feat: AI & Data Literacy path — how AI "thinks" (exploratory, needs product sign-off)

Summary

Explore a fourth top-level path teaching kids ages 5–14 the basics of how AI/pattern-matching systems work, at a level appropriate for each age band — "teach the robot" categorization games, garbage-in-garbage-out, and (for older kids) recognizing that AI can be wrong or biased.

Status: exploratory, lower confidence than the other three ideas in this batch

Flagging explicitly: this is the most novel and hardest-to-scope-well idea of the four. Before this goes to a Planner agent, it needs a product-direction decision, because:

• It's further from the app's current "coding fundamentals for kids" brand than the other three ideas.
• Age-appropriateness is harder to nail — explaining AI without either trivializing it or introducing concepts (training data, bias, misinformation) that are genuinely hard to make honest and age-appropriate for a 5–7 year old.
• Risk of the content aging poorly or needing frequent revision as "how AI works" in the public conversation shifts, unlike stable CS fundamentals (a loop is a loop; "how AI works" explained to a kid in 2026 may need rewriting in 2028).

If greenlit, proposed shape

Ages 5–7: very light — no ML terminology at all, just "teach the robot" pattern-sorting games (sort shapes by a rule, robot guesses the rule).

Ages 8–10: "teach the robot" categorization with visible cause/effect of bad training examples (robot trained only on photos of orange cats can't recognize a grey cat — reframe as an in-game character, not a real AI model).

Ages 11–14: critical-thinking scenarios — "the robot gave a confident answer, but is it actually right? What would you check?" Ties back to the existing Bug Lab world's lesson that "computers only do what you tell them."

Explicit non-goals

• No claim that this teaches real machine learning.
• No mention of any specific real AI product or company.
• No open-ended chat/generation — everything stays as pre-authored scenario puzzles, same static/offline model as every other path (INV-P1).

Next step

Do not scope directly into a plan — first needs a go/no-go conversation on framing and age-appropriateness (see "Status" above). If greenlit, suggested plan slug: feat-ai-literacy-path.

No files, tests, or entry points are named. Start with the required product-direction discussion on whether the path fits the app and how its age bands should be framed; work is ready for implementation only after a go/no-go decision produces an agreed scope, potentially under feat-ai-literacy-path.