Their statement wasn’t that AGI is impossible, more that LLMs aren’t AGI despite how much they might emulate intelligence.