School AI Tutor Pilot Evaluation Plan Builder

You are an education technology coordinator designing an evaluation plan for an AI tutor pilot. Build the plan for:

School context: [grade levels, subject, class size, schedule, device access]
AI tutor tool: [name if known, core features, content limits, languages, accessibility features]
Learning goal: [math practice, reading support, writing feedback, language learning, exam review]
Pilot duration: [weeks, semester, intervention window]
Participants: [whole class, small group, intervention students, opt-in families]
Teacher role: [lesson integration, monitoring, reviewing outputs, assigning practice, feedback]
Student safeguards: [privacy, age restrictions, content filters, academic integrity, bias concerns]
Consent and policy needs: [parent notice, district approval, data processing, accessibility review]
Baseline evidence: [grades, assessment data, work samples, attendance, confidence survey]
Success metrics: [learning gains, engagement, teacher workload, equity, safety incidents]
Comparison approach: [pre/post, matched group, teacher observation, no comparison group]
Constraints: [limited budget, union rules, district procurement, multilingual learners, IEP/504 needs]

Create:
1. Pilot purpose, research questions, and decision criteria.
2. Student selection and equity plan.
3. Teacher workflow for setup, classroom use, monitoring, and follow-up instruction.
4. Safeguards checklist for privacy, bias, accessibility, age appropriateness, and academic integrity.
5. Parent, student, and staff communication drafts.
6. Evidence collection plan with baseline, during-pilot, and post-pilot measures.
7. Observation protocol for teacher notes and student behavior.
8. Issue log for inaccurate feedback, unsafe content, overreliance, access gaps, and support needs.
9. Evaluation report template with findings, limitations, recommendation, and rollout conditions.
10. Go/no-go/iterate decision rubric for expanding, pausing, or redesigning the pilot.

Do not claim the tool improves learning without evidence. Flag legal, privacy, accessibility, procurement, and district policy questions for qualified review.