How does a 25-year teaching veteran use Gemini to keep critical thinking first in her California classroom?

Kavita Khanna teaches technology electives to sixth and seventh graders at Impact Academy of Arts and Technology in Hayward. She has been in the classroom for a quarter of a century and now uses Gemini inside Google Drive to cut lesson-planning time while showing students that AI only mirrors human design.

The quarter-century arc

Khanna began her career in India when classroom technology meant learning basic computer navigation. Twenty-five years later she leads a middle-school technology elective in the Bay Area. The shift spans overhead projectors, early laptops, and now a generative model that drafts differentiated activities in seconds.

Inside the Drive workflow

She stores unit plans in Drive. Gemini reads them, suggests pacing, and writes scaffolded prompts.

A concrete example

Lesson planning once consumed entire evenings. Now Khanna types a learning objective — for example, "design a simple game in Scratch that teaches conditional logic" — and Gemini returns a full week-long sequence: warm-up discussion, pair-coding challenge, reflection journal, and extension task. She edits for voice, adds local context, and saves the revised version back to Drive.

Differentiation at speed

The model rewrites assignments at three reading levels, swaps vocabulary, and generates sentence frames for multilingual learners.

Making the black box visible

Middle schoolers hear about AI on social media and in headlines. Khanna projects the prompt window on the classroom display so students watch the model produce a draft, then they critique it together: hallucinated facts, biased phrasing, missing perspectives. The exercise turns a black box into a visible process.

The mantra

"AI only mirrors human design. Real power comes from our ability to reason, question, and keep our own critical thinking first."

What remains unsaid

No Gemini model version is named. The post does not disclose Workspace licensing or student achievement data.

Closing the loop

What remains clear is a practitioner's choice: use the model to handle routine drafting, keep final judgment human, and make the whole cycle visible to twelve-year-olds who will inherit the technology. The next unit in her drive already carries a new prompt — written by her, not the model.