Make It Mindful: An Education Podcast

# Make It Mindful Transcript
# Guest: Rebecca Winthrop
# All timestamps marked *(approx)*

---

[INTRO — recorded separately]
[00:00:00]

Support for Make It Mindful is brought to you by World Savvy, partnering with K-12 schools, districts, and community leaders to prepare youth to thrive in a complex, interconnected world.

We are proud to feature Svitlo School - a free, extracurricular online school in English, that has educated more than 3,000 Ukrainian children since 2022. Children learn from native English speakers in the UK, US and across the world. Visit SvitloSchool.com to learn more about volunteering.

Welcome to Make It Mindful, the podcast about how students learn, who they're becoming, and how new technologies and global experiences reshape teaching.

I'm Seth Fleischauer, founder of Banyan Global Learning.

My guest today is Rebecca Winthrop, senior fellow at the Brookings Institution and co-director of its Center for Universal Education. She co-authored The Disengaged Teen, which builds a student engagement framework from a decade of research across thousands of students, parents, and educators — and she recently led a global task force examining the risks and benefits of generative AI in student learning across multiple countries.

This episode of Make It Mindful is brought to you by Banyan Global Learning.

At Banyan Global Learning, we design structured live experiences that connect classrooms to the world.

Banyan Global Cohorts is a 4-week journey in global competence. Classrooms join an international cohort of 3–4 schools and meet weekly during regular class time for facilitated, 45-minute sessions built around a compelling Big Question.

Students explore global contexts through virtual field trips, exchange ideas with international peers, and create a final digital artifact that expresses identity, perspective, and connection across cultures.

If you're curious about what this could look like in your classroom or school, we'd love to talk. You can find us at banyangloballearning.com.

And now here's my conversation with Rebecca Winthrop.

---

Seth (00:02:10) *(approx)*
Doctor Winthrop, thank you so much for being here. Welcome.

Rebecca Winthrop (00:02:13) *(approx)*
Thank you for having me, Seth.

Seth (00:02:15) *(approx)*
There's so much ground to cover here. Since I learned of you, I heard you on a podcast, and then was delighted to see that you were the keynote speaker at the IB conference that I went to in Denver in July and learned about your book, The Disengaged Teen — which was going to be called My Day Was Fine. I love that title, but I get why the publishers went there. And you also have this study that you did through Brookings on AI and education, exhaustive across many different countries, and there's a ton of insights and wisdom in there. Let's see if we can pull it all together. Here we go.

Seth (00:02:50) *(approx)*
I want to repeat back to you this thing that I've been saying to everyone since I learned about your framework. One of the things that strikes me about it is how intuitive it feels to people. They hear about it and they're like, yeah, that makes complete sense. So it is an engagement framework. You've got on the bottom axis, engagement, on the vertical axis, agency. You've got these four positions. The passengers are people who are low agency, low engagement — they'll kind of do whatever you tell them to do, but they're not stoked about it, and they're not going to fight it. The achievers are the ones who are high engagement but low agency, which is one of the surprising things about the framework. We think about achievers — these kids who do all the things right, do exactly what their teachers tell them to do — as the apex of learning, yet your framework identifies that they lack agency. Their biggest pet peeve is their teacher not telling them exactly what to do, because then they're not going to know what to do. They just want to know what the path is to the A, and they'll get it. But the resistors — those are the ones that are high agency, low engagement, and they aren't stoked, but they'll tell you about it. The explorers — that's the apex — where it's high engagement, high agency, and those kids are leading and creating their own learning. Another surprising finding is that resistors can move to explorers more easily than achievers can move to explorers. Do I have that all correct?

Rebecca Winthrop (00:04:05) *(approx)*
Yes, it is absolutely about right. It is very close. The only sort of voiceover I would do is the experience of kids in passenger mode is coasting. And they're oftentimes just bored, honestly. And a lot of them are getting good grades. So they may present as achievers, but they're not that engaged emotionally. They don't find it interesting cognitively. And then the explorers — the only other overlay there is this is where they're taking initiative to make the learning environment more interesting and supportive to themselves. And that can be in small ways — could I write my paper on this topic? Or ooh, could I study with a friend? Or ooh, I think I need to look at a YouTube video. This is small ways. So it isn't that explorers are just off the charts, climbing a mountain, doing their own thing entirely. It's that they're taking initiative to mold the learning environment to suit them better. And the contrast with kids in achiever mode is they are engaged at getting good grades. So they're not thinking about their interests per se. They're really good at following instructions and trying to figure out what the teacher wants. Because they're indexed on the output, not the learning journey. Explorers are interested in the learning journey. Achievers are indexed on the output. It's not a bad skill to understand what people in your environment want from you. It just ultimately makes some really good followers, which is not exactly what we need in an AI world.

Seth (00:05:20) *(approx)*
And you mention AI there. It is so alluring to try to latch on to these labels — to say I was an achiever, and I was for the most part. And my daughter's like that. And my son is such a resistor, but I see how he goes into explorer mode so easily, and I see the difference between that and how my daughter can struggle sometimes to get there. But these are modes, not identities. You talk in your book about resisting that temptation and identifying the ways in which we can flow between these modes. And the key to that feels like it's metacognition — we need to be able to see and understand how we are thinking in order to even understand which mode we're in, let alone how to transition from one mode to another, whether that be taking a break from achiever mode to be in passenger mode for a bit, or getting from resistor to explorer. Would you agree that that's the linchpin here?

Rebecca Winthrop (00:06:15) *(approx)*
I really love where you're going and it's where we point toward at the end of the book. What if kids knew about these modes? Because absolutely correct — they are states, not traits. And at the moment, young people sort of show up in the modes based on their external conditions and environments. Do they have a teacher who super believes in them? Do they have parents who are saying, I really need you to get an A because you need to be the one to go to college and get a good job? So they're like, okay, I'm going to be an achiever, do everything right. Do they take a class that is super hands-on and makes them super excited and internally motivated? These are all external conditions. And most of the time, young people are not aware of the modes they are in. They're sort of just being shuffled from one to the next based on how they feel, their internal motivation, whether they're excited, whether they have the persistence skills, whether they have the supports. Toward the end of the book we do point to this idea: what would it be like if educators started talking to students about the different modes of engagement? There's a little bit of evidence of teachers trying to give kids exactly those metacognitive skills to think about how they are engaging, and that it is very helpful. We've found a number of schools and teachers who are taking up the modes and doing things with them, including incorporating them into lesson plans, usually at the high school level, so young people can self-reflect on them and be more intentional about what they need to move from mode to mode. And sometimes it's okay if they need to be in passenger mode. I remember visiting a classroom — Mrs. Carmen in Texas, who had been teaching and scaffolding her students around engagement — and one student raised his hand and said, I'm really, really distracted today. He had something going on. He said, I'm just going to be passively participating. And it's like, okay, we get it. He's self-aware.

Seth (00:08:10) *(approx)*
Another thing you talk about in the book is this idea of transcendent thinking and this opportunity to reflect deeply on meaning and your role in the world. Is that what you see coming out of these schools that are embracing this metacognitive practice around the modes? Are they able to get students to that point of transcendent thinking?

Rebecca Winthrop (00:08:30) *(approx)*
I think some of them are, but just because you give students the tools doesn't necessarily mean the teacher's pedagogy will allow for that deep reflection — which is a criteria for kids to be able to get into this transcendent thinking mode where they're thinking not just about what is the right answer for the worksheet, but what can I do with this math? Or is this principle used in the real world? Or some sort of ethical dilemma that they're pondering. It's not a one-for-one. You still need educators to be able to set up some of those learning experiences.

Seth (00:09:10) *(approx)*
And it feels like a central part of that practice is understanding what good questions are. You talk about this from a parenting perspective in the book — how was your day? is obviously one of the flattest questions you can ask. But really practicing that curiosity: which part of your day was your favorite, what did you do well there? These kind of deeper questions. And I was struck — you had this quote, I guess it was a sign you saw at a conference in Nashville: be the grown-up you needed. As I was reading this, I was thinking, my gosh, all these are great ideas. I don't always do them though.

Rebecca Winthrop (00:09:45) *(approx)*
Yes. Yeah, one thing is knowing, the other is doing, for sure.

Seth (00:09:50) *(approx)*
Yeah. And even in my career — there's someone I work with, Dr. Trevor Saponis, friend of the podcast. I watch him coach and I'm like, that is exactly how you're supposed to do that. The way he just leans into that person's experience and resists giving any kind of advice until they've exhausted all of what's going on with the person. And I remember seeing that and being like, that's how you do it. And then I went and tried to do it and within like two minutes I'm giving advice.

Rebecca Winthrop (00:10:20) *(approx)*
The advice monster — we talk about that in the book. Yeah, it's from a great workplace book.

Seth (00:10:25) *(approx)*
Yeah. And I wonder — how do you become that better adult? How do you become that grown-up that you needed for the young people you work with, whether you're a parent or a teacher?

Rebecca Winthrop (00:10:35) *(approx)*
My goodness, what a question. A lot of the advice we give in the book is: if you want your kids to spend more time in explorer mode — not the whole time, more time — and have more explorer moments, there's things you can do no matter what mode your kids are in. As a parent and a teacher, but there's slightly different things you need to do if your kid is coasting in passenger mode, or driving really hard in achiever mode and melting down if they get an A minus, or in resistor mode — I'm out, not engaging. These kids need slightly different things. And across all of them, I think where parents and teachers go wrong is the source is anxiety and care and love for the kids in front of us — wanting them to get to a place that you want them to get to soon. Not waiting for a long period of time because you're worried about them. And I think the first step is honestly taking a breath, saying: this child in front of me is good, wants to be good, and is going to be okay. And then that gives you a little more space and time. If they don't want to talk about their day that day, try tomorrow. We do often say: start with things you know your kid loves. It could be lunch, it could be recess, but there's usually a class or something they're interested in or good at — and just start there. And maybe that's enough for that day. Because tomorrow you can get to the class where you know they're struggling. If you always start with the worst class, they're failing or they can't do it or it gives them anxiety — how was the math test? And did you need — and shall I tutor you? — it becomes a lot for young people and a lot of young kids can shut down from that. Talk about the content of the day, start with the things you know they are shining and loving. You don't have to talk about everything in their day that day. You can spread it out over a week. I don't like to talk about all the presentations I've bombed in great detail every night when I come home and have dinner with my family. Neither do they.

Seth (00:12:40) *(approx)*
Yeah. It's the presence of mind to resist seeing your kid walking toward a rake and being like, don't step on that rake — and understanding that if they knock themselves in the head with the rake, that is teaching them to fish. That patience, and navigating the discomfort of watching people you care about be in pain or suffering.

Rebecca Winthrop (00:13:05) *(approx)*
Yes. I think we have — in school, because schooling has become so high-stakes for social mobility, largely employment — we know that kids with a four-year college degree usually have a much better chance of higher lifetime earnings than kids without. So because it's sort of this really crucial path that they need to take, it lends itself to not allowing for struggle. Whereas in other domains, we feel comfortable with struggle. My oldest son who just graduated was on the track team. I was not like, oh my gosh, he had a hard workout, he's turning red. I was like, yeah, that's what you should do — you're training to run fast. Maybe channeling that feeling: learning to think well and learning to be in explore mode and to have deeper learning and higher order skills — using your academic knowledge to solve problems with others and collaborate — it's like riding a bike. They have to practice and we cannot do it for them. They will never learn if we're going to hold the back of the seat the entire time.

Seth (00:14:15) *(approx)*
You talk about education as the path toward social mobility. It reminded me of a question I had around explorer mode being labeled the apex of learning. You talk about kids in explorer mode achieving better results in life. And I was wondering: by what criteria are you determining them to be better?

Rebecca Winthrop (00:14:40) *(approx)*
So this really came out of the work of John Marshall Reeve, who works on agentic engagement along with many other colleagues. His fundamental insight was that there are various ways to measure student engagement — emotional engagement, cognitive engagement, behavioral engagement. And he said, you could be engaged on all three and really just be following exactly what the teacher is telling you. Not really learning to take initiative and try to shape the learning environment toward what interests or supports you. So he started looking at kids who do that initiative taking. He's been studying this for 20 years. What he found is less about life outcomes and more about success while they're in school. Young people who have teachers who let them take initiative — you do need a certain little bit of space in the classroom to do that — perform better on achievement over time compared to those who don't take initiative. They feel better. They have more pro-social behavior. So they out-achieve the achievers grade-wise. And they feel better. And they are developing a disposition for proactivity that I really think is essential given we are in an AI world where traditional standards of achievement can be done by a chatbot. All the high-stakes tests are easily completed by AI. So that isn't the right measure for whether we're truly preparing our kids. We want them to learn the knowledge — I'm not saying not — but that's not the end all be all.

Seth (00:16:25) *(approx)*
Yeah, and I guess that's where students in explorer mode who are out-achieving the achievers are doing so based on assessments that may or may not reflect the new world. Yet there's also a certain degree of reading the tea leaves and checking your intuition: we are going to need to be able to drive our understanding and knowledge in a world that's awash with information.

Rebecca Winthrop (00:16:45) *(approx)*
Mm-hmm. Mm-hmm.

Seth (00:16:50) *(approx)*
I want to pivot and talk about this AI study. I think the link between these two bodies of work is the idea that this engagement crisis — this disengagement crisis — is not new. It's been around for a long time, yet it is being intensified by technology. How is that?

Rebecca Winthrop (00:17:10) *(approx)*
Yeah. When we did this big global look at Brookings, we were looking at generative AI — not predictive AI, the new AI, post-ChatGPT. What are the potential risks, what are the potential benefits? And one of the things we did find is a lack of motivation. Because a lot of schools operate the way they have for many years — traditional assessments are snapshots in time, indexed on proxy products: a finished essay, taking an online exam. Given that generative AI can do all those really well — better in many ways than students feel they can do themselves — they're like, well, why am I here? What is the point of all this? And I am really worried about a lot of young people moving into passenger mode because of that. Not all. You've got some kids in explore mode who are like, I've always wanted to learn about this topic — and they're not using it to do the work for them or question the purpose of school, but to go deep on a topic. So I'm worried about AI creating a lot more kids in passenger mode. The problem is not really the kids. It's really this rapid new technology that has shown up and we haven't had time to shift the design of schooling to adapt to it.

Seth (00:18:30) *(approx)*
Gosh, what a time for education — on the heels of COVID to have AI arrive in the way it did. You talk about proxy products. And it feels like there is thinking that we want students to do. And to a certain extent, no matter what the product is, it's a proxy product of that thinking — until we get implants and we know exactly what anyone's thinking, you still need to get it down on a piece of paper, say it. There's always some kind of filter separating whatever they're producing from the actual goal of any given assignment, which is the thinking behind it. We get closer and closer to being able to understand that thinking the less standardized the assignment is — multiple choice being one extreme where the student could have guessed that answer or could have completely understood it. When we have students describe their thinking, practice metacognition, do oral defense of any given presentation or idea — these are ways to, I think you'd call them, durable assignments that are less...

Rebecca Winthrop (00:19:40) *(approx)*
I like that — durable assignments for durable skills, huh?

Seth (00:19:45) *(approx)*
Yeah! And what strikes me is how difficult to assess and standardize they are, how not very scalable. And when you talk about policy, which is your realm, that's all about scale. That's all about understanding what can we say at this level that's going to get translated down to the experience of the student. Do you see that tension?

Rebecca Winthrop (00:20:05) *(approx)*
I agree with you that part of the reason the proxy assessment products have endured so long is because it is easier to administer to large numbers in a more level-the-playing-field way. Nothing is ever really level, of course — even before AI, plenty of people used their older sibling's assignment. But I agree that's part of why they're so often used. But I'm not sure it's not scalable to do things like reflective journaling of the learning process, oral exams, proctored written exams — either with a frozen laptop or by hand in class. That does give an incentive to actually learn the material if you have to reproduce it with no aids. I think projects can work particularly if educators have a good way of sussing out the different roles young people are playing. I would say where it's much harder is higher education. But K-12, there's plenty of proof points that you can do that type of teaching and learning across an entire school or school system. If you can do it in one school, you can do it in others.

Seth (00:21:20) *(approx)*
And your report also talks about these narrow uses of AI that can knock down barriers to engagement with any given content — examples being language barriers, or especially for students with disabilities, voice to text. These technologies that maybe have been around for a while but are getting better. And I'm realizing I've now said the word metacognition four times, but I think about — they call it a spectrum for a reason. Everyone's somewhere on the spectrum, everyone's got some kind of special skill or thing they're working to develop. You don't want AI to do the thinking for you. You talk about cognitive stunting. A lot of people have talked about cognitive bypass, but stunting is different — because I can bypass something I already know how to do. If I'm having it do something I don't know how to do, I may never develop that skill.

Rebecca Winthrop (00:22:15) *(approx)*
Exactly.

Seth (00:22:17) *(approx)*
Is it true then that any given learner's relationship with AI is like a fingerprint — where if I've already developed the skill, maybe I could use AI to knock down this roadblock so that I can engage with this part of the assignment, increase my efficiency or capacity to do so? And so is it less about identifying from a policy level specific narrow uses of AI that we know are good for the general population, and more about teaching kids to understand when AI is doing thinking for them and avoiding those situations?

Rebecca Winthrop (00:22:55) *(approx)*
Yeah, right. So those two paths you talked about — one path is at a policy level, help young people have access to useful AI use in teaching and learning. And broadly, the headline takeaway is: we found that yes, generative AI can be helpful in student learning and development if used very, very narrowly — which is really teach-not-tell machines. It is really unhelpful if we call it wide AI use, where kids are directly interfacing with general-purpose chatbots — which are really tell-not-teach machines. That's the headline. And the question you're asking is: okay, from a policy angle, do you just focus on narrow AI use, or what about helping young people learn when they should use it, when they shouldn't? I think you need to do both for sure. However, I am not in support of just focusing on responsible AI use and putting the burden on young kids who are still developing — and frankly their teachers and parents who are busy. And it's hard to tell at the moment where exactly that line is. It's actually a topic I'm beginning to work on now: where's the line between what you should do with AI and what you shouldn't, and is it different for different kids based on their competency? That is incredibly complicated. It's a huge burden on a young person to say, that's your responsibility — good luck. So I think that should be a complement. And it really does come through critical AI literacy, where young people really learn about AI, they learn how it's made, they learn that it can be changed and designed differently — it's human-made, not the laws of physics. But we absolutely need to make narrow AI use more the default than wide AI use. And that is a policy choice.

Seth (00:24:50) *(approx)*
Within that narrow AI use, do you see a future where research supports certain narrow uses of AI in the context of particular assignments and particular students — maybe even if they have IEP labels? Do you see it getting that granular, where there's someone backed by research or by policy who is making the call about which narrow uses of AI are approved in a given scenario and that gets disseminated throughout the system? Or do you see there being individual choice at the teacher, school, or district level?

Rebecca Winthrop (00:25:30) *(approx)*
In the United States, we have a very decentralized system. So I could envision large-scale rules of the road coming from states or possibly the federal government — particularly around wide AI use that's not meant for learning and not safe for kids. We should try to shut that down, and that'll solve a lot of things. That's not something an individual school can do, because a lot of times kids are accessing it outside of school. So to me, that is an example of big guardrails at scale. But I cannot imagine in the US that every single kid from Tennessee to Texas to Oregon is going to have the exact same line in every English class. I am sure there will be variety in experimentation, which I'm not against. What I hope is that the range of variety fits within the bounds of evidence — so we aren't experimenting with diverse approaches and then five years later saying, my gosh, that was terrible for children's learning and frankly their mental health.

Seth (00:26:50) *(approx)*
And in your research practices — it sounds like that must be in the back of your head, this idea that this may be true for now but this is still a brand new phenomenon. How do you control for that concern — that we may not yet know — yet people are looking for guidance? Your report reads very intuitively. Back when they were teaching how to identify a good website in the mid-2000s, there was also guidance that sounded like it made sense. And then it turned out to be wrong. How do you make room for that possibility while also speaking with authority?

Rebecca Winthrop (00:27:25) *(approx)*
Two answers to that. That's exactly why we decided to do this task force as a collective exercise that was a pre-mortem. It was not evidence-free — it was let's gather the best evidence that is emerging based on what we already know about students' learning and development, because there are things that have fifty, sixty, seventy years of evidence behind them. For example, young people learn through interaction with other people. Humans have evolved for that. So it is unlikely they will learn beautifully all by themselves for eight hours a day just with a laptop. In fact, we tried that with COVID and it didn't work. So we take that as our central spine. And then in the pre-mortem we said: we're going to gather all the emerging evidence and put our finger in the air. Are we heading in the right direction? Because we're going to look at possible potential benefits, possible risks — and we do that by looking at all the emerging studies that have some rigor, and by talking to students, teachers, parents on the ground. We would have picked up a lot of the things we're now dealing with around social media if we'd really talked to families and teachers on the ground a decade ago — they would have said kids are distracted, or my kid comes home on the Chromebook and isn't doing the teacher-assigned YouTube video, they're quickly moving to something else. We would have known that ages ago. And we were very careful to say currently — for now, the way AI is implemented — it looks like risks are overshadowing benefits. But it's early days, and by the way, it's not a done deal. What we do now can shift that.

And answer two: I think we need a different mental model where the burden of proof is on the innovative AI use to show that it's beneficial. I worry about jumping in with one approach too early and scaling it and realizing it was the wrong approach. What we don't have now is a lot of external evaluations on these AI tools. Because of the risks, it behooves us to take a cautious approach — rather than the world is your oyster, we're going to close down as the risks unfold. There's a lot of risks, so we're going to keep going with what we know works for how kids learn. And when the burden of proof is surpassed — this is a really helpful narrow AI use for learning algebra for kids who are behind, or multilingual learners, or dyslexic kids, or exposing rural kids to a profession they never would have been able to job shadow — then that's when we say, okay, now we're bringing that in.

Seth (00:30:40) *(approx)*
One of the most salient findings from your study was around trust. This idea that AI is breaking trust in multiple directions — not just the immediate impact of teachers thinking students were cheating, but now you have students accusing their teachers of using AI too much. You have everyone not trusting AI companies. You have the one thing that Americans can agree on, which is that we hate data centers. You mentioned social learning. It does seem like relationships are the antidote to that lack of trust. Is it that simple? Do we just need to get back to the basics?

Rebecca Winthrop (00:31:20) *(approx)*
One of the things I think we — and I understand why — erred on as a collective sector, early on, was leaning into the students-are-cheating, I-as-a-teacher-have-to-catch-them mode. I think — and this could be wrong — but not long ago I believe something like half of the statewide AI policies coming out of the US say don't do that. Don't focus on anti-plagiarism. First of all, the technology is really hard. Second of all — maybe even more important — it ruptures the relationship. It makes a teacher a policeman. And that is not a conducive environment for anybody. Teachers hate it — that's not why they went into teaching. And students don't like it either. So I do think anything we can do to repair and build strong trusting relationships — among students, educators, their parents, the content that is being learned — is a good thing to do.

Seth (00:32:25) *(approx)*
Well, Dr. Winthrop, thank you so much for being here today. I'll link all of your stuff in the show notes, but I really appreciate your time, your work, your candor. Thanks for being here.

Rebecca Winthrop (00:32:35) *(approx)*
Thank you for having me on, Seth, and good luck with your work.

Seth (00:32:39) *(approx)*
Thank you.

---

[OUTRO — recorded separately]

That's a wrap on our conversation with Rebecca Winthrop.

Her distinction between cognitive stunting and cognitive bypass — whether AI is preventing a skill from forming versus just speeding up work you could already do — is the most useful lens I've found for cutting through the "ban it everywhere" versus "embrace it everywhere" debate. Your relationship with AI is a fingerprint, and it depends on what you've already built.

The trust point is the one I'm still sitting with: it's not just teachers distrusting students anymore. Students are watching whether teachers use AI, and that runs in both directions.

Links to the Brookings report and The Disengaged Teen are in the show notes.

Thanks for listening to Make It Mindful, and we'll see you next time.