So AI researchers still haven't integrated basic concepts like physical limits on exponential growth?

Maybe that's why they haven't contributed anything useful in the past twenty years.