AWS Certified AI PractitionerFoundation ModelsMedium
An AI solutions architect is evaluating different approaches for a client who wants to build a chatbot for customer support. The chatbot needs to provide accurate, up-to-date information from a proprietary knowledge base that is frequently updated. The client is concerned about the model generating outdated or incorrect information. Which foundation model technique is best suited to address this concern?
- ASupervised fine-tuning
- BRetrieval Augmented Generation (RAG)
- CFew-shot learning
- DZero-shot learning
Show answer & explanationAnswer & explanation
Correct answer: B. Retrieval Augmented Generation (RAG)
Retrieval Augmented Generation (RAG) is specifically designed to address the problem of foundation models generating outdated or incorrect information by coupling them with an external knowledge base. It retrieves relevant, up-to-date information from the knowledge base and then uses this information to condition the generation process, ensuring factual accuracy.
Why the other options are wrong
- A. Supervised fine-tuning adapts the model's weights but doesn't provide a mechanism for real-time access to frequently updated external data.
- C. Few-shot learning allows a model to learn from a small number of examples, but doesn't inherently prevent generation of outdated information from its original training data.
- D. Zero-shot learning allows a model to perform tasks it hasn't seen before, but doesn't address factual accuracy from external, dynamic data.
Retrieval Augmented Generation (RAG)
A technique that combines a retriever component with a generator model. The retriever fetches relevant information from an external knowledge base, which then guides the generator to produce more accurate and up-to-date responses.
- Mitigates hallucination and outdated information
- Enables models to access real-time or proprietary data
- Improves factual accuracy and trustworthiness of outputs
Memory trick: Search first, then speak true.