Reasons for optimism around AI/agentic frameworks:
1. I think the training of LLMs will decline as “pretty good” is good enough for everything given a) cheap fine tuning and b) “agents” that correct/refine results.
2. AIs will increasingly run on commodity hardware in “normal” DCs.
3. Open source model development and open datasets will soon achieve similar results to today’s cutting edge models and be easy to extend and audit.

So AI is not crypto imho.
#AI #AgenticFrameworks