From the Koan Team

Reflections on Learning

Thoughts on education, AI, and the invisible process of becoming a thinker. Written for teachers, school leaders, and anyone who believes learning deserves better.

Latest

What They Are Watching

On June 11, Inside Higher Ed published the latest Student Voice survey of more than a thousand undergraduates on how they relate to AI. Six in ten said its real use to them in college is learning support. Four in ten said they are quietly worried about depending on it. One researcher named the finding precisely. The students, she said, are paying attention to what is happening to their thinking.

Inside Higher Ed's June Student Voice survey of 1,038 undergraduates landed without the drama the discourse keeps insisting on. Most students were neither evangelists nor refusers. They were using AI, often daily, while watching themselves do it. The most arresting line in the piece came from the researcher who interpreted the data. The dependence worry, she said, is not fear of the tool. It is students paying attention to what is happening to their thinking.

June 18, 20266 min read

Future of EducationAI in SchoolsLearning VisibilityStudent Voice

Read essay

The Line They Asked For

Oxford University Press surveyed nearly four thousand UK teenagers on AI in their schoolwork. The headlines went to the cheating finding, but the more useful number is buried lower. Only fifteen percent of the students said they have been given enough guidance by their schools on when AI should and should not be reached for. Eighty percent already use it daily. They are not refusing rules. They are asking for them.

Jun 176 min read

The Inner Life

On June 18, the 17th Ed Leadership International Roundtable opens at City Montessori School in Lucknow, the largest school on earth by the Guinness count and a recipient of the UNESCO Prize for Peace Education. The theme is "Growing the Inner Life in the Age of AI." The convenor, Dr. Sunita Gandhi, named the choice in the simplest possible terms. The purpose of school is not to make children faster than machines, but more human than ever.

Jun 166 min read

Where the Thinking Used to Show

A new NPR/Ipsos poll of 545 K-12 teachers, released June 5, 2026, found three things at once. Most teachers now use AI on the job. More than half believe it is making it harder for students to learn to think for themselves. Fifty-seven percent say it is hindering their ability to assess what their students know. The most useful question is not which finding is most worrying. It is what the three, taken together, are describing. A profession's standard instrument is going out of tune.

Jun 156 min read

Two Years to Look

On June 9, a coalition of New York City parents, teachers, and elected officials delivered a letter to the chancellor of the country's largest school system. They want a two-year moratorium on the use of generative AI in the city's 1.1 million-student system. The petition behind the letter has been gathering signatures since October. More than half the City Council has signed on in parallel. The careful question is not whether the pause will be granted. It is what the two years would be asked to do, and whether the field is prepared, when the interval ends, to have learned anything it could not have learned without one.

Jun 145 min read

Before They Ask the Teacher

This week, Common Sense Media released the first AI census it intends to repeat each year. The survey covered 1,204 American kids between nine and seventeen. Eighty-six percent had used a generative AI tool. The headline number traveled. The quieter finding underneath it, the one worth dwelling on, was that nearly a quarter said they would turn to a chatbot for help with homework or a personal problem before they would ask a trusted adult. The relationship a child has with the first person she asks is one of the oldest pieces of school. Something in it is being rerouted, gently, and the field has not yet decided what to do.

Jun 135 min read

The Pain Point

Last week, NBC News reported from inside Google's Mountain View campus, where the company had flown in seventy K-12 teachers and technology directors for a two-day training on Gemini and NotebookLM. The official subject was the tools. The unofficial subject was a persuasion script. The training's leaders taught attendees to overcome colleague resistance by finding pain points and matching them to a Google product. A first-grade phonics packet, the example went, could be generated in two minutes instead of two hours. The saving is real. The framing is the part worth pausing on.

Jun 125 min read

The Slower Clock

On June 1, Maryland's AI Ready Schools Act took effect, the most comprehensive state law yet written about artificial intelligence in K-12 classrooms. The architecture of the law is tidy. State guidance flows down to districts, who have 120 days to adopt policy and name a coordinator. By June 2027, AI literacy will be in the standards. The chain runs at one speed. A seventh grader deciding whether to ask ChatGPT for help on tonight's history paragraph runs at another. The gap between those two clocks is where most of the actual question is being decided.

Jun 115 min read

What the Room Believed

On Monday, the Hechinger Report described a survey done by a University of Chicago undergraduate of 338 of her classmates. Sixty percent said they personally use ChatGPT on their schoolwork. Ninety percent said they believed the average student on campus does. The thirty-point gap is not a measurement error. It is the same misread sociologists have catalogued on college campuses for thirty years, applied now to a new behavior. The imagined room is the room that drives action.

Jun 105 min read

The Rule, Not the Reading

On Monday, Common Sense Media released the first national Census of AI use by American tweens and teens. The headline number is eighty-six percent. The number worth dwelling on is two pages deeper. Three-quarters of kids say their school has told them what they can and cannot use AI for. Only about half say their school has taught them how to tell whether what an AI tool says is true. That gap is the story.

Jun 95 min read

Who Took the Time

On Tuesday, the Mountain Times republished a letter from a high school junior in central Vermont accusing her teachers of using AI to write the worksheets, draft the test items, and generate the lesson plans. The letter is roughly three hundred words long. It is the most precise account of the AI-in-education question to appear in any newspaper this week, and it was written by a sixteen-year-old.

Jun 85 min read

What the Index Has to See

On May 29, the dean of Boston University's Wheelock College and the executive director of Just Horizons Alliance signed a three-year agreement to build something the field has been quietly missing: a rigorous, independent way to evaluate the AI tools that have arrived in American classrooms before anyone could check them. The Index will measure nine dimensions. Eight of them have settled methods. The ninth is the one that will decide whether the project amounts to anything.

Jun 75 min read

What the Diploma Was Proxy For

On Thursday, Matt Barnum at Chalkbeat published a column with a question most school boards have not yet asked out loud. Will more education still pay off in an era when AI is closing the productivity gap between educated and less-educated workers? The harder question underneath is what the diploma used to signal, and whether the thing it signaled can still be seen.

Jun 65 min read

What the Trust Was Made Of

On Friday, NPR and Ipsos released a nationally representative poll of 545 K-12 teachers. The headline number was that nearly three in four believe AI will mean more for education than the internet or computers did. The number underneath was quieter and, in some ways, heavier. Nearly six in ten teachers said AI is eroding the trust between them and their students. Mallory Newall of Ipsos called it one of the biggest red flags in the data.

Jun 55 min read

Not the Minutes

On Tuesday, David Marshall of Auburn and Tim Pressley of Christopher Newport published an op-ed reporting that eighty-five percent of the nearly six hundred American teachers they surveyed in May believe students spend too much time on screens during class. The week before, the Boston Globe reported on the Cambridge Public Schools screen-time audit. The youngest students, the audit found, were on screens for about twenty minutes a day. The vast majority of usage was instructional. Two studies, one week apart, looking past each other at the same question.

Jun 45 min read

What Was Already in the Room

On Saturday, Euronews published a long look at the Estonian schools experiment most American superintendents have not yet read. The piece is called A technorealistic approach to AI literacy in Estonian schools and the word in the headline is the part worth dwelling on. Technorealist. Not optimist, not skeptic. A small country has decided that the binary the rest of us are arguing inside is the wrong one.

Jun 35 min read

He Asked Them How

Three years into the ChatGPT era, most districts are still working their way through one question. Should students be allowed to use AI on their writing, and how do we know if they did. A veteran English teacher in Central Bucks has been asking a different one. When his students turn in a paper, they fill out a transparency survey. Not whether they used the tool. How they used it. Which sentence they reshaped with help. Which paragraph they discarded. Whether the voice on the page still feels like theirs.

Jun 25 min read

The Plan Was for One Child

On May 20, NPR reported that fifty-seven percent of America's special education teachers are now using AI to help draft Individualized Education Programs, up from thirty-nine percent the year before. Fifteen percent rely on AI entirely. The Individuals with Disabilities Education Act requires that an IEP be individualized to one child. The tool was designed to recognize patterns across many. The two definitions of personalization, the legal one and the algorithmic one, do not match.

Jun 15 min read

What the Scale Was Hiding

California State University recently renewed its ChatGPT Edu contract with OpenAI at roughly $13 million a year for three more years, deploying the tool across 460,000 students and 63,000 faculty and staff. The system's own fall 2025 survey of 94,060 respondents shows the harder picture underneath the rollout. A majority of faculty say AI is hurting their teaching. Two-thirds of students say their professors have not shown them how to use it well. Scale was decided. Pedagogy was not.

May 305 min read

What Eighty-Two Percent Are Doing Alone

On May 27, Gallup and the Walton Family Foundation published the newest segment of their Teaching for Tomorrow study. Eighty-two percent of American public school teachers said they have received no formal guidance on how to apply AI to their work. Three in ten of those same teachers are using it weekly, saving roughly six hours a week. The map did not come with the tool.

May 295 min read

What She Did Not Ban

On May 27, Randi Weingarten unveiled a ten-point plan called Devices Down, Eyes Up, Hands-On. The press coverage emphasized the bans: screens out of pre-K through second grade, student-facing AI out of elementary school, companion chatbots out of the lives of anyone under sixteen. The careful part of the speech, the part that will matter five years from now, was the line she chose not to draw.

May 285 min read

The Quieter Number

A team led by researchers at Berkeley and Cornell surveyed 95,513 undergraduates across 20 public research universities and published the results in Science last Thursday. The number that traveled was nine percent: the share of AI users who have used the tool to cheat. The numbers that almost nobody printed were the quieter ones underneath, about who is using the tool at all, and about who is not.

May 275 min read

The Contract Was Signed. The Conversation Wasn't.

Cal State just renewed its OpenAI deal at $13 million a year, the largest agreement OpenAI has with any university. Underneath the headline is a faculty petition with more than 3,300 signatures, a student association that learned about the contract from a press release, and a systemwide survey in which 67 percent of students said their professors don't teach them how to use AI effectively. What does it mean when a tool arrives in a classroom faster than the conversation about why it is there?

May 265 min read

The Skills the Bubble Sheet Couldn't See

A piece in EdWeek MarketBrief this week walked through Skills for the Future, a pilot now running in classrooms across North Carolina, Indiana, Wisconsin, Nevada, and Rhode Island. Thousands of high schoolers are working through a new kind of assessment, one that uses AI to surface the durable skills employers and colleges have asked schools to develop for forty years and that no standardized test has ever quite been able to see.

May 255 min read

The Name Was the Point

This week in Utah, an eighteen-year-old senior at Northridge High School won a petition to keep an AI system called Tassel from reading her class's names at graduation. The same week, similar petitions in Virginia and Texas hit different walls. The story keeps being filed as an AI story. It is closer to a story about what a school means when it says, on the last day, that it sees a student.

May 245 min read

The Training Doubled

This week, Education Week reported that 58% of K-12 teachers have now received at least one AI professional development session, up from 29% two years ago. The doubling is real. The deeper question is whether teachers are being trained to operate a tool, or to read their students more clearly. The two paths look identical on a sign-in sheet and very different on a Wednesday afternoon.

May 235 min read

Closer to the Student

On Wednesday, NPR reported that a majority of special education teachers now use AI to help draft individualized education programs. The number jumped eighteen points in a single year. The reflex is to argue about the paperwork. The more useful question is what teachers do with the minutes the paperwork used to cost them, and whether the classroom is built to receive a teacher who has time again.

May 225 min read

What the Booing Was Saying

In the second week of May 2026, a new ritual quietly appeared at American college commencements. The boo. UCF on the 8th, Middle Tennessee State on the 9th, Arizona on the 16th. Each time, a different speaker, the same shape. The boo is not against the existence of AI. It is against the story being told about it from a stage that has stopped looking at student work.

May 215 min read

The Conversation, Not the Detector

Oakland Unified's draft AI policy, reported on Monday, finally puts language around a year of classroom-by-classroom improvisation. The interesting part is not what the policy allows or prohibits. It is a single instruction about what teachers should do when they suspect AI misuse. Not a scanner. A conversation. The instinct is right. The infrastructure for it does not yet exist in most classrooms.

May 205 min read

Education-Specific

On May 11, twenty-seven countries' education ministers met in Brussels and approved the Council of the EU's first formal conclusions on the role of teachers in the era of AI. The text is short by Brussels standards. Buried in the conventional language about literacy and equity is a phrase that does quiet work: the Council asks member states to promote education-specific AI tools. Not AI tools. Education-specific.

May 195 min read

The Sixth Practice

Michigan released a starter guide and a comprehensive framework for AI in its K-12 districts last Tuesday. The substance is a list of six essential practices. Five of them belong to the world of policy. The fourth, maintain human oversight, and the sixth, support transparency and continuous improvement, describe a relationship between a teacher and a student mediated by a tool no human can audit by hand. They require a thing most schools do not yet have. They require a record.

May 185 min read

The Witness in the Room

Princeton's faculty voted on May 11 to end 133 years of unproctored exams, citing AI as the breaking point. The Daily Princetonian's senior survey found that 28% of graduating seniors had used ChatGPT on assignments that prohibited it, more than double the year before. Forty-five percent knew of an Honor Code violation they had chosen not to report. The deeper story is not about AI cheating. It is about what the Honor Code was quietly leaving out.

May 175 min read

Sticky Help

Researchers from North Carolina State and Carnegie Mellon spent a school year logging 1,437,055 student interactions with an AI math tutor across 339 students in 14 classes. When they asked who the teachers chose to walk over and help, the data was gentle and unsurprising. Teachers tend to return to the students they have already helped. The researchers called it sticky help. Its measurable effect, the data suggests, is bounded to the same class session.

May 165 min read

From the Binder to the Building

On May 14, Teach Plus Illinois and the Illinois Digital Educators Alliance released a brief titled "From 'Rules and Tools' to Schools." Fifty-eight percent of responding educators now use AI for lesson planning. One in four reports no AI professional development at all. The state has until July 1 to publish official guidance under Senate Bill 1920. The quieter argument sits in the report's title.

May 155 min read

Between Relief and Displacement

On May 5, the University of Washington released a study interviewing 22 teachers in Aurora Public Schools, a district investing heavily in Google's Gemini and MagicSchool. The paper does not arrive at a verdict. It arrives at a name for the tension teachers keep encountering: relief or displacement.

May 75 min read

The Funding Pulls One Way. The Generation Pulls the Other.

On May 13, the U.S. Department of Education's new AI priority takes effect, weighting K-12 federal grants toward applications that push AI deeper into classrooms. Three weeks earlier, a survey of 1,572 Gen Z Americans found that anger about AI had jumped sharply and excitement had collapsed. Federal policy and student sentiment are running in opposite directions.

May 45 min read

When the Rollout Itself Goes Dark

A panel at last month's ASU+GSV Summit did something unusual for an industry conference. Three superintendents and district leaders sat on stage and said, in plain language, what they had gotten wrong about their AI rollouts. The mistakes were different. The shape of all three was the same.

May 35 min read

What the Moratorium Is Really Asking

On April 29, parents, students, and teachers filed into a Panel for Educational Policy meeting in lower Manhattan and stayed for seven hours. They asked the country's largest public school system to stop, for two years, before it puts AI any deeper into classrooms. The most surprising voice in the room belonged to the panel's resident technology expert.

May 25 min read

What Cleared Her Was the Process

Eleanor Canina got a 0 and a note that said 'evidence of AI, Please redo.' Three different detectors had returned likelihoods of 62, 75, and 87 percent. She had not used AI. What proved it was not a better detector. It was the writing process itself, captured in her Google Doc revision history.

Apr 305 min read

The Tutor They Liked, Until They Knew

Seven doctoral nursing students rated three blind responses to their statistics questions. They preferred the chatbot's answers. Then they were asked to guess which one came from the bot, and they reliably pointed at the response they had liked least. The paradox is not about better disguise. It is about visibility.

Apr 295 min read

The Studies We Have Not Run

Stanford's SCALE Initiative combed through more than 800 academic studies on AI in K-12, expanded to a repository of 1,100 papers, and found just twenty rigorous causal studies. None examine American schoolchildren in their own classrooms. The silence is not just a research problem. It is a structural one.

Apr 286 min read

When the Teacher Can Breathe Again

Education Week just published a study on what AI does to teachers, not students. The chain it uncovered runs in the opposite direction from the burnout story we keep telling ourselves.

Apr 266 min read

The Students We Don't See

Researchers at NC State and Carnegie Mellon analyzed 1.4 million interactions in AI-powered classrooms and discovered a sticky pattern: teachers help the same students repeatedly while quietly disengaged students go unnoticed. The findings expose a truth about classroom attention.

Apr 256 min read