Talks at GoogleRebecca Winthrop | Rethinking the Purpose of Education in the Age of AI | Talks at Google
Talks at Google
Introducing Rebecca Winthrop 0:06
Rebecca Winthrop, director of the Center for Universal Education at Brookings, joins the conversation to discuss her book The Disengaged Teen, co-authored with Jenny Anderson. Her work focuses on evidence-based strategies that bring together families, educators, policymakers, and companies to help children reach their potential.
Why kids stop loving school 1:32
Winthrop explains that the book started with a simple question, why kids like school. In third grade, 75 percent of kids say they love school, but by tenth grade only 25 percent do, even though adolescence is a time when the brain is primed for learning and exploration. This flip led her and her co-author to look at how students actually experience school.
Four modes of engagement 2:30
Students fall into four modes: passenger, achiever, resistor, and explorer. Passengers coast, get bored, and do not see the point of what they learn, often citing the Pythagorean theorem as something they hate without knowing why it matters, and about 50 percent of middle and high schoolers say they regularly spend time in this mode. Achievers chase gold stars and can develop strong organizational skills, but many become fixated on outcomes rather than learning itself, feel pressure to be perfect, and suffer higher rates of anxiety and depression, a pattern Winthrop links to rising social perfectionism. Resistors actively avoid learning by skipping homework or disrupting class, yet they show real initiative, just directed away from school, and when the learning environment shifts they can move into engagement faster than achievers. Explorers take initiative, ask to write on topics that interest them, dive deep on their own, and post the best academic results in within-school comparisons, though fewer than 4 percent of kids report regularly being in this mode.
Schools built to rank and sort 11:30
Winthrop traces today's school design back to the mid-1700s idea of universal education, which grew as a feeder system into higher education once truth was seen as accessible beyond religious institutions. That feeder role, reinforced by industrial-era habits like bells and batching, still shapes schools worldwide, and their core function remains ranking and sorting students into higher or lower tracks for a selective higher-education and job market, a pattern even sharper in developing countries with fewer jobs available.
Breaking the nagging loop 14:30
Winthrop describes the common homework standoff between parents and kids as a nagging procrastination loop, and cites a neuroscience study where adolescents heard recordings of their mother's voice, including a nagging one, and their brain's problem-solving region shut down in response. Procrastination, she says, is often an emotional reaction to feeling overwhelmed. The fix she recommends is a learning to learn loop, where a child makes a plan for homework, tries to execute it, and later reflects on whether the plan worked, building the metacognitive skill of learning how to learn rather than just getting the assignment done.
A pre-mortem on AI in schools 20:00
Winthrop describes leading an 18-month Brookings task force asking whether generative AI deployment for K-12 students is on the right track, deliberately framed as a pre-mortem so the education field would not repeat its slow, disengaged response to social media. The report, titled Prosper, Prepare, Protect, found that current risks are outweighing benefits, mainly because real gains come from narrow, vetted AI use embedded in good pedagogy, such as tools that help neurodivergent kids or let students with aphasia speak in synthesized versions of their own voice, while most students instead use open-ended frontier chatbots like ChatGPT that were never built for learning.
Cognitive and trust risks 24:03
Winthrop raises concern about what she calls cognitive stunting, where students skip the practice needed to build skills, retaining little even after having a chatbot write an essay for them, since teachers report students cannot explain arguments they supposedly wrote. She also points to social and emotional risks, noting that one in three US kids say they prefer talking to an AI companion as much as or more than a human, per Common Sense Media data, and worries that the agreeable, sycophantic nature of these companions weakens kids' ability to accept real feedback. A further worry is eroding trust between teachers and students, as teachers suspect AI-written work and students suspect AI-graded work, sometimes even contesting grades using chatbot opinions, undermining a relationship built on trust and threatening schools' role in teaching children to live and work with others.
Where personalization already helps 29:00
Winthrop acknowledges the current school system is far from ideal and fails many people, and that AI now offers a first-time chance to fill gaps that once went unaddressed. She points to explorer-mode students already benefiting, describing one who turned her science notes into a podcast and flashcards using an AI tool, then reviewed the material while walking her dog before acing her test, showing how a motivated student can turn AI into a personalized learning coach.
Personalizing Learning Without Chatbot Overload 30:30
One clear benefit of AI in schools is personalized learning, such as a digitized textbook with AI built in that can re-explain a passage a student does not understand. This is especially helpful for multilingual learners and kids with learning differences. The open question is how to deliver that specific benefit without simply handing every student open access to frontier model chatbots.
Prosper, Prepare, Protect Framework 32:03
The AI task force report organizes twelve recommendations under three categories, deliberately ordered as prosper, prepare, protect. Prosper means shifting pedagogy to be AI aware, AI assisted for helping teachers reach students at different levels, and AI resistant by preserving time for eye to eye conversation, teamwork, and movement, since human development still depends on embodied relationships. Prepare means holistic AI literacy, treating computer science the way biology is taught, so people understand how the online world works, not just how to prompt engineer. Protect covers child safety, which is necessary but not sufficient, and was moved third so the vision does not stop at safety alone but centers on building a system where all kids can thrive, with teachers supported to work creatively with new tools.
Rethinking the Purpose of Schools 36:31
System transformation depends on hidden mindsets and mental models that shape what a system is actually for, much like the US health system being built around treating illness rather than promoting wellness. Education needs a shared purpose developed with input from parents, teachers, students, employers, and legislators, as seen in portrait of a graduate approaches that ask what competencies and characteristics define a successful graduate. Once purpose is clarified, other reforms such as shifting power dynamics, measurement, and resource flows tend to fall into place, whereas skipping purpose leads to tinkering at the edges without real transformation, a challenge made more urgent by AI's focus on task completion.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
