I would bet that building a "self-aware" agent from an LLM might be trivial. Add: - an ability to do actions(tasks) - an episodic memory to record it's actio

I would bet that building a "self-aware" agent from an LLM might be trivial. Add: - an ability to do actions(tasks) - an episodic memory to record it's actions and report on those actions - a database of the agents abilities and goals. It's not AGI but would be self aware, I think.