I agree that semantic search combined with LLM would be a huge leap forward. The financial costs of imitating human learning and using it for even bigger projects might be a pipe dream or revolutionary. We'll see in a few years. I hope to be here for the results.