Education Links, 3/27/2026

Adam Lehodey on classical schools; Daniel Buck on same; John Bailey on AI in education; Mark McNeilly on same

Adam Lehodey writes,

Our first class is Latin, one of the two courses that makes South Bronx Classical a “classical” school (the other is debate). The subject is taught daily to students from third grade and above. Learning Latin advances English comprehension, the school believes. For many students, it represents their first exposure to the art of learning for its own sake.

Daniel Buck writes,

Since the pandemic, classical education has undergone a renaissance, boasting 264 new schools founded between 2019 and 2023 alone.

…Only after the progressive movement deconstructed the once-assumed first principles of objectivity and order did a self-conscious movement aimed at restoring them (i.e., classical education) become necessary. So, while Pryor says Brookfield Academy embraces the term “classical,” she sees her school’s mission as simply returning to traditional practices that work.

While in Austin, we met a woman who teaches at a classical school started by Joe Lonsdale, who wanted that education for his children. Kindergarten and first grade are Hebrew immersion, before the children switch to learning Latin and Greek.

John Bailey writes,

the agentic wave sweeping Silicon Valley has yet to reach education in any meaningful way, and its arrival could produce far better tutors. Imagine a coordinated team of AI agents: one diagnosing exactly where a student’s understanding breaks down, another selecting problems tied to the curriculum, another deciding when to offer a hint versus when to let the student struggle, another evaluating the quality of the student’s reasoning rather than just checking answers, and a critic auditing the others to catch errors and prevent the system from becoming a crutch. The single AI chatbot tutor may prove far less effective than a team of specialists working in concert with both student and teacher.

AI works in education when it’s used not to give answers but to orchestrate the conditions for productive struggle. The result is deeper engagement and stronger outcomes.

He says that early studies show that AI used carelessly by educators can be harmful, but AI used carefully can help.

Mark McNeilly writes,

Students report using AI to save time (27%), improve the quality of their work (25%), generate ideas (16%), understand complex topics (13%), check their work (8%), overcome writer’s block (8%), and prepare for exams (2%). That’s not a picture of mass cheating. That’s a picture of students inventing their own personalized tutoring system.

Research on how learners actually prompt AI confirms this. Students are asking questions like:

• “I want to learn by teaching — ask me questions about this topic so I can practice explaining it.”

• “Identify the most important 20% of this material that will help me understand 80% of it.”

• “Create a practice quiz and ask me each question one by one.”

You can certainly encourage students to take this approach, instead of prompting the AI to give them the answer. The AI should be prompted to be prompting the student.

He also writes,

this means you can’t prompt intelligently without knowing what questions matter in your field, what assumptions are embedded in a problem, and what a good answer actually looks like. You also need domain knowledge to evaluate the output — AI produces confident, well-written responses that are sometimes incomplete or wrong. Without expertise, you can’t tell the difference.

This is why domain knowledge isn’t less important in the AI era. It’s more important.

As you know, I am skeptical of this take. I can foresee a time when the general AI is better than the domain expert at the expert’s own game.

substacks referenced above:

@

Does AI Belong in the Classroom?