LLMs can reason to an extent. I believe it’s entirely possible even today to achieve AGI. A complex system made out of LLMs, maybe Recursive Retrieval Augmented

LLMs can reason to an extent. I believe it’s entirely possible even today to achieve AGI. A complex system made out of LLMs, maybe Recursive Retrieval Augmented Generation (use an LLM to retrieve) or Recursive Agent Teams (any agent can assign a job to a new subcommittee of agents) or both. The most fundamental data structure is a tree, after all. And that is how we think, no? One thought leads to another in a tree structure.