Smarter AI in Schools
Chatbots took schools by surprise. Suddenly, students carried an app in their phones that could magically answer almost any homework question or spin up an essay in seconds. As we move through 2026, this initial shock has given way to a more measured exploration of how AI can genuinely enhance learning—without undermining it.
Organizations ranging from OpenAI to UNESCO encourage AI use in the classroom, but many teachers aren’t sure how to handle it. Yet some promising approaches are emerging.
At Cheshire Academy, for example, teachers are trained on general techniques for using AI rather than prescribed tools. They use generative AI to prepare class materials, and one teacher has developed a traffic light system that tells students when they can use AI for assignments. This approach gives teachers flexibility while maintaining clear boundaries for students.
The school has also explored specialized tools, including an AI platform for educators. Here’s how it works—and how other teachers can apply it.
—Peter Hall
A Robot Carnival in Shanghai
In Shanghai, a different kind of AI revolution is on display. A recent exhibition dubbed a “robot carnival” featured dozens of humanoid robots performing tasks ranging from serving tea to dancing in sync. The event, part of China’s broader push to lead in robotics by 2030, highlighted both the impressive progress and the remaining challenges in making robots truly useful in everyday life.
Visitors interacted with robots that could recognize emotions, navigate crowds, and even engage in basic conversation. But experts note that behind the spectacle lies a crucial question: how to scale these technologies from controlled demonstrations to real-world applications.
The Frontier AI Safety Gap
Amid these advances, a sobering reality persists: no one really knows how to safely test frontier AI models. As models grow more capable, existing evaluation methods struggle to keep pace. Researchers are calling for new frameworks that can assess not just performance, but also potential risks—such as misuse or unintended behaviors—before deployment.
This gap is particularly pressing as AI integrates deeper into education, healthcare, and public infrastructure. The promise is enormous, but so are the stakes.
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.