Designing With AI

AI accelerated my work. Twenty years of teaching informed how I used it.

Understand→Design→Try→Refine

After more than a decade in school and district leadership, I returned to classroom practice and spent a year examining where generative AI could meaningfully support teaching and where professional judgment remained essential.

Original classroom artifact: an excerpt from the cumulative assessment.

Using AI Across the Instructional Cycle

This past year, I taught Early American History to 100 students across five sections in the Bronx, including Honors and ENL. I used Claude and Gemini throughout the instructional cycle — not as substitutes for pedagogy, but to accelerate first drafts and test possibilities.

I conducted this work within a full teaching load, without a separate design team or pilot structure. AI accelerated option generation and first drafts; it did not determine which version best matched my students, which concepts warranted additional instructional time, or which activities were most appropriate for a particular class.

Define the goal · Provide context · Generate options · Exercise judgment · Teach · Analyze student evidence · Refine

Design Principles

Where Professional Judgment Remains Essential

Context determines output quality — The strongest improvement came from providing the tool with my actual unit plan, student context, and instructional constraints rather than relying on generic prompts.
Maintain durable instructional context — I configured the tools to retain curriculum, rubrics, and instructional sequence across sessions, reducing the need to reconstruct context for each interaction.
Treat AI output as a draft requiring expert review — Plausible output can still contain misaligned reading levels, ineffective examples, or flawed assessment items; identifying and correcting these issues requires instructional expertise.
Increase access without reducing cognitive demand — Differentiation should provide additional support while preserving the same underlying intellectual task.
Use student evidence as the measure of effectiveness — I analyzed the exam question by question and used patterns in student responses to determine what required reteaching or further instructional support.

Selected Artifacts

AI-Enabled Prototyping Across the Instructional Cycle

Next: back to Home →

AI accelerated my work. Twenty years of teaching informed how I used it.

Using AI Across the Instructional Cycle

Where Professional Judgment Remains Essential

AI-Enabled Prototyping Across the Instructional Cycle

Cumulative Assessment Design

Differentiation & Multilingual Access

From Textbook to Multimedia

Question-Level Data Analysis