AI Safety
Domain: Safety & Ethics
🎯 Motivation
AI systems can cause harm - misinformation, bias, security vulnerabilities. AI safety ensures systems are reliable, controllable, and do not cause damage.
📋 Concrete Example
A chatbot that gives dangerous advice (how to make weapons) is unsafe. Safety measures: filter harmful content, refuse dangerous requests, add disclaimers, human review.
🔗 Analogy
A car’s safety features - seatbelts, airbags, ABS brakes. The car is useful but could be dangerous. Safety features prevent harm. AI safety does this for AI systems.