I agree. I think the emergence of LLMs have shown that AGI really has no teeth. I think for decades the Turing test was viewed as the gold standard, but it's clear that there doesn't appear to be any good metric.