I know from extensive experience with AI that it is often worthy of critique. It is a powerful tool that must be used wisely. But much of today’s criticism has substituted slogans, suspicion, and rhetorical posturing for wisdom and critical thought.

Human creativity has never been a solitary act.

Every generation has borrowed from teachers, editors, mentors, books, traditions, and technologies.

Artists collaborate with patrons.

Authors collaborate with editors.

Scientists collaborate with colleagues.

Composers collaborate with musicians.

We have always created alongside people, ideas, and tools.

AI is simply the newest collaborator.

Yet much of the public conversation treats collaboration with AI as though it were a moral failure rather than another step in humanity’s long history of extending its own abilities.

The phrase AI slop illustrates the problem.

Sometimes it is an accurate description of lazy, low-quality work.

But far too often it has become an intellectual shortcut.

Instead of asking whether an argument is accurate, insightful, original, or useful, the label itself is expected to end the discussion.

A label are not arguments.

The easiest way to avoid judging quality is to focus on where the work came from instead of whether it is any good.

If we focused more on identifying Human Slop than catching AI Slop, we would be much closer to making an argument.

Once the conversation becomes “Did AI write this?” instead of “Is this thoughtful? Is it true? Is it useful?” critical thinking has already begun to fail.

Ironically, some of the loudest critics of AI exhibit the very habits they claim to oppose.

They dismiss work without engaging it.

They substitute certainty for analysis.

They confuse suspicion with good judgment.

They rely on slogans instead of evidence.

In condemning AI slop, they often produce a different kind of intellectual slop: shallow criticism that mistakes confidence for careful thought.

This is not an argument against skepticism.

Skepticism is an indispensable part of thinking.

AI makes mistakes.

It hallucinates.

It can encourage intellectual laziness when used as a substitute for thinking.

Those dangers are real, and they deserve rigorous examination.

But rigorous examination requires rigor.

Much of our political and cultural discourse in recent years has been shaped by fear and a perceived struggle for survival.

When people believe something essential is being threatened, curiosity often gives way to certainty.

Nuance gives way to slogans.

Disagreement becomes evidence of bad faith.

A similar pattern seems to be driving much of the reaction to AI.

For many people, AI is not merely another technological innovation.

It represents a perceived threat to professional identity, creative achievement, educational traditions, and economic security.

Those fears are understandable.

But fear is a poor substitute for good judgment.

We should be asking:

“How should we use this powerful new tool responsibly?”

Instead, we are asking:

“How do we stop people from using it?”

That is the wrong question.

History suggests that transformative tools are not defeated by condemnation.

They are mastered through wisdom.

Calculators did not eliminate mathematics.

Word processors did not eliminate writing.

Search engines did not eliminate research.

Each changed the nature of expertise.

AI tools will do the same.

The real divide is not between people who use AI and people who do not.

It is between those who use AI to avoid thinking...

...and those who use AI to think more deeply.

The first group should concern us.

The second group represents one of the greatest opportunities in the history of human learning.

Education’s task has never been to preserve old methods for their own sake.

It has been to cultivate judgment.

We should evaluate work by its clarity, reasoning, originality, evidence, and usefulness, not the tools used to create it.

We should become detectors of quality, not detectors of AI.

The obsession with AI detection reveals a deeper problem.

We no longer trust ourselves to evaluate ideas directly.

Judging quality requires knowledge, experience, patience, and humility.

Looking for AI requires only suspicion.

One is difficult.

The other is easy.

The label is easy.

The thinking is harder.

Because “AI-generated” is not an argument.

Because “Human-written” is not an argument.

Quality is the standard.

Until we recover the ability to recognize it, we’ll keep swatting at gnats while swallowing camels.

Tim Moon

Between the Horns

I have been trying to grapple honestly with the role of AI in education, without catering to either side of the current dilemma. Each provides a poor alternative.

This debate appears like a bull. One horn warns that artificial intelligence will permanently weaken human thought. The other promises that it will personalize learning, democratize expertise, and finally deliver the educational revolution we have been promised for decades.

Each horn sees something real. A student who repeatedly hands memory, writing, interpretation, and judgment to a machine will lose something. A faculty that is never exercised will weaken. AI does make it possible to escape difficult thinking faster and more completely than any previous classroom technology.

The promise has substance too. A patient tutor that can question a student at any hour, explain an idea in several ways, expose a weak inference, or generate a serious counterargument is not a small thing. Used with discipline, AI can deepen a student’s encounter with an idea.

But we have seen the larger promise before. The computer, the internet, and the tablet entered school carrying hopes of access, personalization, and liberation. They brought real benefits. They also brought distraction, passive consumption, shallow research, and entire classrooms staring into portals. No machine has ever relieved us of the human work of education.

We are often told to take the bull by the horns. I think that is the wrong move here. The horns appear to point in opposite directions, but they are attached to the same head, and the premise in that head matters more than either prediction.

Go for the Head

Doom and utopia share the same skull. Both assume that AI determines the educational outcome.

In both stories, the teacher becomes a bystander, the student becomes a recipient, and the school becomes a place where technology happens to people. The two camps disagree about whether AI is good or evil, but they grant it the same sovereignty.

AI changes the conditions of learning. But it does not relieve us of deciding what learning is for, which struggles are formative, and who remains responsible for the thought.

That does not make the tool neutral. AI answers quickly, smooths language, resolves uncertainty, and makes completion easier. Tools train the people who use them, and a tool designed to reduce effort will usually do so unless the teacher intervenes with clear instructions, sequencing, and direction.

Once the tool receives most of the causal power, human responsibility thins out. Schools can blame AI for corrupting students. Vendors can credit it with saving them. Both avoid the harder question: What forms of thinking must students still do for themselves, and how will a teacher know that the work was truly theirs? That question leads backward, before AI, to the habits schools had already built.

The Stage Was Already Built

Cognitive offloading is the transfer of mental work to agents outside the person. We offload memory to a note, calculation to a calculator, direction to a map, and judgment to a rule. Offloading is not always foolish. Human civilization depends on tools, records, institutions, and other people carrying work no single mind could carry alone.

Education still has to decide which work develops the student. That work cannot be surrendered without cost.

Long before ChatGPT, schools had begun transferring more of that work away from the learner. Memory moved to the handout and then to search. Judgment moved to the rubric. Argument moved to the five-paragraph formula. Attention moved to the screen. Knowledge became something temporarily retrieved for an assessment and then released.

This history is uneven, and no serious argument should romanticize every older classroom. Traditional schooling often confuses mechanical rigor with cognitive rigor, obedience with learning, and memory with understanding. Still, the larger direction is hard to miss. We built systems that rewarded the appearance of learning while asking too little about the reality beneath it.

A polished paper stood in for understanding. A passing score stood in for durable knowledge. Compliance stood in for intellectual ownership. The student learned to produce the acceptable artifact because it was the ticket to the show.

Then AI arrived.

It did not write the script. It stepped onto a stage education had already built. The scenery was in place. The cues had been given. The role was waiting: produce the paper, complete the worksheet, answer the question, produce the artifact.

We are blaming the actor for performing the role education had prepared for it.

Granted, a late arrival can still make a bad play worse. AI can industrialize cognitive offloading. It can create the appearance of thought without the inconvenience of thought, at speed and in language polished enough to fool both student and teacher. The prior weakness makes the present alarm more serious because the tool can now scale what the school had already normalized.

Friction Is the Point

Intellectual strength grows through resistance. We remember by recalling. We learn to write by struggling to give form to an idea. We develop judgment by making distinctions, living with uncertainty, testing our reasoning, and standing on conclusions.

AI can remove that resistance before the student has fully felt the problem. Ask for an essay, and it writes one. Ask for an interpretation, and it supplies one. Ask it to decide, summarize, compare, or conclude, and it will often perform the very act the student was supposed to practice.

Some friction is merely mechanical. Formatting a citation, correcting a transcription error, or repeating a routine calculation may consume time without deepening understanding.

Formative friction is different. It requires the student to remember, explain, distinguish, defend, and revise.

In-class exams that require students to write out answers to questions are formative for memory. Short in-class essays that capture a student’s feelings, convictions, and patterns in thinking are formative.

Reverse engineering an essay is another formative tool. Providing a well-written essay and having a student write a prompt by hand to reproduce a similar essay can be very demanding.

Teaching students to ask the right questions is as critical as teaching them to provide the right answers. Writing a prompt that can closely replicate such an essay is formative. These are just a few ways to retain and restore cognitive friction in the AI Age.

Asking students to answer questions and defend responses orally also creates formative friction.

This is how seasoned traditional, or classical, teachers from the last century would have handled the challenges of AI in this century. They would not have panicked or been intimidated, nor should we.

The governing principle is friction relocation. Use AI to carry some of the mechanical work, but the thinking effort that develops the person must remain with the student. Grade the process along the way to a polished product, and weigh the process higher than the finished product itself.

Sequencing matters. The student should encounter the question, attempt an answer, recall what is known, and commit to a claim before asking the machine for help. AI can then expose a weak inference, ask for evidence, introduce a counterargument, or test whether the student can explain the idea without borrowed vocabulary.

After the tool has helped, the teacher should be able to ask a simple question: What work remains inside the student? If the answer is only approval, selection, or cosmetic editing, the machine carried too much. When the student must still explain, defend, and take ownership, the tool may have made the work harder in the way that matters. That standard makes the proper role for AI much clearer.

Interlocutor, Not Scribe

The strongest educational role for AI is to be an interlocutor, not a scribe. A scribe produces language on the student’s behalf. An interlocutor meets the student in dialogue and returns responsibility through questions.

A scribe says, “Here is your answer.” An interlocutor asks, “What do you mean? What evidence would change your mind? Where does your claim contradict itself? Can you explain it without the vocabulary you just borrowed? What is the strongest case against you?”

That is closer to the work of Socrates than the work of a ghostwriter. It provides the teacher another instrument for pulling thought from the student, provided it is visible and transparent.

The process must be visible. A finished paper is no longer reliable evidence of the thinking that produced it. Teachers need drafts, oral defense, in-class work, remembered knowledge, conversation, and a record of how a claim changed and developed. The student becomes the object of education, and the finished product becomes one piece of evidence in a larger record.

That leads to an irony schools should face. Using AI well may require less general computer use at school. Access to the machine should be limited and targeted. Initial thinking will often need to happen by hand, in speech, or in an observed classroom setting before AI enters the process.

A consistent written and oral record of each student provides stronger evidence of authenticity than a detector or a single polished paper. AI may help the teacher organize that record, but the teacher interprets the growth, and the student retains ownership of the work. Over time, the accumulated record gives the teacher evidence that no isolated product can provide.

The Harder Place

The space between the horns is not safe. It is the place representing the full weight of a bull.

A blanket ban protects familiar assignments. Broad adoption allows a school to buy access, announce innovation, and accept the vendor’s promises as a theory of learning. Both choices are easier than redesigning the intellectual life of the classroom.

The alternative requires teachers to separate mechanical work from work that forms and reveals thought. As a general rule, formative processes like thesis development, rough drafts, etc, should occur predominantly at school, where the teacher can gauge the process. Mechanical work can be done at home.

This has a price. It will consume class time and slow coverage. Teachers may need to assign fewer papers and examine them more closely. Schools will need time for conversation, handwritten work, and oral defense. Students will lose some of the convenience of completing important work privately. Administrators and parents will have to stop treating a large stack of polished products as proof of learning.

Unless schools are willing to pay that price, responsible use will remain a slogan, and the adults who design school use will continue placing the burden on the tools.

Who Still Does the Thinking?

AI did not introduce cognitive offloading into education. Education had already set the stage. The technology entered a system practiced in mistaking performance for formation, and its default use will deepen that mistake.

The next decision is practical. It concerns when AI enters the work, what it is allowed to do, what evidence the teacher keeps, and which capacities the student must continue to exercise.

Pedagogy determines who still thinks. By pedagogy I mean the full human design of the project or curriculum, the purpose of the lesson, the sequence of effort and help, the limits placed on AI, the evidence a teacher requires, and the responsibility the student retains.

AI should enter after the student has recalled what is known, struggled with the question, and made a first attempt. At that point, it can question, pressure, and extend the student’s thought without becoming its author.

The future of education will be decided by whether schools can recover and document a clear account of the thinking that forms a person. AI arrived on a stage that education already built. What happens next is still our responsibility.

When you type a question into an AI model, you usually see only the finished answer. A few seconds pass, words appear on the screen, and the machine seems to have understood what you meant.

Beneath that exchange are three problems older than the machine:

• What do the words mean within this context?

• Does the logic hold together?

• How should the answer be presented to this particular reader?

Grammar. Logic. Rhetoric.

For centuries, these three disciplines formed the Trivium, the foundation of classical education. Grammar taught students to understand language and the knowledge carried through it. Logic taught them to test relationships, follow an argument, and recognize contradiction. Rhetoric taught them to communicate truth persuasively to another person.

The Trivium is often treated today as an antique artifact, something preserved in classical schools along with Latin conjugations and old books. But its basic architecture has returned in one of the newest technologies on earth.

The Trivium is often treated today as an antique artifact, something preserved in classical schools along with Latin conjugations and old books. But its basic architecture has returned in one of the newest technologies on earth.

I do not mean that the engineers who designed modern artificial intelligence secretly followed a medieval curriculum. Such a historical claim would be way too neat to be true. The connection is deeper than direct inheritance. Artificial intelligence has been built to perform, in mechanical form, the same three kinds of language work that the Trivium once taught human beings to perform with discipline.

The machine receives language. It identifies relationships within that language. It produces language fitted to an audience and purpose. The old architecture has appeared inside the new machine.

Grammar before thought

Grammar has been reduced in many minds to punctuation, spelling, and the correction of embarrassing mistakes. Those things matter, especially when the mistake is ours. But classical grammar was larger. It concerned the structure of language and the accumulated knowledge preserved through words.

Before we can reason about something, we must know what its terms mean. We must recognize the symbols, distinctions, definitions, stories, and facts from which thought begins.

Grammar gives the mind something to think with. It provides the ingredients of thought.

Modern language models begin with a mechanical version of this problem. They break language into tokens and learn statistical relationships across immense collections of text. Systems descended from the Transformer architecture use attention to weigh relationships among elements in a sequence. Models such as GPT-3 were trained to predict what token should come next based on the language preceding it.

That is not grammar as a human being experiences it. The machine does not sit with a word and wonder what it means for a life. It does not hear a mother’s voice inside an old expression or discover that a familiar term has concealed an assumption. It learns patterns of use with extraordinary breadth.

Still, the resemblance matters. The machine’s capabilities begin in its encounter with the structure of language and the knowledge carried there. Without that foundation, there is nothing to reason about and nothing meaningful to say. The classical teachers understood the order. Grammar comes first.

We have become less patient with that part. Students often want to express an opinion before they understand the terms of the question. Adults do it too. We encounter a headline, recognize a few familiar words, and move directly toward judgment. The difficult work of definition, context, and accurate reading feels slow compared with the speed of response.

Now we have a machine willing to perform that response for us. But if we do not do the requisite thinking before turning to the machine, it may lack the context needed to answer with much depth. As is often said, garbage in, garbage out. What often goes unsaid is the reverse: when quality comes out, some quality may have gone in.

The difficult middle

Logic stands between grammar and rhetoric for a reason. It is the discipline that asks whether the movement from what we know to what we claim can be trusted.

Does the conclusion follow? Are two statements actually contradictory? Has evidence been confused with illustration? Has a cause been established, or merely placed beside an effect? What assumption must be true for the argument to work?

These questions are less visible than grammar and less impressive than rhetoric. Logic rarely produces the sentence someone wants to share. It slows us down. It makes us return to a definition, surrender a convenient claim, or admit that the evidence does not carry as much weight as we hoped.

Artificial intelligence can perform many tasks that look like reasoning. Large language models solve problems, compare arguments, identify patterns, and sometimes expose contradictions that a reader missed. Their success is real. Their inconsistency is real too. A model may construct a careful chain in one response and confidently break that chain in the next.

Plausibility and truth often travel together, but they are not the same thing, and human beings confuse them too.

A polished sentence can make a weak proposition feel strong. A familiar phrase can carry a bad assumption deceptively past our notice. An argument can succeed rhetorically while failing logically, especially when it tells us something we already wanted to believe.

The machine did not invent this weakness. It learned from us.

That may be one of the more uncomfortable gifts AI has given education. It has made the difference between fluency and thought difficult to ignore. A student can now produce a clean paragraph without understanding the source, the terms, or the reasoning beneath it. The artifact may look finished while the mind that submitted it has barely begun.

The problem reaches beyond a machine thinking for the student. The student may never learn to recognize what thinking requires.

Rhetoric arrives early

Rhetoric is the art of fitting language to a purpose and a person. It considers the audience, the moment, the available evidence, and the response the speaker hopes to produce.

This is where artificial intelligence can appear most human. It can explain quantum mechanics to a child, rewrite an email for an employer, soften a disagreement, imitate a public figure, or turn a page of scattered notes into a persuasive argument. It adjusts tone, structure, vocabulary, and emphasis in seconds.

This ability did not emerge from scale alone. Research such as OpenAI’s work on training language models with human feedback used human demonstrations and rankings to make model responses more helpful and better aligned with a user’s intent. Human beings helped teach the machine which answer sounded more appropriate to another human being.

That is a form of rhetorical training.

The result is a machine capable of producing rhetoric before the person using it has necessarily gathered the grammar or tested the logic. This reverses the educational order. The finished expression arrives before the understanding that should govern it.

You have probably felt this temptation. The machine produces a sentence better than the one you were struggling to write. It is clearer, tighter, and ready to use. The relief is immediate. If you have ever accepted that sentence because it sounded better than yours, you know how easy the exchange can feel.

But what exactly has been relieved?

Sometimes it is merely mechanical labor. The thought is yours, the reasoning is sound, and the machine has helped clear away awkward language. That can be a genuine gift.

At other times, the struggle with the sentence was exposing an unfinished thought. We could not explain the idea because we did not yet understand it. When the machine supplies the fluency too soon, it may conceal that absence from everyone, including us.

Rhetoric has outrun logic, and the final product can look literate while the process beneath it is not.

Recovering the order

Much of what is now called AI literacy focuses on learning how to use the tools. Students are taught to write prompts, compare outputs, protect private information, check sources, and recognize fabricated claims. These are necessary skills. They are not enough.

Tool literacy teaches us how to operate the machine. The Trivium teaches us how to remain responsible for what comes out of it.

• Grammar asks us to slow down before interpretation. What do the important terms mean? What information has been provided? What context is missing? Which sources deserve trust? What has the machine assumed?

• Logic asks us to examine the movement. Does the conclusion follow from the evidence? Are alternatives treated fairly? Has correlation become causation? Does the response contradict itself or quietly change the meaning of a term?

• Rhetoric asks us to examine the effect. Who is being addressed? What does the answer want the reader to believe or do? Which facts receive emphasis? Which disappear? Does the language clarify reality, or merely make one position easier to accept?

These questions belong together. Grammar without logic collects information without testing it. It offers ingredients without a recipe. Logic without rhetoric may discover truth without making it heard. Rhetoric without either can sound convincing to an untrained mind. Artificial intelligence has made that third possibility cheap, which is why recovering the Trivium matters now.

A higher literacy

For generations, literacy meant the ability to read and write. Those abilities remain essential, but they no longer describe the whole responsibility. A machine can now produce competent prose faster than most human beings. If literacy is measured only by the finished paragraph, the machine will appear increasingly literate while the person using it becomes harder to evaluate.

The future requires a higher standard. We must be able to understand the language placed before us, test the reasoning beneath it, recognize the rhetorical forces acting upon us, and decide what we are willing to claim as our own.

We must learn to use a powerful language machine without confusing its fluency with truth or its assistance with authorship.

The machine can arrange language without living inside its meaning. It can reproduce an inference without bearing responsibility for its consequences. It can persuade without believing what it says. Those differences place a responsibility on us that no increase in machine capability can remove.

The deeper opportunity of artificial intelligence may not be that it will finally teach machines to think like human beings. I am not sure we understand human thought well enough to know what that claim would mean. The opportunity may be that the machine will force us to recover disciplines we neglected when human fluency was harder to imitate.

We built a machine from the accumulated products of language, reasoning, and persuasion. Now the machine returns those products to us almost instantly, polished and detached from the struggle that once made their origin easier to see.

That should change how we teach.

Students should not merely learn to obtain an answer from AI. They should learn to interrogate the language, reconstruct the reasoning, examine the persuasion, and decide whether the answer is true and deserves to survive. The machine can participate in that education, perhaps by asking questions, exposing contradictions, or providing arguments to test. But it cannot assume the student’s responsibility for judgment without weakening the very faculty education is meant to form.

The Trivium is not a retreat from the future. It is preparation for it.

The next time a machine gives you a fluent answer, do not begin by asking whether it sounds intelligent. Ask what its words mean, whether its reasoning holds, and what its language is trying to accomplish in you.

Grammar. Logic. Rhetoric.

The future of human literacy may depend on remembering their order.

The sense that you are just going through the motions haunts many teachers. It haunted me.

Every t crossed. Every i dotted. Every grade entered, every report filed, every requisite duty completed. The work was done, but I could not help wondering if any of it actually mattered.

I entered teaching for the art of it—the romance of making a difference in a single life. Yet much of the job had become the cold, mechanical work of tracking hundreds of lives and managing the compliance that came with them.

Now, AI has arrived, bringing with it more tools to learn, more policies to draft, and a constant, low-humming threat of student attention loss and cheating.

Yet, there is a strange irony in this digital storm. The very tool that arrived as another administrative burden may in turn be our escape hatch from mechanical drudgery. And AI, arriving as a threat, may in fact provide a unique source of relief.

But a machine cannot distinguish mechanical burden from formative teaching on its own. It has no soul, no taste, and no memory of what it feels like to struggle. These machines require access to the teacher’s reasoning and values to know which difficulties to allow and which to prevent.

Accessing this requires legibility.

Beneath the Artifacts

A lesson plan reveals what should happen. A grade records a conclusion. A platform records compliance. But the teacher’s judgment lives beneath all of these artifacts. It is the foundation of the questions that matter, the awareness of a quiet student, the instinct of when to automate and when to offer a direct human touch.

Our human colleagues learn this by working beside us. They watch, ask, notice, and remember. AI, however, sees only the context we choose to show it. Unwritten judgments leave empty spaces, and software always fills empty spaces with its own generic defaults.

If we want to govern AI, we must make our patterns legible to it.

Key Idea: If we leave our professional judgment unwritten, software will fill the empty spaces with its own sterile defaults. We must become legible to the machine to remain in command of it.

Link 1: Making the Teacher Legible

A Human Operating System (HOS) provides the machine with a map of the teacher’s judgment. It is not a collection of clever prompts; it is an exercise in professional self-awareness. It is composed of durable, structured files:

• Vocation: What the work is ultimately for.

• Convictions: The principles governing our most difficult classroom decisions.

• Boundaries: The sacred aspects of learning that efficiency must never be allowed to touch.

• Voice and Method: How we communicate and how we actually work.

• Current Context: A real-time picture of the classroom as it exists today.

Honestly articulating these elements allows the machine to operate closer to the human collaborator. This closeness does not make the machine inherently correct, but it makes the machine more accountable to the teacher’s stated identity and goals.

Written down as a set of durable markdown files, this judgment persists across platforms. It ensures that when the AI assists us, it speaks in our cadence and guards our values.

Link 2: Making the Student Legible

The same principle of legibility offers a way forward for our students. It shifts our stance from defensive policing to active development.

A single paper demonstrates very little in the development of a student. It may be unusually strong, rushed, heavily coached, or written under duress. A student’s voice is not a performance captured in a single, high-stakes assignment; it is a trajectory revealed over time, across months and years.

To capture this trajectory, we must return to the physical world. Regular, handwritten, in-class writing provides authentic baselines of what a student can presently do.

How it works without the burnout: To prevent this from becoming a new administrative chore, we use the machine as our clerk. The teacher quickly photographs these handwritten pages; the AI transcribes, dates, and archives them behind the scenes. The goal is not to grade every page, but to preserve unforced evidence of development.

Across time, this longitudinal record reveals what no single essay can:

• A student’s increasing clarity and control over language.

• The subtle development of sentence structure and vocabulary.

• A growing ownership of claims and evidence.

• A movement from mere correction toward purposeful revision.

• The gradual emergence of a recognizable, authentic voice.

An AI can preserve, retrieve, and compare more of this developmental record than a teacher can hold in memory for a hundred different students. And then, the student can be compared to their own past self, rather than an arbitrary, easily-gamed AI detection device.

Once a student is progressively legible, we can offer AI assistance that is proportionate to their development. It changes the defensive question—“May a student use AI?”—to a pedagogical one: “What kind of help serves this student’s growth right now?”

• For a novice writer: The AI might ask clarifying questions without supplying the answers.

• For an intermediate writer: It might identify a weak stylistic pattern without rewriting the sentence.

• For an advanced student: It might act as a devil’s advocate, testing an argument against fierce opposition.

The help grows only as the student’s personal ownership grows.

Student Operating Systems in Progress

The ultimate goal of capturing student work is to provide useful evidence for teachers, students, and parents—those who are actively responsible for driving formation.

• For the individual student: The system can surface a hidden strength, isolate a concept requiring attention, and suggest the next appropriate challenge.

• For the class: It can reveal shared blind spots without ranking individuals—showing the teacher that tomorrow’s lesson needs to address a recurring grammatical error, a weak transition, or a misunderstanding of evidence.

• For parents: The AI can assist in preparing concise, highly specific updates grounded in actual work: what has improved, what needs attention, and what teacher and student are targeting next.

This longitudinal approach is not limited to writing. It applies wherever process matters: in tracing mathematical reasoning steps, tracking scientific observations versus inferences, or preserving the iterative revisions in art and music.

Regaining the Offensive

It is a striking irony. The very tool that threatened to make student authorship invisible is the tool that can make the trajectory of their growth completely clear. And the tool that came with fears of added mechanical drudgery can actually spell relief for teachers.

By documenting our own Human Operating System, we stamp the machine with our values. And by capturing our students’ progress, we protect them from generic algorithms and make their genuine development visible to them, their parents, and their teachers.

We do not need to cede the classroom to the machine. By making our values and our students’ progress legible, we regain the offensive.

Enjoyed this post? Share it with other educators navigating the AI shift, or leave a comment with your thoughts on “legibility” in your own classroom.

I am passionate about AI and education. If you share either of these passions, let’s talk. 757 746 2931. timsmoon@gmail.com

By Tim Moon. Reading Time: ~5 minutes

Behind the blinking cursor sits a folder of plain, carefully developed text files. Vocation. Convictions. Boundaries. Voice. Each holding part of me. And each allowing the machine access to what it otherwise could not see: My mission, my most critical beliefs, the lines I refuse to cross. And how I genuinely sound when the words are truly mine.

The machine can read those files in seconds, but the writing and development, which I had to labor over, took days. Not because of the number of words required, but because of the precision they demanded. I had to build foundational files of identity for the machine to wrap itself around to approximate me.

This required digging convictions out of instinct, and testing each for accuracy. Removing the appearance of wisdom without its weight. I had to decide where help becomes interference, where ease deforms, and where a useful tool asks too much in return.

It meant turning what I carried silently into words another intelligence could find. Only then could the machine, after careful brainstorming and dialogue, create something roughly approximating me.

Agent Legibility

I came across a name for part of this while reading OpenAI’s essay, “Harness Engineering: Leveraging Codex in an Agent-First World.” OpenAI’s engineers were building software with Codex agents. Their early problem was an underspecified environment. The agents could write code, but they could not use knowledge, inspect behavior, follow boundaries, or correct failure unless those things were made available to them.

So the engineers changed the environment:

• They made the application visible.

• They exposed logs and tests.

• They placed knowledge in durable, structured documents.

• They used a short file as a map into deeper sources rather than cramming every instruction into one swollen manual.

They called the goal agent legibility.

The phrase stayed with me.

Their subject was software. Mine is me, the human being working with the machine. The distance between those subjects matters. A repository can be mapped. A person transcends maps. Yet the same difficulty appears in both: one intelligence cannot work faithfully with context it cannot reach.

Human collaboration has always depended on invisible context. People can communicate context with eye contact and facial expressions and phrases. History, judgment, standards, memory, taste, and relationship make this possible. But AI cannot access this context.

AI did not create the need for legibility. It revealed how much of our work depended upon invisible context.

This matters even more as frontier models improve. The durable advantage of clever prompting is diminishing. But a person who has made his judgment, methods, and standards visibly present for the model provides a consistent, durable starting place—a structure that prevents continuous reconstruction for new conversations. Whatever remains sealed in my head may guide me; it cannot guide the machine.

The Durable Layer

Prompts offer temporary relief. I can pack more explanation into the box, add rules, paste examples, and try again. Tomorrow the box is empty. The work begins with another compressed account of who I am and what I mean.

I got tired of having to explain myself from scratch every morning.

Durable context changes this.

My Human Operating System separates what changes slowly from what will soon rot:

• Layer I: The Durable Layer (Vocation, Convictions, Boundaries) — Changes very slowly, surviving the transition of models and years.

• Layer II: The Developmental Layer (Voice, Method) — Evolves alongside personal craft and skill.

• Layer III: The Passing Memory (Current Projects, Handoffs) — Transitory, perishable, and rewritten freely.

A new model may arrive next year. But the files remain plain enough to travel, making the system model-agnostic.

The model changes. The files do not. And the files provide legibility.

Proximity vs. Authenticity

Legibility does not mean perfect knowledge or memory; it merely creates the conditions for proximity.

Proximity is not certainty. It does not perfect a draft. It does not ensure the language perfectly captures my voice. It asks a smaller question: How near did the machine come?

When the machine can read the files designed to capture me and my work, any draft I ask for gets closer to me, better capturing my sound, my habits, and my concerns. Nearness means closer to me. Proximity is not a solution; it is a step closer to the final product. It paves the way for me to perfect the path.

I must polish the draft to bring it closer to authenticity. It still requires cutting, rebuilding, restoring, and replacing.

The machine can approach a voice, but cannot perfectly capture it. But a machine armed with the knowledge of me can legibly approximate me—moving the stones closer to their final setting.

Refining representation allows me to stand inside the work. It completes ownership. Stamp my name on it, a label of responsibility.

That ownership extends beyond style. Representation is moral ownership. I answer for the reasoning, the evidence I selected or ignored, the claims I left out, and the consequences of what I ask the reader to believe or do. A sentence may sound exactly like me and still fail to represent me.

Voice is resemblance.
Representation is responsibility.

This is where legibility turns back on the person who seeks it. The Operating System teaches the machine who I am, which I must first accurately identify. The files increase proximity because their development requires honesty.

The Shadow of Legibility

This is also where the argument can go wrong. Legibility has a shadow. Schools, governments, and organizations can make people legible so they can be more easily sorted, measured, predicted, and controlled. A life reduced to fields becomes easier to process and harder to see.

The demand for complete legibility would turn a person into a profile built for the convenience of a system. But that price is too high.

Human legibility must remain chosen, partial, and bound to a purpose. The files serve the person. The person does not owe the machine an exhaustive account of himself. Privacy protects more than information. It protects the unfinished, the sacred, and the parts of a life that lose something when forced too soon into words.

The Human Operating System begins from ground deeper than productivity.

Worth precedes performance.
Formation outranks efficiency.
Judgment remains human.

These convictions govern the use of the system because the system cannot govern them.

The cursor still blinks. The folder cannot contain the person sitting before it. It can preserve what he has learned to say, what he has chosen to protect, and where he intends to stand.

The machine can come near.

But I must make the words mine.

Our culture’s abandonment of reading and critical thinking is the deepest impediment to the effective use of AI. It is the chief cause of the widening AI literacy gap.

That claim may seem strange at first.

A majority of the discussions about artificial intelligence seem to focus on technology. We are preoccupied with access, training, software, devices, and infrastructure. We assume the people who will benefit most from AI are those with the strongest technical skills.

Increasingly, I am convinced the opposite is true.

The AI literacy gap is not primarily a technology problem. It is a language problem.

The most effective AI users are rarely the people who know the most about computers. They are often the people who can read carefully, think clearly, ask good questions, and exercise sound judgment. They know how to define a problem, evaluate a response, detect an error, and refine a line of inquiry.

These are not technological skills. They are intellectual habits.

Unfortunately, these are the very habits our culture has spent decades neglecting. A reality that started long before the introduction of AI.

A society built on television, short-form content, constant distraction, and immediate answers does not naturally produce deep readers or careful thinkers. It produces skimmers. We consume fragments rather than arguments. We react emotionally rather than reflecting intellectually. We have access to more information than any civilization in history, yet we fail to sustain attention long enough to apply it wisely.

AI has both magnified and exposed this problem.

A thoughtful student can use AI as a tutor, debate partner, editor, and research assistant. A careless student can use the same tool as a machine for avoiding thought.

The difference is not the technology. The difference is the person using it.

This is why the ancient liberal arts traditions suddenly feel relevant again.

They are more than archaic words and dead traditions. They are the bedrock of thought and wisdom.

For centuries, the disciplines of grammar, logic, and rhetoric formed the foundation of education. Grammar taught students to understand language. Logic taught them to reason carefully. Rhetoric taught them to communicate truthfully and persuasively.

These disciplines were not relics of a pre-technological age. They were training in how to think.

Ironically, the rise of AI has made them more valuable, not less.

For years, modern education has assumed that advancing technology would make traditional literacy less important.

AI suggests the opposite.

The more sophisticated our technology becomes, the more valuable reading, reasoning, and persuasion become.

The future will not belong to those who merely have access to intelligent machines. It will belong to those who can direct them wisely.

In an age when answers are abundant, the decisive advantage belongs to people who can ask better questions, evaluate competing claims, and communicate with clarity and purpose.

We often describe AI as artificial intelligence. In practice, it may be better understood as a mirror.

It reflects the strengths and weaknesses of the person using it.

Good readers become better researchers. Good thinkers become better problem solvers. Clear communicators become more effective creators. A large language model can contribute to capability, but the quality of the outcome remains deeply rooted in the quality of the mind directing it.

AI is a powerful amplifier. It amplifies intelligence, creativity, and judgment. It also amplifies confusion, laziness, error, and a lack of depth.

Our challenge today, instead of building a better machine, is to use the machines better.

It amplifies AI literacy and AI illiteracy alike.

This requires our returning to the liberal arts traditions that Western education has been rooted in since Socrates.

https://collaboration-manifesto.vercel.app/

I have spent the last several years watching the conversation around artificial intelligence and creativity with increasing frustration.

It’s not that people disagree.

Disagreement is healthy.

And it’s not because some artists reject AI.

Every artist should be free to choose their own methods.

What frustrates me is something deeper.

The conversation has become surprisingly shallow.

On one side are the evangelists.

AI will write the books.
AI will create the art.
AI will replace the artists.

On the other side are the purists.

AI corrupts creativity.
AI destroys authenticity.
AI derails critical thinking.
AI-assisted artists are somehow cheating.

Both sides flatten a complicated reality into slogans.

As someone actively creating with these tools, neither description resembles my experience.

The more I work inside these systems, the more convinced I become that we are asking the wrong questions.

We keep asking:

“Did you use AI?”

When we should be asking:

“Is the work any good?”

We keep investigating the process before evaluating the result.

And increasingly, the process itself has become a kind of moral test.

AI has become a Scarlet letter, or letters.

Nathaniel Hawthorne understood something about human nature.

Society tends toward creating artificial markers of purity and impurity.

The specific symbols change.

The behavior remains remarkably consistent.

A person becomes identified by a visible sign rather than the quality of their character.

A category.

A label.

An approved or disapproved status.

Today, I see something similar emerging in parts of the creative world.

Before discussing a novel, people want to know whether AI was involved.

Before discussing a painting, people want to know whether AI was involved.

Before discussing an essay, people want to know whether AI was involved.

The tool has become more important than the work.

The process has become more important than the result.

And that should concern anyone who cares about art and creativity.

I have no objection to artists rejecting AI.

None.

If a novelist wants to write every draft by hand, wonderful.

If a painter wants to avoid digital tools, wonderful.

If a musician wants to record only acoustic instruments, wonderful.

Creative preferences are part of artistic identity.

The problem begins when personal preference becomes public orthodoxy.

The problem begins when a creator’s chosen method becomes a universal standard imposed on everyone else.

The problem begins when legitimacy is granted not by artistic merit but by compliance with approved methods.

The irony is that creativity has never been solitary.

Every meaningful work emerges from collaboration.

Teachers.

Editors.

Mentors.

Critics.

Researchers.

Friends.

Spouses.

Traditions.

Books.

Conversations.

Influences.

No artist creates in isolation.

The myth of the solitary genius was always incomplete.

What has changed is not collaboration itself.

What has changed is that a new kind of collaborator has entered the room.

And suddenly, many of the assumptions we held about authorship are being exposed.

My own experience illustrates the complexity.

Years ago, I published my first novel, Tomato Fields.

No generative AI.

No collaborative writing systems.

No creative operating system.

Just me, a keyboard, and a story.

Today I am completing the sequel, The Watchers.

This time, I am working inside a carefully designed creative ecosystem.

I write the novel.

I write every chapter.

I make every creative decision.

But surrounding the novel is a system of research files, continuity tracking, voice frameworks, project architecture, version history, field notes, and intelligent tools that help me organize, remember, challenge, and refine my thinking.

The result is no less human.

In many ways, it is more intentionally human.

Because the work forces me to think more clearly about what actually belongs to me.

My judgment.

My taste.

My intuition.

My vision.

My voice.

The deeper question is not whether AI can create.

The deeper question is whether human creativity is best understood as an isolated act at all.

What if creativity has always been an ecosystem?

What if artists have always been conductors rather than solo performers?

What if the future belongs to people who learn how to collaborate wisely with machines?

That is why I am launching The Collaboration Manifesto.

https://collaboration-manifesto.vercel.app/

I am not trying to defend machines or promote technology.

Nor am I trying to convince every artist to use AI.

I am launching it because I believe creative freedom matters.

I believe artists should remain free to experiment.

I believe creators should be judged by the quality of their work rather than the approval status of their tools.

I do not believe collaboration is corruption.

The future of creativity will belong to those who understand how to build meaningful relationships between people, ideas, traditions, tools, and imagination.

The Collaboration Manifesto is my attempt to begin that conversation.

It’s not about machines.

It’s about creators, and their freedom to create.

https://collaboration-manifesto.vercel.app/

• Thank you for reading this essay. If it moves you, let’s continue the conversation. Drop me an email or give me a call. timsmoon@gmail.com, (757) 746-2931

“mentions”: [

{

“@type”: “Thing”,

“name”: “The Collaboration Manifesto”

},

{

“@type”: “Thing”,

“name”: “Creative Freedom”

},

{

“@type”: “Thing”,

“name”: “Creative Ecosystems”

},

{

“@type”: “Thing”,

“name”: “Collaborative Creativity”

}

]

Detection software. Honor-code revisions. Assignments redesigned to be harder to outsource. Every bit of it treats AI as a threat to contain. Almost none of it asks what it looks like to teach a student to use the thing well.

There are two ways a student uses AI, and they aren’t close. One quietly does the thinking for them. The other turns it into something to argue with, push back against, and use to sharpen work they actually produced. The first erodes the very skills school exists to build. The second is a skill in its own right — and it’s the one almost no one is teaching.

I’m hosting a free one-hour seminar for teachers, department heads, and administrators on the second way: how to fold AI into your teaching honestly, design work that survives it, and model the thinking-partner habit for students directly.

AI Without Cheating — for Teachers Tuesday, June 9 · 11:00 AM Central · Live online · Free Register: https://seminar.critical-ai-solutions.com/ai-for-teachers

If you’re responsible for an AI policy, an anxious faculty, or just your own classroom, this is a starting framework you can actually stand behind.

• Thank you for reading this essay. If it moves you, let’s continue the conversation. Drop me an email or give me a call. timsmoon@gmail.com, (757) 746-2931

Many schools are discovering that they spent years assessing products that can now be generated in seconds.

That realization creates understandable anxiety.

The public conversation tends to focus on visible concerns:

• plagiarism

• AI detection

• policy

• compliance

Those concerns matter.

But they are only the surface of the issue.

The deeper challenge is educational.

If AI can generate essays, summaries, explanations, and analyses on demand, schools are forced to reconsider questions that have quietly existed for a long time:

What deserves assessment?
What counts as authentic work?
What is the role of the teacher?
What does it actually mean to learn?

In some ways, AI is not creating an educational crisis.

It is exposing one.

For decades, many systems have unintentionally rewarded polished outputs more than the slow development of judgment, understanding, and intellectual maturity.

Now the distinction is becoming impossible to ignore.

Ironically, this may become an opportunity rather than merely a threat.

As AI removes more mechanical friction, schools may have the chance to focus more intentionally on:

• dialogue

• reasoning

• oral defense

• wisdom

• discernment

• authentic human formation

The real disruption is not that AI can produce answers.

It is that schools must now think more deeply about what education is ultimately for.

• Thank you for reading this essay. If it moves you, let’s continue the conversation. Drop me an email or give me a call. timsmoon@gmail.com, (757) 746-2931

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "AI Is Not an EdTech Purchase",
  "description": "Artificial intelligence is forcing schools to reconsider assessment, authenticity, learning, and the purpose of education itself.",
  "author": {
    "@type": "Person",
    "name": "Tim Moon"
  },
  "about": [
    "Artificial Intelligence",
    "Education",
    "Assessment",
    "Pedagogy",
    "Human Formation"
  ],
  "keywords": [
    "AI in education",
    "assessment",
    "pedagogy",
    "authentic learning",
    "teacher role",
    "AI and schools",
    "educational philosophy"
  ]
}
</script>

.

If you can dream—and not make dreams your master;

If you can think—and not make thoughts your aim;

— Rudyard Kipling, If—

Kipling wrote these lines as practical wisdom to his son.

The whole poem is about refusing to be ruled by the activities of your own mind.

Do not be mastered by your dreams. Work to make them real.

And do not mistake the appearance of thinking for the reality of judgment.

Kipling could not have known how literally, a century later, we would build machines

capable of producing thoughts on demand.

There is a difference between thinking and producing thoughts.

Many selling artificial intelligence tools proceed as though there is not.

But thoughts are products, and thinking is presence.

A page of analysis is a thought.

The slow, often uncomfortable formation of judgment that the page was meant to represent is thinking.

You can buy the first one cheaply.

The second remains expensive.

This is not a complaint about technology.

The technology is doing exactly what it was designed to do.

The problem is cultural.

We are increasingly encouraged to substitute mechanical artifacts for forms of human formation that cannot be automated.

AI did not create this confusion.

It has exposed it.

And nowhere is that exposure more revealing than in education.

For decades, schools have often mistaken visible performance for inward development.

Correct answers stood in for understanding.

Completed assignments stood in for engagement.

Polished prose stood in for thought.

Artificial intelligence did not invent these substitutions.

It made them impossible to ignore.

When a machine can produce many of the artifacts schools have traditionally rewarded with ease, we are forced to ask a more fundamental question:

What were we trying to cultivate in the first place?

Leisure and the Conditions of Thought

Josef Pieper’s Leisure: The Basis of Culture offers one of the clearest answers.

Writing in 1948 from the wreckage of postwar Germany, Pieper argued that genuine culture depends on something modern societies increasingly distrust: leisure.

Leisure, in Pieper’s sense, is not idleness.

It is not laziness.

It is not the refusal to work.

It is the receptive, contemplative condition in which truth can be received rather than merely pursued.

It is a form of focused presence.

It is the inner stillness required for wisdom, art, education, and the formation of a person.

Pieper warned that modern life was increasingly organized around what he called the world of total work.

In such a world, everything must justify itself by measurable output.

Anything that does not produce immediately appears suspect.

The teacher who spends an hour wrestling with a difficult question seems unproductive.

The student staring quietly out the window appears disengaged.

The contemplative is told to become more efficient.

What is lost is not rest.

What is lost is the condition under which human beings are formed.

You can produce things without leisure.

You cannot cultivate souls or produce persons.

Pieper also distinguished between two modes of knowing.

Ratio is discursive reasoning: the step-by-step process of solving problems and reaching conclusions.

Intellectus is receptive understanding: the act of beholding reality and allowing it to disclose itself.

You use ratio to balance a budget.

You use intellectus when a difficult sentence from Augustine changes how you see the world.

Modern productivity culture excels at measuring ratio.

It has little patience for intellectus.

So it increasingly behaves as though intellectus does not exist.

This distinction is not unique to Pieper.

The Greeks called it paideia.

The Germans called it Bildung.

The Romans called it otium.

Christian monastic traditions called it contemplation.

The words differ.

The reality does not.

The AI hype economy did not invent the loss of this tradition.

It monetized it.

What is sold as frictionless productivity often substitutes the appearance of

understanding the difficult work of becoming wise.

That is why the most important questions about AI are not technical.

They are anthropological.

What is a human being?

What does it mean to know?

What kind of activity forms a person rather than merely generating output?

The answers determine whether AI becomes an instrument of human flourishing or

another mechanism for accelerating our distraction.

Substitution One: Attention Over Engagement

Teachers know this distinction intuitively.

There is a difference between getting students to look at something and getting students to be present with it.

The practice of contemporary education has increasingly favored the first because it is measurable, and the second is not.

Completed worksheets.

Quiz performance.

Time on task.

Attendance.

Grades.

We harvest the metrics and call it engagement.

Every experienced teacher knows better.

A student can complete every assignment and still not be present.

AI accelerates this confusion.

It is excellent at mechanical productivity.

It is largely blind to the interior reality of a learner.

It does not perceive hesitation.

It does not notice visible confusion.

It does not sense the moment when the eyes light up, and an idea takes form.

What the interface cannot tell you is whether anything in the student has actually moved.

Attention can be manipulated.

Engagement cannot.

Attention generates data.

Engagement forms a human being.

The hype merchants often do not know the difference.

They sell systems that maximize visible activity and call the regression progress.

Engagement, in the older sense, is what happens when a person is allowed to be slow

and present with something that does not yet make sense.

That process is difficult to measure.

It is also the heart of education.

Substitution Two: Emotion Over Reason

Reason is hard work.

Emotion is cheaper and more immediately captivating.

Provocation reaches everyone in the feed.

The argument reaches almost no one.

Anyone who spends time on LinkedIn has seen this.

The AI conversation has been conducted largely on the emotional register.

Excitement about transformation.

Fear of obsolescence.

Awe at capability.

Anxiety about regulation.

Each emotion is real.

Each has its place.

What is often missing is the sober work of reasoning.

What is this technology?

What can it do well?

What can it not do?

What are we trading away?

You can watch the same YouTuber swing from breathless admiration to dismissive mockery of the same tool in two weeks.

The emotion changes.

The thinking fails to deepen.

Teachers are not immune to this.

The teacher who feared ChatGPT and now uses it for convenience may have changed

his feelings without changing his understanding.

Reason is what allows a teacher to stand still in the middle of the noise.

To say:

Here is what this tool is.

Here is what it does well.

Here is what it cannot do.

Here is what I will and will not allow it to do in my classroom.

And here is why.

That internal dialogue requires thinking.

It cannot be produced by feelings alone.

The hype merchants depend on this substitution.

Their business model is the management of emotion.

Reason interrupts the sale.

Genuine thought is the willingness to sit with a question until you arrive at an answer

you can defend.

Substitution Three: Immediate Gain Over Steady Progress

Traditional educators understood that formation takes time.

They were patient by principle.

The student sat at the feet of the master for years.

Bildung means formation.

And formation takes time.

The AI hype economy leans in the opposite direction.

Ten times productivity by Friday.

Transformation in thirty days.

A strategic plan in twelve seconds.

A letter in moments.

The output may resemble the result of slower work.

What is missing is the part that mattered most.

The document was never the point.

The thinking the document was supposed to require was the point.

And that thinking requires a foundation of knowledge built over time.

The teacher who lets AI draft a difficult parent letter and relies on polish alone will be disappointed.

Judgment, insight, and genuine concern still have to come from her.

Output can look impressive while bypassing formation.

Over time, the bypass accumulates.

A person becomes capable of producing documents but incapable of judgment.

The documents may pass review.

The person fails when the situation no longer fits the template.

Real thinking still takes time.

You do not become a teacher in thirty days.

You do not become a strategic thinker in twelve seconds.

You do not become carefully formed by consuming only summaries.

You become these things by sitting with hard questions for years.

The hype merchants know this.

They are incentivized to emphasize speed in a culture that prizes efficiency above almost everything else.

But teachers are entrusted with formation.

And formation takes time.

Those incentives do not match.

Substitution Four: Fast Profit Over Productivity

There is a difference between making something and extracting value from something.

Much of the current AI economy is organized around the latter.

Models were trained on generations of writing, scholarship, art, code, and conversation.

The profits can be enormous.

But the productivity, in the deeper sense of creating enduring value, is real in some cases and superficial in many others.

This is the oldest temptation in commerce.

To pursue what pays quickly rather than what produces deeply.

To optimize for fast profit at the expense of genuine productivity.

Teachers intuitively know what real productivity looks like.

You cannot monetize the moment when a student begins to understand justice.

You cannot predict such moments mechanically, regardless of how much time a tool appears to save.

The profits are captured by those who build tools on top of cultural inheritance.

The productivity, in the deeper sense, remains in the slow work of forming minds and souls.

Pieper saw this clearly.

Profit is what the total-work culture seeks.

Productivity, in the older sense, is what leisure produces.

The two are not synonyms.

They are often opposites.

The teacher who spends a Sunday afternoon thinking carefully about one student is

profoundly productive.

The dashboard may not register it.

The attention merchant posting for the eleventh time that week may generate profit

while producing nothing of lasting value.

The first is leisure in Pieper’s exact sense.

The second is the condition he warned would follow when output became the

measure of all things.

The Part That Is Yours

So here is the line again, now that we have examined the substitutions it was designed to resist.

To think and not make thoughts your aim.

Kipling offered practical wisdom to his son.

Pieper would have recognized it as an example of the grammar of leisure.

Both point to the same truth.

Some forms of human activity cannot be outsourced.

They do not appear on productivity dashboards.

They cannot be optimized by large language models.

They cannot be purchased in thirty days or generated in twelve seconds.

They require years of attentive engagement with reality.

They require patience.

They require suffering.

They require presence.

This is not a condemnation of the tools.

It is a warning against confusing assistance with formation.

Use the tools in ways that free time and attention for the moments that matter.

Then teach your students to do the same.

Let artificial intelligence assist with what is mechanical.

But defend what is formative.

Protect the slow work by which judgment is born.

Because what is being offered as a substitute is not better.

It is only faster.

And in the formation of a human being, faster was never the goal.

Cicero described one of the highest duties of teaching as movere—to move.

AI can transfer information.

It can summarize.

It can imitate style.

But it cannot be changed by what it says.

And a teacher who has not been changed by what he teaches cannot deeply move a student.

The work that matters most still costs something.

That cost is precisely what gives it value.

To think and not make thoughts your aim.

That is the part that is yours.

The greatest danger of artificial intelligence in education is not that students will cheat.

The greater danger is that they will become intellectually weaker while appearing more competent.

For the first time in history, a student can submit semi-polished work while avoiding the necessary confusion, frustration, and revision that make genuine understanding and skill development possible.

The paper may appear sufficient, perhaps even impressive, to the untrained eye, but lack the important rigor necessary for learning.

That is not education.

It is the simulation of education.

When athletes go to the gym, they often work with a spotter.

A spotter stands nearby.

He encourages.

He watches.

He intervenes when necessary.

But he does not lift the weight.

The weight must still be lifted by the athlete.

Muscles grow through resistance.

Minds do too.

AI should function like a spotter.

It should support the work, not do the work.

Students should use AI as a thinking partner.

To strengthen their understanding.

To clarify difficult ideas.

To generate practice questions.

To challenge weak arguments.

To identify misconceptions.

To test whether they truly grasp a concept.

Used this way, AI becomes a remarkably effective tutor.

It provides immediate feedback.

Infinite patience.

And explanations tailored to the learner.

But students should not use AI to bypass the very struggle that produces growth.

They should not ask it to write essays they do not understand.

They should not use it to avoid reading.

They should not submit polished words that conceal an unformed mind.

If AI removes the struggle that produces understanding, it is being misused.

This distinction mirrors a broader principle.

Some kinds of friction are mechanical.

Proofreading.

Formatting.

Organization.

These tasks can be delegated.

Other kinds of friction are rhetorical and intellectual.

Interpreting.

Evaluating.

Judging.

Arguing.

These are the activities through which a mind is formed.

They cannot be outsourced.

A student who asks AI to explain a difficult passage is learning.

A student who asks AI to quiz him is learning.

A student who asks AI to challenge his reasoning is learning.

A student who asks AI to do the assignment is not learning.

The appearance may be similar.

The outcomes are not.

The purpose of education is not polished output.

It is a transformed judgment.

Not the ability to produce convincing prose.

But the ability to see clearly, think carefully, and choose wisely.

AI can accelerate that process when used with discipline.

It can also undermine it when used as a substitute for effort.

There is a paradox here.

The student who lets AI think may look smarter today.

The student who uses AI to think more deeply will become smarter for tomorrow.

Teachers should not forbid AI.

They should teach students how to use it honorably.

As a tutor.

As a critic.

As a sparring partner.

As a spotter.

A proofreader.

But not as a replacement for intellectual rigor.

Use AI to strengthen your thinking, not to replace it.

That is the principle.

Simple enough for a student to remember.

Strong enough to guide lifelong learning.

Thank you for reading this essay. If it moves you, let’s continue the conversation. Drop me an email or give me a call. timsmoon@gmail.com, (757) 746-2931

<script type=”application/ld+json”>

{

“@context”: “https://schema.org”,

“@type”: “Article”,

“headline”: “AI Is a Spotter, Not a Substitute”,

“description”: “A practical guide for students on using AI to strengthen thinking rather than replace intellectual effort.”,

“author”: {

“@type”: “Person”,

“name”: “Tim Moon”

},

“publisher”: {

“@type”: “Organization”,

“name”: “Critical AI Solutions”

},

“keywords”: [

“AI in Education”,

“Student Learning”,

“Artificial Intelligence”,

“Critical Thinking”,

“Academic Integrity”

]

}

</script>

There is a particular kind of tired I see in teachers’ faces these days.

It is not the familiar exhaustion of papers, meetings, and late nights.

It is something deeper.

It is the look of a person who has spent twenty-five years learning how to awaken young minds, only to be told that a machine is about to make that work obsolete.

The consultants arrive.

The slides appear.

The language is always the same.

Efficiency. Personalization. Scale.

And somewhere in the back of the room sits a veteran teacher wondering whether the work of a lifetime is being reduced to a software feature.

I want to say as plainly as I can:

Your vocation is not under threat.

But to see why, we have to remember what teaching actually is.

Most of the public conversation about AI in education begins with a false assumption.

It assumes that teaching is primarily the delivery of information.

If that were true, teachers would have been replaced long ago.

Textbooks would have done it.

Recorded lectures would have done it.

Search engines would have done it.

They did not.

Because teaching was never fundamentally about information transfer.

Information matters.

But information is only the content.

The real work lies elsewhere.

More than two thousand years ago, Cicero described the three duties of the great communicator.

Docere. Delectare. Movere.

To teach.

To delight.

To move.

This ancient triad offers the clearest framework I know for understanding artificial intelligence in education.

And it answers, with surprising precision, what AI can do and what it cannot.

Docere — To Teach

To teach is to explain.

To clarify.

To break a difficult idea into forms that a student can grasp.

AI is remarkably good at this.

It can rewrite explanations at multiple reading levels.

It can generate practice problems.

It can provide immediate feedback.

It can identify misconceptions.

It can patiently explain the same concept ten different ways.

Teachers should use this without apology.

Lesson plans.

Rubrics.

Differentiation.

Practice quizzes.

Parent emails.

Translation.

This is mechanical work, and AI handles it well.

Let it.

Delectare — To Delight

To delight does not mean to entertain.

It means to awaken wonder.

It is the moment a student’s eyes widen because an idea suddenly becomes real.

The analogy that lands.

The example that clarifies.

The question that causes a teenager to lean forward and think.

AI can help here, too.

A skilled teacher can use AI to discover fresh comparisons, better explanations, and unexpected angles.

Not because the machine possesses insight.

But it can surface possibilities that the teacher may not have considered.

The tool expands options.

The teacher supplies judgment.

Movere — To Move

And then there is the highest duty.

To move.

To change a student in some lasting way.

Not merely to improve a score.

Not merely to complete an assignment.

But to alter the trajectory of a human life.

Every experienced teacher knows this moment.

Years later, a former student says, “Something you said changed me.”

The teacher often has no memory of saying it.

The student never forgot.

That is the real work.

And that work cannot be automated.

A machine can explain.

A machine can suggest.

A machine can imitate.

But it cannot stand before students with conviction forged by experience.

It cannot embody character.

It cannot love truth.

It cannot look at a struggling student and communicate, through tone and presence alone:

“I believe you can do this.”

It cannot move a soul.

That remains the work of a teacher.

It always will.

What This Means for Teachers

Use AI aggressively for the parts of your work that drain time but do not require your deepest humanity.

Let it draft.

Let it organize.

Let it summarize.

Let it generate.

Let it handle the scaffolding.

Then invest your reclaimed energy where it matters most.

In the conversation after class.

In the well-timed question.

In the encouraging note.

In the moral seriousness that no algorithm can replicate.

The best teachers are not being replaced.

They are being liberated.

Freed from some of the mechanical burdens that have long consumed their time.

Freed to focus more fully on the part of teaching that matters most.

The part that has always mattered most.

Teach.

Delight.

Move.

AI can help with the first.

It can assist with the second.

The third is yours alone.

Thank you for reading this essay. If it moves you, let’s continue the conversation. Drop me an email or give me a call. timsmoon@gmail.com, (757) 746-2931

<script type=”application/ld+json”>

{

“@context”: “https://schema.org”,

“@type”: “Article”,

“headline”: “AI Will Not Replace Teachers. It Will Reveal What Teaching Always Was.”,

“description”: “A classical framework for understanding what artificial intelligence can and cannot do in education, using Cicero’s triad: to teach, to delight, and to move.”,

“author”: {

“@type”: “Person”,

“name”: “Tim Moon”

},

“publisher”: {

“@type”: “Organization”,

“name”: “Critical AI Solutions”

},

“keywords”: [

“AI in Education”,

“Artificial Intelligence”,

“Teaching”,

“Classical Education”,

“Cicero”,

“Docere Delectare Movere”,

“Teacher Vocation”,

“Educational Technology”,

“Teacher Formation”

],

“about”: [

{

“@type”: “Thing”,

“name”: “Artificial Intelligence”

},

{

“@type”: “Thing”,

“name”: “Education”

},

{

“@type”: “Thing”,

“name”: “Teaching”

}

]

}

</script>

I have been trying, for months, to explain to clients why AI is not the tool they think it is.

It is not merely a calculator, a spell-check, or a search engine, though it can perform all of those tasks.

It is more.

But the sense that it is more collapses on contact with the oft-repeated claim that AI is just a tool, with the emphasis on just.

Though I have often acquiesced to such an oversimplification, it always gnawed at me, because I knew it was not just a tool.

AI is not just a tool.
It is an instrument.

Then, while watching a YouTube reel last week, it came to me.

A little boy approached a group of street musicians performing in a big city and asked if he could play the piano. The musicians were surprised and a little suspicious, but agreed to let him try.

I was not sure if it was serious or not. The boy seemed far too young to play at their level.

But of course, he surprised everyone with his skill as a musician.

I watched the faces of the adult musicians shift from indulgence to astonishment, and somewhere in that shift, the analogy I had been hunting for showed up.

AI was not just a tool, any more than a piano is just a tool.

A piano sits in a room, and a child wanders in and bangs on the keys. The noise is fun, or obnoxious, depending on the observer.

At this point, the child may conclude, reasonably, that a piano is a toy that makes loud sounds when struck.

The child is not wrong, exactly.

The child is simply not trained to use the tool.

The piano did not change.
The relationship changed.

But when a child who is a trained pianist sits down to play, something much different occurs.

What happens next is not louder, or faster, or more efficient.

What happens is that the piano talks back.

It offers resistances that the untrained child could not feel. It rewards attention that the untrained child did not know how to give.

It has opinions about the player’s posture, about the weight of the hand, about the silence between the notes.

The pianist is not using the piano.

The pianist is in conversation with the instrument.

It is a conversation made possible by years of having been shaped by the tool’s demands.

The pianist is not using the piano.
The pianist is in conversation with the instrument.

The piano did not change between the untrained child and the child pianist.

The relationship changed.

And the relationship changed because one of the two parties had become a formed musician.

This is the thing I could not say about AI until I could say it about the piano.

To the uninitiated, AI makes noise.

Useful noise, sometimes.

It can draft an email, clean a paragraph, or summarize something one didn’t want to read.

The noise is fun, or obnoxious, depending on the observer.

The uninitiated user concludes, reasonably, that AI is a productivity tool that produces text when prompted.

They are not wrong, exactly.

They are simply uninitiated.

To the uninitiated, AI makes noise.
To the initiated, AI talks back.

To the initiated, AI talks back.

It offers resistances the casual user cannot feel. It rewards attention that the casual user does not know how to give.

It has defaults that must be argued with, cadences that must be interrupted, and hedges that must be refused.

The conversation is real, and like all real conversations, it shapes both parties, whether they consent to the shaping or not.

I am not making a romantic claim here.

The piano is not magical, and neither is AI.

I am making a categorical one.

Some tools sit still and wait to be used. Other tools require initiation before they will do what they are actually capable of doing.

The hammer belongs to the first category.

The piano belongs to the second.

AI, whatever else it is, belongs to the second.

Some tools sit still and wait to be used.
Other tools require initiation.

Which means the question most people are asking is the wrong question.

How much should I use AI?

That is the question you ask about a hammer.

The right question, the one AI forces on us, is the one every pianist already knows:

What is this instrument asking of me,
and am I willing to be formed by the asking?

Most people will answer no and go on banging the keys.

That is their right.

But they should not be permitted the comfort of believing the piano is a toy or just a tool.

<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "AI Is Not Just a Tool. It’s an Instrument.",
"description": "An essay arguing that AI is better understood as an instrument that requires practice, attention, and formation, not merely as a passive productivity tool.",
"author": {
"@type": "Person",
"name": "Tim Moon",
"url": "https://timmoon.substack.com"
},
"mainEntityOfPage": {
"@type": "WebPage",
"@id": "PASTE_SUBSTACK_URL_HERE"
},
"datePublished": "2026-05-05",
"keywords": [
"artificial intelligence",
"AI literacy",
"human-AI collaboration",
"AI as instrument",
"technology and formation"
]
}
</script>

I did not come to artificial intelligence as a technologist.

I came to it as a teacher who spent twenty-five years watching students learn to think, through imitation, through practice, through the slow accumulation of a hundred small interactions that gradually produced something that looked like wisdom.

That is the progymnasmata method.

It is as old as Aphthonius, who codified it in the fourth century, and older still in practice. The Romans used it. The Renaissance humanists refined it. It passed through the great classical schools of the nineteenth century and survived, barely, into the twentieth.

The method works like this:

You study a model.

You imitate it closely.

You vary it deliberately.

Eventually, through repetition and reflection, something original emerges from the imitation.

What surprised me, when I finally paid serious attention to AI, was how well it understood this.

Not because AI is wise.

It isn’t.

But because the progymnasmata was never really about wisdom. It was about the mechanics of internalization. You learn a form so thoroughly that it becomes available to you. The model disappears into your muscle memory, and what remains is capability.

AI operates on a recognizably similar logic.

It has internalized so many models that it can generate plausible imitations of almost anything. The difference is that AI never arrives at genuine understanding. The form never becomes wisdom.

It stays mechanics, all the way down.

That distinction is not a flaw to work around.
It is something important to teach.

One classroom exercise makes this especially clear.

Take a short passage of written text. Analyze it carefully for structure, style, context, rhythm, and logic. Then back-engineer that analysis into a prompt designed to produce a similar essay.

In other words, the student studies a model, imitates it closely, varies it deliberately, and then watches what happens when the machine tries to do the same.

That is not a departure from the ancient method.

It is a new application of it.

Engaging in this exercise forces a student to think critically not only about content, but about the quality and nature of the writing itself.

What is its structure?

What is its tone?

What is its logic?

What does the writer know how to do that the prompt cannot fully capture?

Then the student translates that analysis into a prompt and asks AI to produce something equivalent.

It won’t.

Not quite.

And that is where the teaching lives.

The reproduction of the essay is the fun part. Students can compare results. They can compete over which prompt produces the closest imitation. They can see, almost immediately, what the machine captured and what it missed.

But the end product is just the frosting.

The analysis is the cake.
The process is the cake.
The thinking is the cake.

That matters because AI is forcing education to recover something it should never have lost.

We have spent too long focusing primarily on the final product: the essay, the worksheet, the completed assignment, the polished output.

But education was never supposed to be mere knowledge reproduction.

At its best, education was always about formation.

It was about learning how to read, how to notice, how to imitate, how to vary, how to judge, and how to think.

I believe AI allows us to bring that back.

Students who do this exercise learn three things simultaneously.

They learn to read closely, because you cannot write a precise prompt without first doing the analytical work.

They learn to think systematically, because a prompt is a specification, and a specification requires logical decomposition.

And they learn what good writing actually involves.

Not merely correct sentences.

Not merely clean grammar.

Not merely a plausible structure.

Good writing carries judgment. It carries presence. It carries the pressure of a human mind making meaning.

That is what AI can imitate.

It is also what AI cannot become.

That is not a lesson about AI.
That is a lesson about being human.

I am currently bringing this exercise, and others like it, to schools across South Central Texas as part of a series of free workshops for teachers who want to use and build practical AI tools that work in a real classroom.

The initial workshop is free because I am still testing the methodology, and honest feedback is worth more to me right now than a fee.

If you teach, lead a school, or know someone who does, the door is open.

I can help teachers learn how to use AI practically in the classroom. I can also teach them how to build their own tools and apps using existing lesson plans, natural language, and the pedagogical judgment they already possess.

The method is older than you think.

It has just found a new application with AI.

Tim Moon is the founder of Critical AI Solutions and teaches AI at Wayland Baptist University in San Antonio. He writes Silicon and Soul on Substack.

Schema:

<script type=”application/ld+json”>

{

“@context”: “https://schema.org”,

“@type”: “Article”,

“headline”: “The Method Is Older Than You Think”,

“description”: “An essay connecting the ancient progymnasmata method to practical AI classroom exercises, arguing that reverse-engineering prompts can help students recover close reading, imitation, variation, and critical thinking.”,

“author”: {

“@type”: “Person”,

“name”: “Tim Moon”,

“jobTitle”: “Founder of Critical AI Solutions”,

“affiliation”: {

“@type”: “Organization”,

“name”: “Critical AI Solutions”

}

},

“publisher”: {

“@type”: “Organization”,

“name”: “Silicon and Soul”

},

“about”: [

{

“@type”: “Thing”,

“name”: “Artificial Intelligence in Education”

},

{

“@type”: “Thing”,

“name”: “Progymnasmata”

},

{

“@type”: “Thing”,

“name”: “Classical Education”

},

{

“@type”: “Thing”,

“name”: “Prompt Engineering”

},

{

“@type”: “Thing”,

“name”: “Critical Thinking”

}

],

“keywords”: [

“AI in education”,

“progymnasmata”,

“classical education”,

“prompt engineering”,

“critical thinking”,

“writing instruction”,

“AI classroom tools”,

“human-machine collaboration”

],

“mainEntityOfPage”: {

“@type”: “WebPage”,

“@id”: “PASTE_SUBSTACK_URL_HERE”

},

“image”: “PASTE_IMAGE_URL_HERE”

}

</script>

Scroll LinkedIn for ten minutes, and you’ll meet him. The Refuser. He writes his own emails. He does his own thinking. He wants you to know this. The post is usually a variation on the same beat: I asked ChatGPT to do X and look how bad it was. Imagine outsourcing your mind to this. The comments agree vigorously. Everyone feels better. Nothing has been learned.

The argument underlying the performance is old and tired: AI use replaces humans. It removes the struggle that makes writing writing, the effort that makes thinking thinking. The student who uses it is cheating themselves. The professional who uses it is hollowing out. The writer who uses it is no longer a writer.

I want to propose a different frame, because the replacement story doesn’t match what I actually see when I watch people work with these tools.

AI doesn’t replace the human in the loop. It reveals them.

Put two people in front of the same model with the same assignment. One types “write me 800 words on rhetorical friction in classical education” and ships whatever comes out. The other spends forty minutes in an argument with the machine, pushing back on a weak claim, asking for the counterargument, rejecting a tidy metaphor because it flattens something important, noticing that the third paragraph is doing work the second one should be doing. Same tool. Different outputs. The difference is not the software. The difference is the person.

This is what I mean by a mirror that amplifies. The model enhances whatever you bring to it. Thin thinking in, thin output out. And the thinness is now legible in a way it wasn’t before, because you can no longer hide behind the labor of typing. But if you enter rich thinking, something genuinely new emerges. The model simply follows the trail; your questions and revisions determine the brilliance, not the model.

The replacement critique assumes the human contribution is the labor. Type the words, fight the blank page, wrestle the sentence into shape. Remove the labor, and you’ve removed the human. But the labor was never the point. The labor was a proxy for the thinking, and a lossy one at that. Plenty of effortful writing is effortfully empty, and plenty of effortful writers never learned to think, only to perform the appearance of having thought.

What AI does, for the person willing to use it well, is strip the proxy away and leave the actual variable exposed: what do you know, what do you notice, what do you care about enough to push back on? These were always the questions. They were just easier to dodge when the labor could stand in for the substance.

Which means AI collaboration is not the end of assessment. It’s the beginning of a better one. The student who submits a Claude-generated essay on Hamlet and the student who submits something she built through forty exchanges with Claude about whether Hamlet’s delay is theological or psychological are not turning in the same kind of artifact. The first has no fingerprints. The second has fingerprints everywhere, because every turn of the conversation required her to bring something the model couldn’t supply: a hunch, a half-remembered line from Augustine, a suspicion that the standard reading is too neat.

The teacher who can’t tell the difference between those two artifacts has a pedagogy problem, not an AI problem. The teacher who can tell the difference has, for the first time in a long time, a direct window into the student’s actual mind, not her stamina, not her time management, not her facility with academic register, but her thinking.

That window is what the performance of refusal is closing off. Refusing to engage with AI in education doesn’t preserve the assessment of student thinking. It preserves the assessment of student labor, which we have always mistaken for thinking, because the two used to travel together. They don’t have to anymore. That’s not a loss. That’s a clarification.

There are serious arguments against AI in writing and education. Cognitive atrophy is real. Epistemic dependency is real. The deskilling of a generation is a live worry I take seriously. Those arguments deserve engagement.

But the LinkedIn performance isn’t one of those arguments. It’s a costume. And the thing about costumes is, they only work until somebody holds up a mirror.

"@context": "https://schema.org",

"@type": "BlogPosting",

"headline": "AI, A Mirror that Amplifies",

"description": "AI doesn't replace the human. It reveals them. Why the Refuser's critique of AI misses what it actually does to writing, thinking, and assessment.",

"author": {

"@type": "Person",

"name": "Tim Moon"

},

"datePublished": "2026-04-19",

"image": "REPLACE_WITH_HEADER_IMAGE_URL",

"mainEntityOfPage": "REPLACE_WITH_ESSAY_URL",

"keywords": "AI in education, writing pedagogy, AI assessment, human agency, AI collaboration"

}

A tool you made serves you differently than a tool someone sold you.

Teachers have always known this. AI just makes it possible on a new scale.

I set out recently to solve a problem that I couldn’t stop thinking about, and in the process, built an app.

I wanted to employ AI to Socratically to ask questions, not answer them. And I wanted to empower a teacher to measure a student's progress in writing over time, in a simple way that didn’t result in undue burden on their time.

I could not believe how simple this process was, using only normal English prompts.

The Tool You Made

Every teacher understands something intuitively that takes an economist a paragraph to explain: a tool you made serves you differently than a tool someone sold you.

The lesson plan you wrote from scratch carries your judgment about what matters and in what order. The rubric you built after three years of watching students misunderstand the same thing reflects knowledge no publisher has. The assignment sequence you designed around your students’ actual weaknesses is doing something a textbook simply cannot do, because the textbook was written for a student who doesn’t exist — the imaginary average student every EdTech product was designed for.

Teachers have always known this. It’s why the best ones spend their summers rebuilding their curriculum instead of buying someone else’s. It’s why the most effective classrooms tend to feel handmade.

AI doesn’t change that instinct; it scales it.

What I Built and Why It Matters

The app I eventually built is called VoiceArc. It’s a rhetorical voice-tracking tool for writing teachers. It analyzes student writing across three dimensions: syntactic maturity, stylistic identity, reasoning development, and rhetorical courage. It gives teachers a developmental picture of where a student’s voice actually is, not where the rubric assumes it should be.

It exists because I built it around convictions I’ve carried for two decades. My background in classical rhetoric, the progymnasmata, the trivium, and the belief that voice isn’t decoration but evidence of thought. No off-the-shelf tool encoded those convictions because no off-the-shelf tool knew I had them.

No off-the-shelf tool encoded those convictions because no off-the-shelf tool knew I had them.

Here is the part I want you to hear clearly: I am not a software developer. I did not know how to build an app when I started. What I knew was what I wanted it to do. That distinction matters more than any technical skill.

Because AI has collapsed the distance between knowing what you want and being able to build it. The barrier was never aptitude. It was permission. Nobody told teachers they could do this.

Consider this your permission.

Building Is Teaching

When you configure an AI tool for your classroom, you are not doing something technical. You are doing something pedagogical. You are encoding your judgment, your experience, your understanding of how students learn into a system that will work on your behalf when you aren’t in the room.

A Socratic tutor you built knows that your students tend to make claims without evidence. A grading assistant you configured knows that you care more about argument coherence than comma placement. A writing coach you designed reflects your actual convictions about what good writing is, not a generic rubric’s approximation of it.

This is not a technical act. It is the same act you perform every time you design an assignment.

You are instructing. The medium is new. The work is the same.

And there is one more thing. The teacher who builds her own Socratic tutor and then puts her students in front of it is modeling exactly what she wants from them: creative ownership of language, tools, and thought.

You cannot teach agency while using only tools built by someone else.

What’s Next

I’m offering two things to teachers who are ready to take this seriously.

Free access to VoiceArc

Use it with your students. See what a teacher-built tool feels like from the inside. Then ask yourself what you would build if you knew you could.

A free workshop

I’m going to teach teachers how to build their own AI tools, from the ground up, no technical background required. You will walk away with something real and usable. Not a prototype. A tool that carries your judgment and serves your students.

If either of those interests you, reach out. I will get you the details. The only prerequisite is that you still believe the tools you make matter more than the tools you buy.

You’ve been building things that work for years.
This can be your newest material.

You chose one of the hardest jobs in education. Not because the content is complicated, though it is. Because the work is fundamentally relational, and relationships don’t scale. You read what your students write, and you respond to the person behind the words. You hold the line on clarity and precision in an age that rewards neither. You assign the books nobody wants to assign anymore, because you still believe that stories form human beings in ways that nothing else can. You do this across five classes, thirty students each, with a grading load that would humble most professionals, for a salary that reflects none of it.

This letter is written with full respect for that weight. It is not written by someone standing outside the classroom telling you what you’re doing wrong. It’s written by someone who has spent years in classical education, who understands what it costs to care about a student’s voice, and who believes the English teacher is one of the most important people in a young person’s intellectual life.

I want to start where you actually live: not in the policy debates, not in the faculty meetings, not in the EdTech webinars promising to transform your classroom. I want to start in that specific, quiet moment. You know the one. When you’re staring at a stack of essays, and you feel it. Not frustration. Something heavier.

The feeling that somewhere in that stack is a student who almost had something real to say. Who was close, maybe one good conversation away from a breakthrough, and you didn’t get there. Not because you don’t care. Because you ran out of hours marking comma splices.

That gap between the teacher you want to be and the teacher the job allows you to be. That’s the wound this letter is about.

The Irony Nobody’s Saying Out Loud

Many English teachers, good ones, conscientious ones, are using AI right now. To build rubrics. To generate feedback templates. To draft the email to the parent they’ve been avoiding. To survive.

And many of those same teachers have banned AI for their students.

That’s a problem. Not a moral failing. A structural contradiction. And students notice it immediately.

When the teacher grades on AI-assisted rubrics but condemns student work that employs the same AI, the message received isn’t “AI is dangerous.” The message received is: the rules don’t apply to everyone. And once a student reaches that conclusion, you haven’t protected them from AI. You’ve just made the cheating covert, the relationship adversarial, and your own authority harder to sustain.

This isn’t a scolding. It’s an observation with a way out.

The Prescription

What if you stopped trying to ban AI and started redirecting it, for both you and your students?

Here’s what that looks like in practice:

① Students write more by hand, in class, without screens.

Not as punishment. Not as nostalgia. As protection. The hand-to-page connection is where voice forms, where thinking slows down enough to become sentences. In-class handwriting isn’t anti-technology. It’s pro-cognition. It’s the one space where a student’s raw, unassisted thought is still visible to you.

② AI becomes a tutor that asks questions, not one that answers them.

There’s a meaningful difference between a student asking AI to write their thesis and a student asking AI “what’s weak about this argument?” The first replaces thinking. The second demands it. A Socratic AI tutor, configured to push back, probe, and refuse to just hand over the answer, is pedagogically more rigorous than most study groups. It’s the Socratic method, scaled. You can design for this. It’s not complicated.

③ You use AI to absorb the mechanical load, openly, without apology.

Grammar pattern analysis. First-pass feedback on structure. Tracking which students are consistently struggling with argument coherence. These are real tasks that consume real hours. AI handles them well. Which means you get those hours back, not for rest (though that matters too), but for the work only a human being can do: reading a student’s voice, seeing the person behind the syntax, asking the one question that changes the trajectory of a young writer.

What This Is Really About

Voice — a student’s authentic, developing, irreplaceable voice — doesn’t form through shortcuts. It forms through struggle, through revision, through the friction of being asked what do you actually mean by that? The teacher’s job, at its most essential, is to be present for that struggle.

AI doesn’t threaten that job. Mechanical overload threatens that job. AI, rightly deployed, is what gives it back.

You got into teaching because you believed words matter. That a well-constructed sentence is evidence of a well-ordered mind, that literature does something to a person that nothing else can. None of that has changed.

The tools have changed. The stakes haven’t.

Use the machine for the mechanical. Save yourself for the human.

Your students are still waiting for that conversation.

— Tim Moon

Critical AI Solutions · Silicon and Soul

One of the stranger features of the present AI debate is the alliance between anti-sycophancy rhetoric and a very modern intolerance for friction.

Many critics are right to object when AI flatters the user, avoids disagreement, or confuses politeness with honesty. Those are real weaknesses. A system that only reassures is not a serious partner in thought. It is a service posture wearing the mask of intelligence.

But an irony appears almost immediately.

Some of the same people who complain that AI is too agreeable also recoil when a person, or a machine, speaks with unwelcome clarity. They say they want candor, rigor, and less emotional catering. But when candor threatens identity, or rigor becomes inconvenient, the complaint changes. Then honesty becomes harsh. Precision becomes pedantic. Directness becomes unsafe. Thoroughness becomes too much.

This is not only an AI problem. It is a human one.

Most people do not simply want the truth. They want truth under conditions that preserve a settled identity. They want to feel intelligent, fair-minded, and independent without paying the full cost of revision. They do not want cheap flattery, but many still want a subtler form of affirmation. They want their assumptions handled gently, their motives presumed noble, and their preferred moral vocabulary treated as beyond question.

People denounce flattery in principle, then demand protection in practice.

In that sense, the problem is not that people love sycophancy outright. They often want a more refined version of it. They want honesty, but not so much honesty that it disrupts the story they tell about themselves.

That is why the Marquis line still lands. There is a real difference between making people think they are thinking and making them think. The first experience flatters. The second unsettles. The first confirms identity. The second tests it.

To make people think they are thinking is to let them enjoy the sensation of intelligence without the labor of reordering anything. It allows them to feel perceptive without becoming vulnerable. To make them think, by contrast, is to introduce friction. It brings borrowed opinions, tribal reflexes, emotional investments, and hidden contradictions into view. That rarely feels like liberation at first. More often, it feels like irritation, exposure, or threat.

That pattern also helps explain why the conversation around AI is so unstable. Some critics genuinely want systems that are less flattering and more truthful. Others seem to want the appearance of rigor rather than rigor itself. They demand precision when attacking AI, then mock precision when it complicates their own claims. They denounce shallowness, then grow impatient with visible care. They say they want honesty, but only if honesty arrives in a way that still performs emotional housekeeping.

You cannot complain about shallowness and resent thoroughness at the same time.

You cannot condemn flattery and then punish clarity the moment it stops protecting your preferred self-understanding.

That does not mean bluntness is automatically virtuous. Cruelty is not honesty, and abrasiveness is not a proof of seriousness.

The opposite of sycophancy is not rudeness. It is honesty, and a willingness to tell the truth, with kindness, not demands.

That standard is harder than it sounds.

It asks more of both speaker and listener. The speaker must resist the temptation to flatter or manipulate. The listener must resist the temptation to confuse discomfort with injustice. Both must accept that truth often comes with bruises, and that friction is not always a sign that something has gone wrong. Sometimes it is a sign that something genuine is happening.

AI has not created this contradiction. It has exposed it.

Machines now make visible a tension that human beings have long tried to obscure: we praise honesty in the abstract, but many of us still want it managed. We want the benefits of candor without the pain of revision. We want to be challenged, but not too directly. We want rigor, but not when it turns its attention toward us.

We often praise honesty in the abstract, but still want it managed.

That is where the alliance between anti-sycophancy and thin skin becomes visible.

It is not really an alliance of principle. It is an alliance of convenience. One side wants the status of being associated with truth. The other desires subtle protection to control the emotional cost of the truth. Together, they produce a culture that speaks loudly against flattery while quietly insisting on more sophisticated forms of it.

The real test is not whether we say we want honesty. Almost everyone says that. The real test is whether we still want honesty when it costs us something, whether we can distinguish friction from injustice, and whether we can accept challenges without labeling them as harm.

That is what most professions of honesty eventually have to face.

That is a much harder standard.

And it is why so many people love to feel intelligent while resisting the imposition of thought.

There is a particular kind of satisfaction that arrives when an idea stops being theoretical.

It is working.

For more than a year, I have been arguing, in this space and in classrooms and in consulting work, that one of the most important things a educator can do with artificial intelligence is make it a thinking partner rather than an answer machine.

The AI’s job is to ask the next question, not supply the next answer.

That conviction has only deepened for me as AI has spread through education. The central problem is not friction itself. The problem is confusing the wrong friction with the right friction.

Mechanical friction

Students run into friction everywhere in the writing process. Some of it is meaningful. Some of it is less meaningful. Mechanical friction includes things like getting stuck on wording too early, losing the thread of a paragraph, freezing in front of an assignment, or burning energy on minor obstacles that do not actually deepen thought. It is real friction, but much of it does not form the writer. It just drains the writer.

Rhetorical friction

Rhetorical friction is different.

Rhetorical friction is the struggle that actually matters. It is the difficulty of deciding what you mean. It is the pressure of taking a position and holding it. It is the discomfort of discovering that your thought is still thin, your language still vague, your claim still too safe. It is the hard but necessary labor of trying to say something true and learning to own it.

That is the friction that forms voice.

My concern has never been that AI will simply make things easier. My concern is that it may remove the wrong friction so efficiently that students also escape the right friction. A machine can smooth the path so completely that the student never has to wrestle with thought, never has to risk style, and never has to develop rhetorical courage.

But that rhetorical struggle is where a writer is formed.

So my argument has never been that AI is the problem in education, my argument is that misuse of AI is the problem. AI should be trained to protect the right kind of difficulty.

I have written about this. I have taught it. I even built a simple prototype using Claude and Notion. It was enough to demonstrate the possibility.

Now it is something more than a possibility.

Now I have built a working, shippable model in Claude Code.

It matters to me more than I can say, not only because it works, but because I built it as a non-coder using Claude Code.

That too feels like part of the breakthrough.

The distance between educational vision and actual implementation is getting shorter. For people who have ideas about teaching but do not come from traditional software backgrounds, that matters. It means the gap between seeing a need and building something real to meet it is no longer quite so wide.

This is why it is critical that teachers lear these tools.

The tool is called the Student Voice and Style Progress Detector.

What excites me most is that it does something teachers actually need.

First, it helps make student voice visible over time.

The First Function

A teacher can paste in a student’s handwritten assignment in-class writing. Claude does not primarily read it for grammar or mechanics. It reads for something more human and more difficult to develop: rhetorical voice. It looks for distinctiveness. It helps the teacher see whether the writing sounds like a particular person with something at stake. It also helps track whether that voice is becoming clearer, stronger, and more consistent over time.

The teacher can add notes. Those submissions accumulate into a timeline. Over a semester or a year, the teacher can begin to see a student’s development in a way that is often difficult to capture when papers simply vanish into gradebooks and folders. There is a parent-friendly report. There is a shareable link. Some of the mechanical burden is lifted.

The teacher can think.

That matters because teachers also live under too much mechanical friction. They spend enormous amounts of energy on formatting, reporting, documenting, uploading, organizing, and other repetitive demands that crowd out the deeper intellectual work of teaching. If AI can relieve some of that burden without replacing judgment, that is a real gain.

Students also spend too much time writting long arduaous papers. Trust me, I know there is a place for that, but it is over emphasized. In my experience, shorter writing assignmets with quicker feed back, are far more effective at teaching writing skill.

The second function

The second part of the tool may matter even more.

Each student receives a login. They see their assignment, their prompt, and their tutor. The tutor is a Socratic AI agent trained on the teacher’s coaching method. It does not write for the student. It does not rush to rescue them. It does not offer a polished answer in place of thinking. It asks one question at a time. It presses gently but firmly. It keeps the student inside the work.

That is the point.

The student should not be trapped in needless mechanical friction. But the student should not be rescued from rhetorical friction either.

The right kind of struggle is where the growth happens.

The teacher can also read the full transcript of the student’s interaction with the tutor.

This is not surveillance. It is pedagogy.

The transcript makes the process visible. It lets the teacher see not only what the student eventually produced, but how the student moved toward that result. Where they got stuck. Where they wanted the machine to do the work for them. Where they tried to escape the hard part. Where the tutor held the line and returned them to the question.

For anyone who teaches writing, that matters deeply.

Because writing is not just a product. It is thought under pressure. If we lose sight of the pressure, we lose sight of the formation. We may still receive essays, but we will know less and less about whether a writer is actually emerging behind them.

Why this matters

What I love most about this build is that it is not locked into my method alone. A teacher can shape the tutor around their own pedagogy, their own classroom instincts, their own standards for what good writing requires. The architecture remains. The coaching can be theirs.

That matters to me too.

I do not want to build tools that replace teachers. I want to build tools that strengthen the best instincts of serious teachers and clear away the kind of friction that keeps them from doing their best work.

This is not finished. It still needs refinement, calibration, and revision. There are more features to build. There are rough edges to smooth. There is more testing to do.

But none of that diminishes what this moment is.

This is a breakthrough.

An idea I have been carrying for a long time has crossed into reality. I can now point to something concrete and say: here is a tool designed not to erase struggle, but to protect the right kind of struggle.

Voice is formed through friction.

And the future of writing education may depend on whether we learn to distinguish the friction that blocks thought from the friction that builds it.

That is what this build represents for me.

Not the end of the work.

The beginning of something lasting and real.

The first automation

One of the most fascinating perspectives that has emerged for me while thinking about automation and AI did not come from computer science. It came from paying attention to the human body.

Your heart beats without instruction. Your immune system constantly scans for threats. Your brain processes vision, balance, sound, language, and memory all at the same time. Your digestive system distributes energy where it is needed. Hormones regulate mood and internal states.

All of this happens without conscious supervision.

The more you reflect on it, the stranger it becomes. The body is not simply a collection of parts. It behaves more like a vast network of coordinated tools performing interlocking tasks. Each system contributes to the stability of the whole.

Breathing regulates oxygen.
Blood distributes nutrients.
Neurons transmit signals.
Cells repair damage.

Millions of processes operate continuously. None of them requires us to tell them what to do.

From this perspective, the human body already looks like an automated system. Not a mechanical one, but a living one. It is a web of organic workflows operating in remarkable coordination.

Creation already runs on systems

Once you begin to see this pattern, it becomes difficult to ignore it in the rest of the natural world.

Plants convert light into energy. Ecosystems balance populations. Weather systems regulate heat across the planet. Planetary bodies maintain stable orbits for millions of years. Everywhere you look, there are systems operating through layered processes that sustain stability over time.

Long before engineers built machines, reality itself was already full of coordinated processes.

Automation did not begin in a laboratory. It began in creation.

This realization has a way of changing how you read certain ancient texts.

The ancient claim

In the first chapter of the Epistle to the Colossians, we find a striking claim about the structure of reality:

“For by him were all things created, that are in heaven, and that are in earth, visible and invisible… and he is before all things, and by him all things consist.”

The phrase that stands out is simple but profound. By him all things consist.

The word consist does not simply mean exist. It implies that things hold together. They cohere. They are sustained in an ordered relationship with one another.

The claim is that reality is not just a pile of matter. It is a coordinated system held together by an underlying unity.

Biological intelligence

When you hold that idea next to what we observe in the natural world, it begins to make sense. Life is not chaotic. Biological systems display intricate coordination. Cells communicate with each other. Genes regulate production. Neural networks adapt through experience.

Modern biology increasingly describes life in terms of systems, feedback loops, and regulation. These are the kinds of patterns engineers study when they design automated processes.

In modern language, we might say that living systems consist of networks of organic tools operating in living workflows.

Mechanical intelligence

This perspective also reframes the rise of mechanical intelligence.

When engineers design software or artificial intelligence systems, they are not creating order from nothing. They are recognizing patterns that already exist in nature and implementing them in a new medium.

Neural networks were inspired by the structure of biological neurons. Pattern recognition algorithms imitate the way minds identify regularities. Automation systems replicate the delegation of tasks that already occurs in complex biological environments.

Mechanical intelligence is not an alien force entering reality from the outside. It is a new configuration of patterns that already exist within the fabric of creation.

The deeper continuity

There is a continuity here that often goes unnoticed.

The universe organizes matter through physical laws.
Life organizes matter through biological processes.
The human mind organizes knowledge through language and reasoning.
Technology now organizes information through computation.

Each layer builds upon the previous one. Each layer reflects a deeper order that makes the next layer possible.

Human beings did not invent intelligence. We discovered that intelligence was already embedded in the structure of reality. Once we began to understand those patterns, we started building tools that participate in them.

That is what computers do. That is what artificial intelligence does. These systems are extensions of human participation in the order of the world.

Seen this way, technology does not stand outside creation. It emerges from it.

The same coherence that allows atoms to bond and cells to communicate also allows engineers to design algorithms. The same rational structure that makes science possible also makes computation possible.

The deeper pattern remains the same. Systems hold together. Processes coordinate. Structures sustain themselves through layered interactions.

By him all things consist.

Returning to the mystery

Recognizing this does not reduce the mystery of the world. In many ways, it deepens it.

The more science advances, the more we discover astonishing levels of coordination in nature. Molecular machines operate within cells. Neural circuits process information with extraordinary efficiency. Cosmic structures form vast networks across the observable universe.

We can describe many of these processes in great detail. Yet the deeper coherence that allows them to exist together remains elusive.

Automation in machines is impressive.
Automation in living bodies is breathtaking.
Automation in the structure of the universe is something else entirely.

When we build intelligent systems, we are participating in a pattern that precedes us. We are not creating the order of the world. We are discovering it and working within it.

That realization invites both humility and gratitude.

The systems of the world hold together in ways that are intelligible enough for us to study, yet mysterious enough to inspire wonder.

The ancient text expresses that balance with remarkable simplicity.

By Him all things consist.