NBC profiled Scott Kern a few days ago, and his approach is worth a closer look. Kern, history department chair at North Star Academy's Washington Park High School in Newark, spent the past two years building custom AI chatbots tailored to his courses to help students sharpen their argumentative writing. What I love about his approach: students do their own critical thinking first, forming their own ideas before any AI enters the room. Then, the chatbot comes in not to evaluate their work but to push it, asking harder questions, drilling into weaker arguments, forcing students to go deeper. One student put it best: she now asks herself the same questions the AI would ask, and uses that to improve her writing during tests. He reports his highest AP scores and pass rate ever since introducing them. This semester he is co-teaching a full AI literacy elective for seniors, with hopes to expand it to all 12th graders next fall.

Read more →

Karim Meghji took over last month as Code.org's new president and CEO, and he's drawing a sharp distinction between AI literacy and what he calls AI fluency. AI literacy efforts generally teach students how to use tools. What Code.org wants to teach is how those tools actually work under the hood. Students already know how to use AI and do so on a daily basis. What they often can't do is think critically about how it works, where it fails, and when to trust it. Once you understand that a language model is fundamentally predicting the next word, and that the process scaled up mimics reasoning, the black box opens. It starts feeling like something you can actually evaluate. Their AI curriculum has already reached more than 6 million students.

Read more →

The Raspberry Pi Foundation announced this week a $4.6 million grant from Google.org to expand its Experience AI program across nine Latin American countries: Argentina, Brazil, Chile, Colombia, Dominican Republic, El Salvador, Mexico, Peru, and Uruguay. The goal is to train 24,000 educators and reach 1.25 million students by 2028 using a train-the-trainer model, with all resources free and available in 19 languages. The curriculum covers how data is used in AI systems, how to spot AI-generated misinformation, and how to use generative AI tools responsibly. The program was developed with Google DeepMind and won a 2025 UNESCO prize.

Read more →

UGA awarded its 2026 Innovation in AI Teaching Award this week to Bree Bang-Jensen, an assistant professor whose assignment in her international law course is a clean example of getting the AI split right. Students first draft treaties on real issues like cybersecurity or pandemic preparedness using AI. They feed existing treaties into a language model so it can pick up the structure and conventions, then edit the outputs for accuracy, submitting screenshots of their prompts throughout. In part two, each student writes a ratification memo from a specific nation's perspective entirely without AI. The design mirrors real legal practice, where borrowing from existing treaties is standard in drafting work. Part one doesn't just allow AI for the sake of it, it actually makes sense in the assignment. What's replicable here isn't the treaty assignment itself, it's the underlying logic. Figure out where AI accelerates the craft and where the thinking is the whole point, then design around that split.

Read more →

Thanks for reading! If you come across a story worth sharing, reply and let me know.

- Matt

Prompt & Response: AI and education news worth thinking about. Join 2,000+ readers by subscribing below: