The District and the Problem
Riverside Unified School District is a suburban K-12 district with 12 schools and approximately 340 classroom teachers. Like most mid-size districts, Riverside was caught in a tension familiar to every instructional leader: the expectation that teachers differentiate instruction for diverse learners, and the reality that differentiation takes time that most teachers do not have.
In a post-pilot survey conducted by the district, teachers reported spending an average of 11.4 hours per week on lesson planning, materials preparation, and differentiation tasks outside of instructional hours. For a full-time teacher already managing 25 to 32 students per class across multiple preps, that number represented a significant portion of evenings and weekends.
"Teachers were burning out on the logistics of differentiation," said Priya Nambiar, Riverside's Director of Curriculum and Instruction. "They understood the why. They believed in meeting students where they were. But creating three versions of the same worksheet, at different complexity levels, for a lesson they also had to plan from scratch, that was unsustainable."
The Pilot: 12 Schools, 340 Teachers, One Semester
Riverside launched a one-semester pilot of AI lesson planning tools across all 12 schools, with participation from every grade level and subject area. The pilot was not voluntary, the district made a deliberate choice to deploy broadly rather than to self-selected early adopters, because they wanted data representative of the full teacher population, not just the most tech-comfortable staff.
Teachers received three hours of initial training, followed by a one-hour follow-up session at week four. After that, usage was self-directed.
The tools deployed in the pilot included:
Lesson Plan Generator. Teachers described their learning objective, grade level, subject, and any relevant constraints (available materials, class time, IEP accommodations needed) and received a complete lesson plan with a hook activity, instruction sequence, practice tasks, and formative check. Most teachers reported generating a usable first draft in under four minutes.
Choice Board Generator. For teachers who wanted to offer students multiple pathways through a concept, a key differentiation strategy, the choice board tool generated a 3x3 or 2x3 grid of activity options at varying complexity levels and learning modalities. A fifth-grade teacher could specify her learning objective and receive nine activity options she could review, edit, and print in a single workflow.
Differentiated Instruction Planner. For lessons where the teacher needed to explicitly plan for students at different readiness levels, the differentiated instruction planner generated parallel versions of core activities, the same concept, the same learning goal, but scaffolded at three complexity tiers. Teachers specified which students (by initials or group label, never by full name) were at each tier, and the tool generated materials accordingly.
What Teachers Reported
After the pilot semester, the district surveyed all 340 participating teachers. The results were consistent enough to be actionable.
Average weekly time savings: 6.1 hours. Time spent on lesson planning fell from 11.4 hours to 5.3 hours per week on average. The largest savings came from differentiation tasks: teachers who previously spent 3 to 4 hours per week creating tiered materials reported that the differentiated instruction planner reduced that task to 45 to 60 minutes.
23% more differentiated materials produced. Despite the time reduction, teachers produced more differentiated materials during the pilot semester than in the comparable semester the previous year. Having a faster path to differentiated content meant teachers who had previously skipped differentiation on some assignments, because they simply ran out of preparation time, now differentiated more consistently.
Lesson plan quality variance decreased. The district's instructional coaches evaluated a sample of 60 lesson plans from before and after the pilot using a standard rubric. Plans generated with AI assistance scored higher on clarity of learning objective, alignment between objective and assessment, and coherence of the instructional sequence. Variance in plan quality across teachers narrowed.
What Changed for Students
The 6 hours per week that teachers recovered did not disappear into extra rest (though no one would fault them if it had). The district tracked how teachers reported using recovered time.
The most common response: more time in individual and small-group conversations with students. Several teachers described spending the last 15 minutes of their planning period reviewing student work more carefully rather than scrambling to finish the next day's materials. A middle school math teacher in one of Riverside's higher-need schools described it this way: "I used to spend Sunday afternoon building worksheets. Now I spend that time calling three parents whose kids are struggling. That matters more."
Instructional coaches observed that classroom feedback quality improved during the pilot. When teachers were not racing through preparation, they were less fatigued during instruction and more responsive to the moments when a concept needed re-teaching.
The District's Rollout Decision
At the end of the pilot semester, Riverside's instructional leadership team reviewed the data and voted unanimously to make the AI lesson planning tools a permanent district resource. The district is now integrating the tools into its standard new-teacher onboarding curriculum, new hires learn to use the lesson plan generator in their first week.
"The argument we heard most often against AI tools in education was that they would reduce teaching to a mechanical process," said Director Nambiar. "What we actually found was the opposite. When the mechanical parts of planning got faster, teachers got more human. That was the outcome we were hoping for."
The district has begun exploring the AI assessment tools as a second phase, with a particular interest in how AI-graded exit tickets could give teachers real-time data about student understanding before the next lesson.