Why would deep NNs or LLMs not work based on our theoretical understanding? What theoretical understanding are you referring to?

Tasks like using a CNN to detect digits has been well understood since the 2000s. The explosion in the capability of LLMs is very surprising, sure, but where is the concrete proof that such systems "should not work"?

> All planners start with executing just the next step.

token != step.

Just you try executing a complex command one word at a time.