Black Box
Domain: Safety & Ethics
🎯 Motivation
When the AI’s decision-making is opaque - you cannot see inside. Deep learning models are often black boxes.
📋 Concrete Example
Neural network: input goes in, output comes out, but the 175B parameters in between are a mystery. You cannot easily explain why it made a specific decision.
🔗 Analogy
A vending machine - you put in money and get a snack, but you cannot see the internal mechanism. It works, but you do not know how. Black box AI works but does not explain how.