Seems to me that a level of verbal capability comparable to Gemini or ChatGPT would be sufficient, when combined with the neural net capabilities of Version 12 of the FSD suite -- what is missing, it seems, is the ability to read and react to signs on the road, including those never seen before...