What is the status on continual learning for LLMs?

I belive if this is solved, agi is very close. So id like to know if its solvable