> IMO this is a context window issue. Not really. It's a consequential issue. No matter how big or small the context window is, LLMs simply do not have the con

> IMO this is a context window issue. Not really. It's a consequential issue. No matter how big or small the context window is, LLMs simply do not have the concept of goals and consequences. Thus, it's difficult for them to acquire dynamic and evolving "skills" like humans do.