Part 8/11:
Shapiro’s exploration reveals a recurring theme: the complexity of creating a truly general-purpose, mind-like AI. While many technical solutions bleed into the realm of research, his focus remains pragmatic. For example, focusing on:
Hierarchical storage: Raw logs for detailed queries, summaries for quick insights, knowledge graphs for relationships.
Scaling search infrastructure: Using Azure Search, vector search, and embedding techniques to ensure relevance.
Minimal dependency on large models: Offloading as much as possible to faster, more economical systems.