Randomized controlled trials are the wiki's gold-standard evidence: genai-policy-prompting-rct, genai-can-harm-teaching-rct-2026, access-not-enough-ai-tutoring-2026, and lets-chat-chatbot-outreach-2026 are pre-registered RCTs demonstrating that AI's educational effects are real, heterogeneous, and sometimes negative (RCT, efficacy-study).