Answer: Reinforcement learning with rewards for positive customer feedback

## Explanation **Correct Answer: B - Reinforcement learning with rewards for positive customer feedback** Reinforcement learning is the most appropriate AI learning strategy for this scenario because: 1. **Self-improvement capability**: Reinforcement learning enables an AI agent to learn through trial and error by receiving rewards or penalties for its actions. The chatbot can learn from past interactions by receiving positive feedback (rewards) for good responses and negative feedback (penalties) for poor responses. 2. **Continuous learning from interactions**: As the chatbot interacts with customers, it can continuously improve based on customer feedback, which aligns with the requirement to learn from "past interactions." 3. **Adaptation to changing patterns**: Reinforcement learning allows the chatbot to adapt to new types of inquiries and changing customer needs over time. **Why the other options are incorrect:** - **A. Supervised learning with a manually curated dataset**: This requires human intervention to label data and doesn't provide true self-improvement from ongoing interactions. - **C. Unsupervised learning to find clusters**: This helps identify patterns but doesn't inherently improve response quality based on feedback. - **D. Supervised learning with continuously updated FAQ database**: While this incorporates new information, it still requires manual updates and labeling, not autonomous learning from interactions. **Key AWS Service Context**: AWS offers reinforcement learning capabilities through services like Amazon SageMaker RL, which can be used to build intelligent chatbots that learn and improve over time.

Author: Ritesh Yadav