become more “capable,” they become more convincingly wrong. So, what does a hallucination wrapped in the claim of reasoning mean? It means the AI’s language is fluid and coherent. Its logic may appear sound. Yet, the underlying claims are false. Perhaps you ask why or how this happens. The central refrain of the article is that we don’t know precisely why this occurs, but we know that it does.

Most public commentary treats hallucinations as a bug. Something to patch with better training, alignment, or more careful prompting. But these diagnoses are too shallow. To call these factual errors “hallucinations” is already to smuggle in a metaphor that obscures more than it reveals. Perhaps this is less of a concern in technical contexts where the term may serve as a pragmatic shorthand for denoting the divergence from known/trusted ground truth. Still, it remains eminently misleading in public and philosophical discourse.

As Ludwig Wittgenstein observed, philosophical confusion arises when “language goes on holiday.” That is to say, philosophical confusion, and I would argue confusion around AI more generally, occurs when words, untethered from their ordinary use, are repurposed into abstract problems that bear little resemblance to lived meaning. Take the term hallucination. It gestures toward some psychological malfunction—seeing things that aren’t there. This implies deviation. That is, something separate from the way things typically are, and that may be resolved. When, in truth, the system has no structurally embedded norm of truth to deviate from. In a hallucination, the model is not confused. It is not erring. It is simply continuing a sequence of likely tokens based on its training data. That’s what it was built to do. The result is words and sentences that are statistically relevant, but factually inaccurate. Let's call that taking the long way round to being wrong, factually.

You may be saying, "Statistically relevant token selection aside, wrong is wrong. So what's gone missing in these reasoning models? Aren't they supposedly more knowledgeable?”

What's gone missing is not the model's grasp on truth, but our grasp on what we mean by truth in this context at all. The language of cognition—reasoning, belief, knowledge—has been extended to statistical systems without a firm footing. So, we end up trying to think about hallucinations as if they were epistemic failures, rather than seeing them as artifacts of probabilistic generation. The developers of these systems are aware of this distinction, but it doesn’t seem the public is, and that matters.

If we are to think clearly about AI, we need to return language to its proper work: description grounded in function, not analogy. Not “What does the model believe?” but “What distribution is it sampling from?” Not “How can we make it reason better?” but “What scaffolds reduce compounding error in token prediction?” These are less exciting questions. But they are clearer, and clarity is the first requirement of responsible use. If you squint, you can see a clear and rigorous curriculum emerge from this misuse and confusion, since the technology is most definitely here to stay.

In summary, when an LLM gets something wrong, it’s not failing. It’s succeeding on different terms than we assume. And the more reasoning steps you introduce (via techniques like chain-of-thought prompting), the more space you create for this kind of drift. Statistical artifacts compound. Thus, in the end, the system may sound "more rational" while becoming less reliable.

And so we chase “alignment” like a ghost, all while misnaming the phenomenon we observe. We seek fixes for hallucinations instead of acknowledging that our models are epistemically hollow. They produce not knowledge, but coherent, convincing, and dare I say, human, noise.

Dear Reader: Thank you for the recent influx of subscribers to the newsletter. I endeavor to continue delivering essays and short-form posts worthy of your attention. I hope to engage with you all in the comment threads to learn how you're thinking about these topics.

Stay tuned, we have a few posts building on this one. We'll be discussing the evaluation benchmarks and going through some mental models for practically evaluating AI responses for educational purposes.

Originally published on the Eduaide.Ai Blog.

A new paper (Chen et al., 2025) claims that Generative AI tends to reproduce outdated pedagogy by default, but can be steered, if intentionally designed, to reflect contemporary instructional values like student autonomy, peer collaboration, and dialogic learning.

Their controlled experiment evaluated 90 AI-generated lesson plans from three conditions: vanilla GPT-4, MagicSchool, and School AI. Each output was coded using Vaughn’s framework of student agency (dispositional, motivational, positional) and Alexander’s taxonomy of classroom talk (rote, recitation, exposition, discussion, dialogue). This demonstrated that current outputs lean heavily toward teacher-centered instruction. That is, instruction dominated by rote tasks and expository talk. Their prompt-engineered prototype shows measurable improvements across both domains—student agency and classroom dialogue.

This is the kind of contribution I like. The paper identifies a real problem and proposes a plausible, actionable solution. Their coding framework is rigorous. Their design intervention is concrete. There is obvious utility here for tool developers. In fact, we’re developing some tooling for Eduaide based on this critique.

But the solution, while practical, is also partial. The paper assumes that AI can be redirected toward ideal pedagogy through better prompt engineering. I expect we’ll see gains from this approach, but they will be limited and, in the end, marginal. We must seriously engage the possibility that some educational values are not computable. In treating AI limits as a design problem, we neglect the epistemological one right under our noses. In short, the paper underestimates the irreducibility of teaching to algorithmic systems. Again, there are simple design moves we can and should make to improve outputs, but the fundamental limitations facing AI in Education are left relatively unchanged.

The bias of the tool reflects the bias of the practice. Pedagogical bias is sociological in nature. It arises from accountability structures in the institution, class sizes, standardization, professional incentives, and a bevy of teacher effects. Sure, the paper acknowledges the need for teacher involvement, but treats it as a supplement to tool design. I’d argue it is the precondition for any successful implementation. A tool is a mirror of its uses.

So, yes, let’s build better prompts. But let’s also be clear-eyed about the limits. Not just of AI, but of the broader system it reflects.

References

Alexander, R. J. (2008). Towards dialogic teaching: Rethinking classroom talk (4th ed.). Dialogos.

Chen, B., Cheng, J., Wang, C., & Leung, V. (2025, in press). Pedagogical biases in AI-powered educational tools: The case of lesson plan generators. The Social Innovations Journal.

Vaughn, M. (2020). What is student agency, and why is it needed now more than ever? Theory Into Practice, 59(2), 109–118. https://doi.org/10.1080/00405841.2019.1702393

Too Long; Didn’t Listen: Brandon Hendrickson on Wonder, Storytelling, and the Forgotten Tools of Learning

For those of you who don’t have an hour for the full conversation, here’s the “Too Long; Didn’t Listen” of my episode with Brandon Hendrickson. Brandon is the creator of Science is Weird, a science program for kids built around the idea that nothing is boring (even the boring stuff), everything is weird, and wonder is worth preserving. I write preserve because after talking with Brandon, I came away possessed with the idea that wonder is something we can lose, but also something we can learn to recover. The primary methods of this recovery, if you are to take Brandon or his major influence, the educational philosopher Kieran Egan, as a model, are found in story, metaphor, and play. In these, we do not merely convey information about the world. The framing of the story, the metaphor, and the game create their own context.

What follows are some of the key concepts from the episode.

§1 Curiosity alone is not enough.

1.1. We often treat curiosity as a default state, but Brandon argues that unstructured appeals to student curiosity can lead to chaos and shallow learning. Instead, curiosity should be cultivated, not assumed, and that means making subjects shiny again. That’s his word for learning that’s emotionally resonant, rich in meaning, and intellectually alive.

§2 Teaching, by its nature, must be adaptive and as varied in approach as the needs of the students.

2.1. The best teaching doesn’t always look progressive or classical or fully guided or like that of a Montessori or Waldorf school.

2.1.1. Throughout his work and indeed throughout our conversation, Brandon draws from an array of educational traditions—Montessori, Classical, Waldorf, Unschooling—but resists dogma. He critiques modern classical education for giving kids adult-level content before they’re ready, and progressivism for focusing too much on student-led learning without adequate structure. The sweet spot, he suggests, lies somewhere at the confluence of the traditions and in honoring the intellectual tools kids possess from an early age—narrative, metaphor, riddle, and song, which he suggests are more ubiquitous than traditional academic forms like the lecture or socratic dialogue.

§3 Literacy changes the brain, but it isn’t everything.

3.1. While Brandon agrees that literacy is essential to accessing higher-order thinking, he pushes back on the idea that it’s the sole foundation for deep learning. Oral traditions, embodied experiences, and storytelling predate literacy and remain powerful modes of understanding and conveying meaning. He argues that our over-reliance on text may crowd out more emotionally engaging and developmentally appropriate forms of learning, especially for younger students.

§4 Storytelling isn’t fluff—it’s a structure.

4.1. Kieran Egan’s central insight, Brandon says, is that learning follows a narrative arc. Even in technical fields like botany or physics, the best teaching frames knowledge through stakes, characters, and conflict. For example, instead of diagramming flower anatomy, he teaches plant reproduction as a biological drama: trees can’t move, sunlight fries pollen, yet they must reproduce despite the stakes. He suggests that a lesson that frames the narrative arc thus lands and sticks better for novice learners than a worksheet or diagram could.

§5 Is education solved?

5.1. Surprisingly, Brandon says yes, with a caveat. He believes we already know enough to build classrooms where students become deeply educated and genuinely interested in learning. What’s missing isn’t knowledge, but execution: classroom-ready curricula, teacher training, and instructional tools that make knowledge about teaching and learning usable at scale.

5.2.Brandon laments the decline of shared storytelling, riddle culture, and joke-telling as vehicles for deep cognitive development. These weren’t seen as educational in the classical tradition, but they offered powerful, emotionally anchored ways for children to make meaning. He calls for a return to these forgotten tools, not in opposition to academic rigor, but as its foundation.

In this wide-ranging conversation, Brandon joins Thomas to explore why curiosity is not enough, what most people get wrong about progressive and classical education, and how stories, songs, riddles, and emotion shape cognition. They discuss why the trivium isn’t built for children, how literacy rewires the brain, and what’s missing in the way we teach science, history, and language.

The conversation also touches on Brandon’s experiences teaching from his apartment, his evolving view of homeschooling, his critiques of Montessori and Sayers, and his ambitious effort to build a new, Egan-inspired curriculum that reconnects students with the “shiny” richness of reality.

Read an enhanced transcript or the TL;DL summary at Eduaide.Ai.

Recorded 25 March 2025

In this conversation, Katie joins Thomas to probe the hopes and fears we project onto new technologies and how to ask more nuanced questions about our relationship with tech than “How much screen time is too much?”

In this conversation, Dylan joins Thomas to explore what makes for effective math instruction, why curriculum fidelity has become a concerning trend, and how teachers can balance procedural fluency with conceptual understanding. Drawing on his classroom experience and reflective practice, he discusses everything from the limitations of administrative oversight in education to specific teaching strategies like pre-teaching, choral response, and retrieval practice.

Kane also addresses why educational blogging has changed over the years, what schools get wrong about communicating with parents, and what Generative AI might mean for the future of computation and math education. Whether you're a math teacher looking for practical approaches, an administrator thinking about curriculum implementation, or anyone interested in how students learn mathematical concepts, you'll find valuable insights in this honest conversation about the realities of teaching.

Check out Five Twelve Thirteen here: https://fivetwelvethirteen.substack.com/

Read an enhanced transcript with helpful links at Eduaide.Ai.

Recorded 27 February 2025

In this conversation, they join Thomas to explore why strong research makes for better teaching, how to identify the most effective study strategies, and what it really takes to improve student success.

Drawing on their shared experiences as researchers, authors, and instructors, they discuss everything from overcoming common barriers to learning (like poor note-taking or misconceptions about memory) to designing courses that build in retrieval practice, spacing, and timely feedback. They also tackle how teacher training can benefit from a deeper understanding of cognitive processes, examine the promises and pitfalls of AI-enabled feedback, and address what personalized learning could look like when done well. Whether you’re teaching in a K–12 setting, designing higher-ed courses, or simply trying to study more efficiently, you’ll come away with powerful insights for turning learning research into real-world classroom practice.

Read an enhanced transcript with helpful links at Eduaide.Ai.

Recorded 09 December 2024

Richard joins Thomas to discuss what makes a well-designed PowerPoint, why transfer is so elusive in education, how to balance cognitive load in instructional design, whether AI-generated feedback can be effective, what virtual reality gets wrong about learning, how prior knowledge impacts multimedia instruction, why game-based learning often misses the mark, whether younger generations are better at processing digital media, how to separate education fads from real breakthroughs, what he’s learned from decades of collaboration, and what he’s working on next.

Read an enhanced transcript with helpful links at Eduaide.Ai.

Recorded 15 January 2025