Research Appendix · CPA AI Strategy

Calvary Preparatory Academy | AI in Education Research Packet

Source research streams

This packet was synthesized from six concurrent X + web research runs executed April 27, 2026 using Nomion AI’s research pipeline (Grok x_search + web_search + Claude Sonnet 4.6 analysis). Each run covered both X discourse and institutional/academic sources. Full reports preserved at the paths below.

#	Research Stream	Full Report
1	AI tutoring & personalized learning K-12 (effectiveness, design, hybrid models)	`Workshop/research/topics/ai-tutoring-personalized-learning-k-12-high-school/20260427_081408/20260427_report.md`
2	AI literacy curriculum for high school students (workforce, frameworks, gaps)	`Workshop/research/topics/ai-literacy-curriculum-high-school-students-teachi/20260427_081410/20260427_report.md`
3	Christian school AI integration (theology, ethics, frameworks, discipleship)	`Workshop/research/topics/christian-school-ai-ethics-theology-classroom-inte/20260427_081412/20260427_report.md`
4	AI tools for K-12 teachers (lesson planning, grading, workflow, adoption)	`Workshop/research/topics/ai-tools-for-teachers-lesson-planning-grading-feed/20260427_081413/20260427_report.md`
5	Academic integrity, AI cheating, detection, college admissions policy	`Workshop/research/topics/ai-cheating-academic-integrity-high-school-college/20260427_081415/20260427_report.md`
6	Future workforce, AI skills, entry-level demand, employer expectations	`Workshop/research/topics/future-workforce-ai-skills-employers-demand-entry/20260427_081416/20260427_report.md`

Each report contains: live X discovery results (real posts with URLs), web search results, full analysis, and complete source citations. Total cited sources across the six runs: ~250 unique URLs.

Curated must-read list (~30 sources)

If the AI Working Group reads only these, they will have the working knowledge needed for Phase 1.

Foundational evidence on AI in K-12 (read first)

Stanford SCALE Initiative — The Evidence Base on AI in K-12 (2026). The single most authoritative synthesis. Distinguishes scaffolded AI (works) from general-purpose AI access (harms). scale.stanford.edu
Brookings Institution — What the Research Shows About Generative AI in Tutoring (January 2026). Synthesizes four RCTs across multiple countries; strongest evidence-based case for hybrid AI tutoring. brookings.edu
CMU Heinz College — Human Tutoring Improves the Impact of AI Tutor Use (September 2025). Year-long RCT; 350+ seventh graders; 0.36 grade-level gain for hybrid vs. AI-only. heinz.cmu.edu
Nature Scientific Reports — AI Tutoring Outperforms In-Class Active Learning (June 2025). RCT, n=316, double the learning gains in less time. nature.com
Wharton/PNAS (Bastani et al.) — the “crutch” study. Open ChatGPT access produced 17% worse exam scores despite 45% practice gains. The single most-cited cautionary study. Anand Sanwal summary on X
MIT/Oxford/CMU — AI assistance reduces persistence and independent problem-solving. Rohan Paul summary on X (paper at arxiv.org/abs/2604.04721)
Education Week — Why AI May Not Be Ready to Write Your Lesson Plans (June 2025). Analysis of 310 AI-generated civics lesson plans found systematic bias toward lower-order thinking. edweek.org

AI literacy curriculum & frameworks

OECD/European Commission — Empowering Learners for the Age of AI Framework (May 2025). Most comprehensive international AI literacy framework. Four dimensions + ethics. ailiteracyframework.org
ETS Research Report — Preparing K-12 Students With AI Literacy (November 2025). Grade-9–12 progressions; assessment-focused. rr.ets.org
NSF/PAR — DAILy Curriculum Effectiveness Study. Strongest empirical evidence that structured teacher-led AI literacy curricula produce measurable student gains. par.nsf.gov
EdWeek — Are AI Literacy Lessons Now the Norm? (March 2026). The defining paradox: 80% of schools claim to teach AI; only 5.4% of students receive formal training. edweek.org
Pew Research — How Teens Use and View AI (February 2026). The definitive survey; 54% use AI for schoolwork; only 25% confident in AI outputs. pewresearch.org

Christian school context

IACE — The W.I.S.E. Framework: A Theological Ethic for AI Tools in Christian Education (February 2026). The most actionable document in the field. Worship, Image, Sent, Equipped axes. The starting point for Calvary Prep’s policy work. iace.education
ACSI/Cardus — Navigating AI in Christian Schools (May 2024). Only large-scale survey of actual Christian educator AI adoption. cardus.ca
The Gospel Coalition — Don’t Hand Education Over to AI (August 2025). The clearest articulation of the discipleship-formation objection. Required reading as the loyal-opposition voice. thegospelcoalition.org
CESA — The Question of AI and the Christian School (June 2024). Creation/dominion theological framing for board-level conversations. cesaschools.org
MDPI Religions — Artificial Intelligence in Religious Education (April 2025). Most comprehensive academic review; recommends “human-in-the-loop” models. mdpi.com
Tim Hwang / ICMI — Christian theology and AI alignment. The most intellectually distinctive emerging voice. Christian theology as a technical resource for AI safety. X/@timhwang

Teacher AI tools & adoption

Gallup/Walton Family Foundation — Three in 10 Teachers Use AI Weekly (2025–26). The single most-cited teacher AI adoption study; 60% adoption, 6 hrs/week saved. gallup.com
Stanford SCALE — What K-12 Educators Are Actually Prompting to AI (March 2026). Real prompt-log analysis showing teachers use AI mostly for content generation. scale.stanford.edu
PDK/EdWeek — Americans Grow More Skeptical of AI in K-12 Schools (August 2025). Public support for AI in lesson planning fell from 62% to 49% in one year. edweek.org
SREB — Designing AI Professional Learning for Educators (2026). Four-phase PD model (build, explore, apply, iterate). sreb.org

Academic integrity & assessment redesign

Turnitin — Heavily AI-generated submissions data (February 2026). 15% of essays >80% AI-generated, up from 3% in 2023. prnewswire.com
AAC&U Faculty Survey (January 2026). 95% of college faculty fear AI overreliance; 74% say AI degrades degree value. aacu.org
ACT Research — High School Students’ Views of AI in College Admissions (January 2026). 70% of high schoolers view AI in applications as unfair advantage. act.org
Inside Higher Ed — The Best Defense Against AI Cheating (April 13, 2026). Argues redesigned assessment outperforms detection. insidehighered.com

Workforce & college outcomes

NACE — Demand for AI Skills in Entry-Level Jobs Nearly Triples Since Fall 2025 (April 2026). The defining workforce data point. naceweb.org
Stanford Digital Economy Lab — Canaries in the Coal Mine (November 2025). 13% employment drop for 22–25 year-olds in AI-exposed jobs since 2022. digitaleconomy.stanford.edu
PwC Global AI Jobs Barometer (2025). 56% wage premium for AI-skilled roles. pwc.com
Handshake — Class of 2026 Outlook. 70% of hiring managers consider generative AI skills critical for entry-level hires. joinhandshake.com

How this packet was built

Six parallel X + web research runs executed via Workshop/research/x_research.py on April 27, 2026
Each run covered the previous 30 days (45 days for Christian school stream) of X discourse plus web search
Discovery: Grok 4 with x_search and web_search tools (live X data with citations)
Analysis: Claude Sonnet 4.6
Total cited sources across runs: ~250
Synthesis time: ~25 minutes
Total cost (API): ~$15

The methodology, queries, and full discovery output are preserved in the topic directories. This is reproducible and updatable — the same pipeline can be re-run quarterly to track how the landscape evolves.

“Wise people store up knowledge, but the mouth of the foolish is near destruction.” — Proverbs 10:14 (NKJV)

This is the storehouse. The work now is to read, deliberate, and decide.