SourceYouTube

Published January 10, 2026

DW Shift

Hosts:DW Shift

This is an AI-generated summary. The source video includes demos, visuals and context not covered here. Watch the video → · How our articles are made →

In Brief

DW Shift explores how two countries are racing to bring AI into classrooms. Estonia became the first country to roll out ChatGPT Edu nationwide, while India plans to teach AI and computational thinking from third grade to 250 million students. The documentary balances optimism with concern: personalized AI tutoring shows promise, but cheating cases at universities are tripling, and Rebecca Winthrop of the Brookings Institution warns that student work quality is declining. UC Berkeley professor Hany Farid argues the answer lies in moderation, not in banning or unrestricted access. For related coverage, see Why Banning AI in Classrooms Misses the Point and Alpha School: AI Tutors Teaching Kids 10x Faster.

The central claim

The documentary's core argument is that AI in education is inevitable, and the only real question is how to use it well. AI is already changing the way people learn, and it is doing so faster than schools and universities can react (0:00). Banning AI tools is impractical because students will use them regardless. The countries moving fastest, Estonia and India, are betting that bringing AI in on their own terms works better than pretending it doesn't exist.

Estonia: the first nationwide rollout

Estonia became the first country to give all 10th and 11th graders access to ChatGPT Edu, a version of OpenAI's chatbot designed for schools (2:50). The initiative is part of the country's AI Leap 2025 program, and builds on existing digital culture: one in four Estonians already actively uses ChatGPT (2:58).

At Mustamäe State Gymnasium in Tallinn, German teacher Karmen Kisel uses a straightforward approach. Students write sentences first, then ask the AI to correct them. But Kisel stresses that AI-based tasks must be carefully designed, or the AI simply provides answers instead of supporting learning (3:31). That means more preparation for the teacher, not less (3:48).

Educational technologist Emma Loore Sinitamm confirms that most teachers currently use AI to plan lessons rather than teach them (4:08). Many go to ChatGPT for a lesson plan, get a starting point, and build from there. The technology helps, but the knowledge gap among teachers is real.

India: AI from third grade

India is taking an even more ambitious approach. Starting with the new school year, AI and computational thinking, a way of breaking down problems into steps a computer can follow, will be taught from third grade onward (5:42). The scale is staggering: roughly 10 million teachers and 250 million students (5:58).

At Sapphire International School in Delhi, teacher Shabori Singh uses an AI app called Chrysalis to track each student's progress (6:22). The app categorizes children as beginners, progressing, or proficient, and suggests activities based on their level. In crowded Indian classrooms, this kind of personalization would be difficult without AI.

Computer scientist Anubha Gupta from IIIT Delhi sees AI as a gamechanger but warns about shortcuts. A Large Language Model (LLM) is an AI system trained on vast amounts of text to generate human-like responses. If students simply ask an LLM for answers instead of working through problems themselves, their critical thinking skills suffer (8:03).

The cheating crisis

Not everyone is integrating AI smoothly. In the UK, AI-related cheating cases at universities tripled in just one year (9:03). In the US, a study of over 1,000 students found that nearly half use AI in ways that violate university rules, and many don't see it as a problem (10:00).

Rebecca Winthrop describes a noticeable shift. Before ChatGPT, student papers were stronger and full of unique ideas. After ChatGPT, the ideas clustered together and became less diverse (9:33). This echoes concerns raised in a recent article about Stanford students drawing their own lines on AI use.

The dynamic creates a trust problem. "Professors don't want to be detectives. Students aren't criminals," Winthrop says (10:19). Some instructors are responding by shifting toward in-class exams, group work, and presentations that AI cannot easily replicate (10:37).

The case for moderation

Hany Farid rejects both extremes. Banning AI doesn't work because it's impossible to enforce. Unrestricted use is equally problematic (0:52). His approach: different rules for different levels. In introductory courses, students must do the work themselves to build foundational skills. In graduate courses, they get more freedom because they already have the knowledge base (1:08).

Farid also shares a compelling example from his own flipped classroom, a model where students watch lectures at home and do exercises in class. One student uses ChatGPT alongside his lecture videos, asking questions like "What does that mean?" and "Test me on this" (11:06). She essentially has a private tutor for every lecture. Farid calls it "incredible."

How to interpret these claims

The documentary presents powerful examples, but several questions deserve careful consideration.

Early-mover results are unproven

Estonia's AI Leap launched in 2025. India's curriculum starts in 2026-27. The real results won't be visible for years. Enthusiastic teachers at Mustamäe State Gymnasium and Sapphire International School are encouraging signs, but these are schools that volunteered to participate early. Whether the benefits hold when AI is rolled out to less-prepared schools remains an open question.

The cheating numbers need context

The tripling of cheating cases in the UK could partly reflect better detection rather than more actual cheating. Universities have invested in new AI detection tools during the same period. The numbers confirm a growing problem, but they don't tell us how much of the increase is new cheating versus newly caught cheating.

Teacher readiness is the real bottleneck

The documentary returns to this point repeatedly. Estonia's educational technologists say many teachers don't know how to use AI in class. India's plan covers 10 million teachers, but training infrastructure is still being built. The technology is advancing faster than the training programs designed to support it. Without adequate teacher preparation, even the best AI tools risk becoming sophisticated cheating aids rather than learning tools.

Practical implications

For educators

The most effective approach in the documentary is designing tasks where AI supports the process rather than providing answers. Kisel's German class model works well: students write first, then use AI to correct and learn from mistakes. The key is making AI a tool for reflection, not a shortcut.

For parents and students

Students who use AI as an active learning partner get the most value. That means asking it to explain concepts, test understanding, and provide examples. Those who use it to skip the learning process risk weaker critical thinking skills and potential academic consequences.

Glossary

Sources and resources