Bloom's Two Sigma problem — solved at scale. How teachers design for AI tutoring systems, adaptive practice platforms, and agentic learning agents to reach every student.
In 1984, Benjamin Bloom proved that one-to-one tutoring outperforms group instruction by two standard deviations — an effect so large it has haunted education ever since. AI is the first scalable solution.
Students who received one-to-one tutoring performed two standard deviations better than students taught in conventional group settings. That means the average tutored student outperformed 98% of control-group students. The challenge: one-to-one tutoring at scale is economically impossible — until now.
AI does NOT replace teachers. It handles personalized practice, adaptive challenge, and immediate feedback — freeing teachers to do what AI cannot: build relationships, facilitate discussions, recognize emotional needs, and make the high-stakes instructional decisions that require human judgment. Teachers become learning designers and interpreters, not just deliverers.
Before teachers can use AI purposefully, they need to understand WHY AI matters — not as a technology trend, but as the answer to one of education's most persistent research problems.
| Time | Activity | Description | Format |
|---|---|---|---|
| 0:00–0:15 | LaunchThe 1-to-1 Thought Experiment | "Imagine you could spend 30 minutes a day working with each student individually — just you and them, perfectly calibrated instruction. What would change?" Teachers write for 2 minutes, then share. Surface: pacing, feedback, knowing each student's exact gap. That's the 2 Sigma promise. | Write → share |
| 0:15–0:35 | TeachBloom's 1984 Study | Present the 2 Sigma study: design, findings, and why it mattered. The 2 standard deviation effect. Why hasn't education solved this? Cost, scalability, teacher-to-student ratios. What approaches came close? Mastery learning (Bloom's own solution — worked but slow). Now: AI. | Whole group |
| 0:35–0:55 | ExploreHow AI Approximates 1-to-1 | Walk through the Diagnose → Decide → Act → Reflect → Advance loop. At each step: what does a human tutor do vs. what does AI do? Where are the gaps? Where does AI actually exceed human performance? (Speed of feedback, consistency, patience, 24/7 availability, no social pressure). | Whole group |
| 0:55–1:15 | DemoLive AI Tutoring Demonstration | Principal (or CIL) demonstrates a live AI tutoring session using Khanmigo or equivalent. Show: adaptive questioning, immediate feedback, pacing adjustment, Socratic prompting. Teachers observe using an observation protocol: "What does the AI do that I can't scale? What does it miss that I would catch?" | Observation + discussion |
| 1:15–1:40 | DiscussThe Teacher's New Role | Discussion: "If AI handles personalized practice and immediate feedback, what is MY highest-leverage role?" Build the list together: relationship-building, high-stakes judgment, emotional attunement, discussion facilitation, critical thinking coaching, curriculum design. AI is the assistant. The teacher is the architect. | Whole group |
| 1:40–2:00 | CommitMy AI Readiness Baseline | Each teacher completes a brief AI readiness self-assessment: current comfort, current use, concerns, and one question they want answered before using AI with students. Principal collects — this drives Sessions 2–4 content priorities. | Individual |
Not all AI tools are equal. This session builds teacher fluency with the specific AI tutoring platforms appropriate for K–5, including how they work, what they're best for, and how to evaluate new tools.
| Time | Activity | Description | Format |
|---|---|---|---|
| 0:00–0:15 | SortAI Tool Taxonomy | Teachers sort 12 common ed-tech tools into three categories: (1) Drill/Practice (Duolingo, flashcard apps), (2) Adaptive Learning (Khan Academy, IXL), (3) AI Tutoring/Conversational (Khanmigo, Socratic). Discuss: how are these different? Which is closest to Bloom's tutor? | Groups |
| 0:15–0:35 | TeachThe AI Tool Evaluation Framework | Present a 5-criterion framework: (1) DoDEA/DoD approval status, (2) Student privacy compliance (FERPA/COPPA), (3) Evidence of learning impact, (4) Adaptive responsiveness (does it actually adjust?), (5) Teacher visibility into student progress. This framework applies to any new tool that emerges. | Whole group |
| 0:35–1:05 | ExploreHands-On: Be the Student | Teachers use an approved AI tutoring tool as if they were a student. Complete a 20-minute session in a subject area. Observe: How does it check understanding? How does it respond to a wrong answer? How does it pace? What feedback language does it use? Debrief: what worked? What felt off? | Individual |
| 1:05–1:25 | DesignMatch Tools to Learning Goals | Grade-band groups: for each major instructional goal this semester, which AI tool (if any) is the right match? Not all goals need AI. Build a simple decision tree: "If my goal is [X], the right tool is [Y] or [teacher-only]." Prevents AI from being used where it doesn't fit. | Band groups |
| 1:25–1:45 | DiscussPrivacy, Safety & Student AI Use | Critical discussion: What do parents need to know? What do students need to understand about AI? What are the age-appropriate expectations for K, 1–2, and 3–5? Draft 3 classroom norms for AI use that you could post in your room and explain to families. | Whole group |
| 1:45–2:00 | CommitOne Tool, One Week | Each teacher identifies one AI tool to pilot with students in the next two weeks. They complete a "pilot plan" card: which students, which skill, which tool, how they'll monitor, what data they'll bring to Session 3. | Individual |
AI tools only produce 2 Sigma results when teachers design instruction that integrates them intentionally. This session builds the design skills teachers need to make AI effective — not just present.
| Time | Activity | Description | Format |
|---|---|---|---|
| 0:00–0:20 | DebriefPilot Reports | Teachers share: "I piloted [tool] with [students] for [skill]. Here's what happened. Here's one surprise. Here's one question." Principal listens for patterns. Celebrate bravery. Surface common concerns to address. | Groups → whole |
| 0:20–0:40 | TeachThe AI-Integrated Lesson Structure | Present the recommended structure: (1) Teacher I Do (explicit model — AI can't do this), (2) Brief We Do (teacher + students together), (3) AI-supported You Do (students with AI tutor while teacher monitors and pulls small groups), (4) Teacher synthesis and closure. AI lives in phase 3. Teacher is present and responsive throughout. | Whole group |
| 0:40–1:05 | DesignBuild an AI-Integrated Lesson | Each teacher designs one complete AI-integrated lesson for next week: explicit I Do script, transition to AI session (instructions, expectations, duration), monitoring protocol (what the teacher watches while students are with AI), and closure/synthesis. Share with a partner for feedback. | Individual + pairs |
| 1:05–1:25 | MonitorThe Teacher Dashboard Habit | For each approved AI tool: demonstrate the teacher dashboard. What data is available in real time? What does it mean? Practice: given a dashboard showing 3 students struggling with the same concept and 4 students racing ahead — what does the teacher do RIGHT NOW? Role-play the decision. | Whole group |
| 1:25–1:45 | PlanFamily Communication About AI | Teachers draft a 3-sentence parent explanation of how AI is used in their classroom. Must answer: What tool? What does it do? How is my child's information protected? Share, critique, refine. Principal collects the best versions — these become the template for the school-wide parent letter. | Individual → share |
| 1:45–2:00 | CommitAI Design Goal + Coaching Preview | Teachers commit to teaching one full AI-integrated lesson before Session 4. Principal previews the coaching visit: "I'll be there to observe the transition into the AI session and your monitoring behavior — not the AI itself." This is the highest-leverage observation moment. | Whole group |
The DoWEA director specifically named agentic AI for Tier 2 gifted learners. This session gives teachers the framework to design genuinely challenging, AI-extended learning experiences for high-growth students.
| Time | Activity | Description | Format |
|---|---|---|---|
| 0:00–0:15 | DefineWhat is "Agentic" AI? | Present the distinction: adaptive AI responds to student input → agentic AI sets goals, makes strategic decisions, reflects on learning, and self-directs within guardrails. For gifted learners: agentic AI becomes an intellectual peer or research partner, not just a tutor. | Whole group |
| 0:15–0:35 | TeachThe Diagnose→Decide→Act→Reflect→Advance Loop for Gifted Learners | Return to the agentic AI loop. For Tier 2 students: AI doesn't just adjust difficulty — it presents genuine intellectual challenges, asks "What if?" questions, invites the student to justify reasoning, introduces counterexamples, and proposes transfer to new domains. Show examples at grades 3–5. | Whole group |
| 0:35–1:00 | DesignAgentic AI Learning Experiences by Grade | K: not appropriate — focus on adult-supervised adaptive tools. Primary: semi-structured AI research conversations with teacher-designed guardrails. Intermediate: extended AI-guided inquiry projects where students pose questions, research, analyze, and present — with AI as a thinking partner throughout. Groups design one experience. | Band groups |
| 1:00–1:20 | EstablishGuardrails & Teacher Role | Critical: agentic AI requires more teacher oversight, not less. Design the guardrails: clear project scope, required teacher check-ins, output format expectations, reflection requirements. The teacher's role shifts to: project approver, progress reviewer, quality challenger. Role-play a teacher-student-AI three-way conference. | Whole group |
| 1:20–1:40 | AI LiteracyTeaching Students to Use AI Ethically | What do students need to know about AI? Design age-appropriate AI literacy mini-lessons: K (what is a computer program?), Primary (AI makes mistakes — here's why), Intermediate (AI bias, citation, and the difference between AI assistance and AI replacement). Teachers draft one lesson per band. | Band groups |
| 1:40–2:00 | Plan90-Day AI Integration Roadmap | Each teacher maps their next 90 days: which tools, which students, which skills, which checkpoints, and what they'll share with the principal at each coaching cycle. The GES AI Roadmap becomes a living document — revisited at every coaching cycle. | Individual |