What Project Zero taught me—and why it applies to your SPED & ELL classroom right now
I want to start with a question I’ve been sitting with for a while now.
If AI can generate the answer—and increasingly, it can generate a pretty good answer in seconds—then what exactly are we teaching when we teach?
I don’t ask that to be dramatic. I ask it because I think it’s the most honest professional question any of us can sit with right now. And I ask it because my experience with Project Zero’s Cultures of Thinking work at Harvard Graduate School of Education gave me a framework for answering it—one that feels, to me, more urgent in 2026 than it did the day I first encountered it.
The answer isn’t complicated. But it does require us to be honest about something we sometimes avoid in education: the difference between a student who has the right answer and a student who actually knows how to think. Those are not the same thing. And AI has made that distinction impossible to ignore.
What I Learned From Project Zero—And What It Changed
I’ll be upfront about something, because words matter and I believe in saying things accurately: my connection to Harvard has been through Project Zero, the research center at the Harvard Graduate School of Education, and its professional learning community. I’m not a Harvard alumna in the traditional degree-granting sense. But the experience of learning within that global community of educators shaped my practice in ways I’m still unpacking.
Cultures of Thinking, developed through the research of Dr. Ron Ritchhart and his colleagues, is built on a deceptively simple premise: a classroom culture doesn’t happen by accident. Every environment communicates something to students about what is valued, what is expected, and who gets to participate in intellectual life.
The work identifies eight cultural forces that shape learning environments: time, opportunities, routines, language, modeling, interactions, expectations, and environment. Read that list slowly. Because each one is a choice a teacher makes—consciously or not—every single day.
What we give students time to do tells them what we think matters. The language we use communicates whether ideas are welcomed or dismissed. Our expectations communicate what we believe students are capable of. And the environment—often without a single spoken word—tells students whether this is a place where their thinking belongs.
That last one hit me hardest. Because in a bilingual special education classroom—where students have historically been underestimated, misidentified, and sorted into environments designed for compliance rather than intellectual growth—the environment question isn’t philosophical. It’s an equity question.
The equity question: Are we designing learning environments that make thinking accessible—or environments that unintentionally make the thinking of our most vulnerable students invisible?
Thinking Routines: Not an Activity. A Culture.
When most people hear “thinking routines,” they picture a graphic organizer. Maybe a See-Think-Wonder posted on a bulletin board. Maybe a Claim-Support-Question template filed in a binder somewhere.
That’s not wrong—but it misses the point almost entirely.
A thinking routine only becomes part of the culture when it happens repeatedly. When students stop seeing it as an activity assigned by the teacher and start seeing it as a natural part of how the class makes sense of things together. That shift—from technique to culture—is everything. And it doesn’t happen the first time you use a routine, or the third. It happens when the question “What makes you say that?” stops being something the teacher asks and starts being something students ask each other.
The routine I return to most often, especially with my bilingual SPED students, is one of the simplest: I Used to Think… Now I Think. On paper, it’s five words. But think about what it actually asks a student to do.
- Revisit their prior thinking without shame
- Compare it against something new they’ve encountered
- Name what changed—and what caused the change
- Articulate the evolution of their own understanding in their own words
That’s metacognition. That’s self-monitoring. That’s the kind of executive function work that shows up explicitly in IEP goals for many of my students—and it’s happening inside a simple two-part sentence.
For my English learners, it’s also a language development tool. The sentence frame is predictable and safe—which means cognitive load drops and the student can actually focus on the thinking rather than the task of constructing an academic sentence from scratch.
This is the thing about thinking routines in SPED and ELL classrooms that I don’t see written about enough: the scaffold and the rigor aren’t in opposition. When you give a student a sentence frame, you’re not lowering the ceiling. You’re building a ramp to the door.
What This Looks Like on a Monday Morning
I promised you practical. So here it is.
These are four thinking routines I use regularly in my bilingual SPED classroom, adapted with intention for students with disabilities and English learners at varying proficiency levels. None of them require special materials. All of them can happen in under ten minutes—and all of them build something AI cannot replicate.
1. I Used to Think… Now I Think
Use it: At the end of a lesson, a unit, or after a discussion that shifted something. Works beautifully as an exit ticket for students with IEPs targeting written expression. Pair with sentence strips for students who need visual support.
Why it matters now: AI can summarize what was learned. It cannot reflect on how thinking changed. That’s a deeply human cognitive move.
2. What Makes You Say That?
Use it: During read-alouds, discussion, or when a student gives an answer—right or wrong. The question is the same either way. You’re not probing for errors; you’re building the habit of evidence-based reasoning.
For ELL students: pair with a sentence frame—“I think ___ because I noticed ___ in the text.” For students with language-based learning disabilities: allow verbal responses, recorded voice memos, or AAC device responses. The mode matters less than the thinking.
3. Claim–Support–Question
Use it: With any informational text. Students make a claim, find supporting evidence, and generate a genuine question the text didn’t answer. That last step—the question—is the one AI genuinely struggles with, because authentic curiosity requires caring about something.
In a mixed SPED/ELL classroom: differentiate the text complexity across versions, but keep the routine structure identical. Same scaffold, three entry points. Your below-grade reader and your enrichment reader are doing the same intellectual work through different doors.
4. Think–Puzzle–Explore
Use it: To open a new unit or introduce a new concept. Students share what they think they know, what puzzles them, and what they want to explore. For students with intellectual disabilities: use picture cues alongside the prompts. For newcomer ELs: accept home-language responses and bilingual pairs.
This routine does something AI genuinely cannot do for students: it positions them as the ones who are curious. The starting point isn’t AI’s summary of the topic. It’s the student’s own wondering.
🛒 CLASSROOM TOOLKIT | If you want the ready-made version of what I described here—differentiated thinking routine templates adapted for SPED and ELL learners, sentence frame banks by WIDA level, and IEP goal alignment guides for each routine—I built the complete classroom toolkit. Grab it on TpT here → teacherspayteachers.com/store/maria-angala-nbct
Why AI Makes This More Urgent—Not Less Necessary
Here is the part I want you to sit with.
AI can generate an essay. It can summarize a text, answer a comprehension question, produce a first draft, and in many cases produce a better-organized paragraph than many students can write independently. And this is going to keep getting more capable, not less.
So if we keep designing classrooms where the primary evidence of learning is a product—an essay, a worksheet, a multiple-choice test—we are building on sand. Because the product can be AI-generated, and increasingly, it is.
But here’s what AI cannot do—at least not in any way that serves a learner’s development:
- It cannot notice that a student’s thinking changed between September and March
- It cannot help a student recognize the moment they stopped being confused
- It cannot build a student’s confidence in their own judgment
- It cannot create the experience of being heard in a classroom discussion
- It cannot make a student curious—it can only respond to curiosity that already exists
Thinking routines do all of those things. Not because they’re magic. But because they create repeated, structured opportunities for students to externalize, examine, and revise their own thinking—over time, in community, with a teacher who knows them.
The more powerful AI becomes at producing outputs, the more important it becomes that our classrooms are places where thinking is the point—not just producing.
Dr. Ritchhart’s research puts it plainly: thinking is often invisible. A student can sit quietly for twenty minutes, and we have no idea what’s happening inside. Thinking routines make that internal process visible—audible, discussable, teachable. In a classroom where AI is available to every student on every device, making human thinking visible isn’t just good pedagogy. It’s how we know learning is happening at all.
The SPED & ELL Dimension Nobody Talks About
Here is something I’ve believed for twenty-three years and feel more strongly about now than ever: thinking is not a skill that belongs only to students who are already academically successful.
We have a long history in special education of giving the advanced student the open-ended question and the struggling student the worksheet. Of asking one student to explain their reasoning while another is simply expected to identify the correct answer. That hierarchy is embedded in how we’ve structured classrooms—and it has to end.
Every student is a thinker. The question has never been whether students with disabilities or English learners can think. The question is whether we have designed the conditions that allow them to show us their thinking.
A student may have an insight but not yet have the English to express it. A thinking routine with a sentence frame gives them the structure to get it out. A student with dyslexia may understand a concept deeply but struggle with written output. A verbal or visual response to a thinking routine captures the cognition the written page couldn’t.
A dual-identified student—English learner and a student with a disability—is navigating language acquisition and a learning profile simultaneously. Thinking routines, done well, reduce that double load by making the structure predictable. When the sentence frame is the same every Friday, the student doesn’t spend cognitive energy figuring out the format. They can spend it on the thinking.
That is access. That is what A.C.C.E.S.S. means to me as a framework—not lowering the ceiling, but making sure every student can reach it.
References & Further Reading
1. Project Zero — Cultures of Thinking Project (Harvard Graduate School of Education)
2. Ritchhart, R. (2015). Creating Cultures of Thinking — overview via Harvard Education Press
3. AI in the Classroom: What It Means for Critical Thinking — Edutopia (2025)
Frequently Asked Questions
Q1: Can thinking routines really work with students who have significant disabilities?
Yes—and I’d push back gently on the framing of the question. Thinking routines work precisely because they are flexible enough to meet students where they are. A non-verbal student can respond to “What Makes You Say That?” through AAC device output, picture selection, or a gesture system. A student with an intellectual disability can respond to “I Used to Think… Now I Think” using a two-picture visual prompt. The routine provides the cognitive structure; the mode of expression is yours to adapt. That’s the whole point of thinking routines—they scaffold the thinking, not the communication form.
Q2: What’s the connection between thinking routines and IEP goals?
More direct than most people realize. IEP goals in areas like metacognition, self-monitoring, executive function, academic language production, and evidence-based reasoning all map naturally onto specific thinking routines. “I Used to Think… Now I Think” targets self-monitoring and reflection. “Claim–Support–Question” targets evidence-based reasoning and academic language. “Think–Puzzle–Explore” targets initiating tasks and generating questions—both common IEP goal areas for students with executive function challenges. When you build thinking routines into your instructional routine, you’re addressing IEP goals inside grade-level content. That’s good special education practice.
Q3: If AI can answer questions faster and more accurately, why should students still practice thinking routines?
Because the goal of a thinking routine isn’t to arrive at the correct answer faster. It’s to develop the student’s capacity to evaluate, question, revise, and generate meaning—skills that are becoming more important as AI becomes more capable, not less. AI can produce an answer. It cannot tell a student whether that answer is trustworthy, what assumptions underlie it, or how their own understanding of the topic has changed. Those are human cognitive moves that require practice. Thinking routines are where that practice happens.
The Most Important Investment You Can Make Right Now
I’ve been in classrooms for twenty-three years. I’ve watched a lot of “next big things” arrive with fanfare and leave quietly. I’ve also watched genuinely important ideas get buried under the noise.
Cultures of Thinking is not new. Project Zero’s work has been building for decades. Thinking routines have been in research-informed classrooms for years. But I’m more convinced now than I’ve ever been that they are exactly what our SPED and ELL students need—precisely because AI is here, and it’s not going anywhere.
The question isn’t whether to use AI in our classrooms. That ship has sailed, friends. The question is what we protect, prioritize, and make irreplaceable alongside it.
I think the answer is this: the capacity of a student to examine their own thinking, change their mind with evidence, generate a genuine question, and understand why they know what they know—not just that they know it.
AI cannot give a student that. But a classroom built on a culture of thinking can.
Start with one routine. Pick the simplest one—I Used to Think… Now I Think. Use it every Friday for a month. Don’t grade it. Just make space for it. Watch what happens when students start to notice their own thinking shifting.
That’s the classroom we’re building. And that’s the one AI will never be able to replace.
You Might Also Like
📖 Designing Lessons for Real Kids, Not Pinterest: What Actually Works in My SPED Classroom
📖 From “Cheating” to “Solving”: The New AI Mindset
📖 How I Teach Vocabulary to ELL Students with Structured Supports
📖 Why the A.C.C.E.S.S. Literacy Framework Is Built Like a Bridge—And Why That’s the Whole Point
📖 10 Executive Functioning AI Prompts for SPED Teachers, OTs and SLPs
📬 Join Educators Who Think Deeply About Teaching If this post made you think—even a little differently—there's more where that came from. Join thousands of SPED and ELL educators who receive classroom-tested strategies, framework updates, and honest reflections on teaching in the age of AI. No noise. No filler. Just the kind of professional thinking that actually makes Monday better. → Sign up at: www.bilingualsped.com
Before You Go — A Reflection Question
💬 For You: Think about one student in your classroom right now—a student whose thinking you suspect is richer than their written or verbal output suggests. What cultural force in your classroom might be making their thinking invisible? And what is one thinking routine you could introduce this week that might change that? Drop your answer in the comments. I read every one. And if you're already using a thinking routine with your SPED or ELL students—tell me what you've noticed. Those stories are what this community is built on.