LLMs Are a Step But Not a Path to AGI

Large language models are impressive and represent a step, but they are a dead end with respect to achieving AGI without a fundamental breakthrough; scaling up data and compute alone will not get us over the necessary hump.