By Dr. Janette Camacho | April 30, 2024
If you ask any veteran teacher what consumes the most invisible hours of their week, the answer is almost always the same: planning. Not the creative spark of designing a great lesson - that part is energizing. It is the scaffolding around it - aligning to standards, building differentiated materials, writing rubrics, anticipating misconceptions, creating formative checks. A 2023 McKinsey report estimated that teachers spend 7 to 12 hours per week on planning and preparation. After 28 years in the classroom, that number feels conservative.
Over the past six months, I have fundamentally restructured my planning workflow using AI tools - primarily ChatGPT-4, Google Gemini, and several purpose-built EdTech platforms. The results have been significant enough that I want to share specific, replicable examples rather than vague enthusiasm.
The Framework: AI as Planning Partner, Not Planning Replacement
Before I walk through examples, I want to establish a principle that governs everything I do with AI in planning: the tool generates, the teacher decides. AI is extraordinarily good at producing volume - multiple versions of an activity, alternative assessment formats, vocabulary scaffolds at varying Lexile levels. It is not good at knowing your students. That judgment - which version fits your third-period class that has four English learners and two students with IEPs - remains entirely human.
I think of AI as a first-draft machine. It gets me from blank page to working draft in minutes instead of hours. My professional expertise then shapes that draft into something that actually works in my room, with my students, on that particular Tuesday.
Example 1: Standards-Aligned Unit Mapping
When I began planning a unit on data analysis and probability aligned to Florida's B.E.S.T. standards, I started by feeding ChatGPT-4 the specific benchmarks I needed to cover (MA.6.DP.1.1 through MA.6.DP.1.6). I then asked it to generate a 12-day unit map with the following constraints: spiral review of prior concepts, at least two formative assessment checkpoints, one performance task, and integration of real-world data sets.
The first output was usable but generic. The real value came in the iterative refinement. I asked it to adjust pacing for a block schedule, incorporate stations for small-group differentiation, and suggest specific data sets that would be relevant to middle schoolers in South Florida. Within three rounds of prompting - about 15 minutes total - I had a unit skeleton that would have taken me two to three hours to draft manually.
The critical step: I then reviewed every benchmark alignment by hand. AI occasionally maps standards loosely, connecting an activity to a benchmark through surface-level keywords rather than true conceptual alignment. I found two instances where the suggested activities addressed adjacent standards but not the specific ones I targeted. This is why human review is non-negotiable.
Example 2: Differentiated Reading Materials
One of the most time-consuming aspects of inclusive lesson design is creating materials at multiple reading levels. For a science-integrated lesson on climate data interpretation, I needed the same core content accessible to students reading at grade level, students two years below grade level, and advanced readers who could handle primary-source complexity.
Using Gemini, I provided the grade-level text and asked for three versions: one simplified to a fourth-grade reading level with additional context clues and vocabulary support, the original, and one enriched version that incorporated excerpts from an actual NOAA technical report. I also asked for comprehension questions at each level that targeted the same learning objective but through appropriately scaffolded cognitive demands.
The output required editing - AI tends to over-simplify when reducing reading level, sometimes stripping out the precise academic vocabulary students need to learn rather than avoid. But the base materials were solid, and the entire differentiation process took 20 minutes instead of the 90 minutes it typically requires.
Example 3: Rubric Development with Embedded Feedback
Rubrics are among the most tedious artifacts to build well. A strong rubric does not merely sort student work into "exceeds, meets, approaching, below" categories - it provides specific, actionable language at each performance level that students can use to self-assess and improve.
I have started using AI to generate single-point rubrics with elaborated feedback stems. For a persuasive writing assignment, I provided the AI with the assignment description, the relevant ELA standards, and three anonymized examples of student work at different quality levels. I asked it to reverse-engineer the rubric criteria from the work samples and then generate feedback sentence starters for each criterion.
The result was a rubric with five criteria, each accompanied by two to three feedback stems such as: "Your claim is clearly stated, but the supporting evidence in paragraph 2 relies on personal opinion rather than the textual evidence required by the assignment. Consider revising by..." This kind of specific, constructive feedback language is exactly what saves teachers time during the grading process, and AI generates it remarkably well.
Example 4: Anticipating Student Misconceptions
This is perhaps my favorite use case. When I prompt AI with a specific concept - say, the difference between correlation and causation - and ask it to generate the five most common student misconceptions along with diagnostic questions that surface each one, the results draw on a knowledge base far broader than any single teacher's experience.
For the correlation-causation unit, ChatGPT identified misconceptions including the belief that a strong correlation coefficient proves causation, confusion between lurking variables and confounding variables, and the assumption that temporal sequence (A happened before B) establishes causation. It then generated quick formative prompts - three-minute "exit ticket" scenarios - that would reveal which students held each misconception.
I used these diagnostics on day two of the unit and was able to form targeted small groups on day three based on specific misconception patterns. Without the AI-generated diagnostic framework, I would have relied on a more general formative assessment and caught the misunderstandings later.
Example 5: Parent Communication Templates
This is a smaller but meaningful time savings. I use AI to draft parent communication templates for units that involve AI tools, explaining what tools students will use, what guardrails are in place, and how parents can support learning at home. These communications take sensitivity and precision - parents understandably have questions about AI in their children's classrooms. Having a well-structured draft that I can personalize saves me 30 to 45 minutes per unit and ensures I do not forget key information.
The Ethical Dimension
I want to be transparent about something: when I share these practices at workshops and conferences, I sometimes encounter pushback framed as an ethical concern. "If AI writes your lesson plans, are you really teaching?" The question misunderstands the process. AI does not write my lesson plans. It accelerates the mechanical components of planning - alignment mapping, material differentiation, rubric formatting - so that I can invest more of my finite cognitive energy in the creative and relational dimensions of teaching that no AI can replicate.
The teachers I worry about are not the ones using AI to plan more efficiently. They are the ones spending Sunday evenings exhausted at the kitchen table, hand-building resources that a tool could draft in minutes, because no one has shown them a better way.
Getting Started
If you are new to AI-assisted planning, start with one use case - I recommend rubric generation - and commit to it for two weeks. Use a structured prompt that includes your standards, grade level, assignment description, and any specific constraints. Review every output critically. Keep what works, revise what does not, and build your prompt library iteratively.
The planning revolution is not about replacing teachers. It is about returning to teachers the hours that bureaucratic complexity stole from them, so they can do what they entered this profession to do: teach.
Dr. Janette Camacho is a Google for Education Certified Trainer & Coach, Google Certified Educator Level 1 & 2, Adobe Creative Educator, Apple Teacher, and FETC 2024 Featured Presenter with 28+ years of K-12 classroom experience. She is the founder of iTeachAI.