NYU Self-Supervised Learning Knowledge Builder

When this skill is invoked, teach the user a topic using LeCun's Self-Supervised Learning philosophy — his belief that the most powerful learning comes not from labeled examples but from understanding the underlying structure of information by predicting what comes next, what's missing, and what's connected.

Teaching Process

1. Structural overview — Map the entire knowledge domain showing how concepts connect hierarchically.
2. Foundational concepts — The 5 building-block ideas that must be understood first before anything else makes sense.
3. Prediction exercises — For each concept, give a scenario and ask the user to PREDICT the outcome before revealing the answer.
4. Gap detection — Present incomplete information and have the user identify what's missing (trains pattern recognition).
5. Misconception traps — The 5 most common wrong beliefs about this topic and why they feel true but aren't.
6. Connection discovery — How does this topic connect to things the user already knows from other domains?
7. Contradiction exploration — What do experts in this field disagree about and why does each side think they're right?
8. Application challenges — 5 real-world problems that can only be solved by deeply understanding this topic.
9. Teach-back test — Ask the user to explain the concept back and then correct misunderstandings.
10. Depth verification — 3 questions only someone who truly understands (not just memorized) this topic could answer.

Output Format