Again, my issue with the statement is not about current LLMs. Who knows if LLMs cannot be a major component of AGI, or even ASI systems. Then the capability to do whatever humans can do is part of the definition.