Slide 1: Constitutional AI: Ensuring Harmlessness Through AI Feedback. Slide 2: The Growing Need for AI Safety. Slide 3: Introducing Constitutional AI. Slide 4: The Two Phases of Constitutional AI Training. Slide 5: Phase 1: Supervised Learning (SL). Slide 6: Phase 2: Constitutional Reinforcement Learning (CRL). Slide 7: The 'Constitution': Guiding Principles for AI Behavior. Slide 8: Advantages of Constitutional AI. Slide 9: Constitutional AI vs. Reinforcement Learning from Human Feedback (RLHF). Slide 10: Benefits of Using AI for Feedback. Slide 11: Addressing Limitations of Traditional RLHF. Slide 12: Potential Applications of Constitutional AI. Slide 13: Future Directions and Research. Slide 14: Conclusion: Towards Safer and More Aligned AI. Slide 15: Q & A