Anant Srivastava argues that prompts, retrieval and fine-tuningFine-tuning adapts a pretrained model by updating its parameters on examples chosen for a particular task or subject. solve different problems. He uses an assistant with obsolete product information to show how facts can remain in weights after a catalog changes.
He recommends prompts for stable behavioral instructionsPrompt engineering is the practice of designing and refining instructions, context, examples, and constraints to obtain useful AI outputs., scoped retrievalRetrieval is the process of selecting relevant stored information and returning it to an AI system for the current task. for changing or citable knowledge, and fine-tuning for stable patterns where capability or cost justifies it. He describes how memory, context and learned behavior can inform one anotherAI agent memory is stored information that an agent can retrieve and use across steps, sessions, or changing contexts. while monitoring for drift.
Watch on YouTube
