"The question really becomes, why did symbolic ai research die when it's so effective at specialization?" I think we want "AI" that scales faster than humans c

"The question really becomes, why did symbolic ai research die when it's so effective at specialization?" I think we want "AI" that scales faster than humans can implement. Game "AI" looks nothing like machine learning and looks a lot more like some programmer just smashing out some code and trying it out for a while. Some autotuning can be done, of course, but it's largely a program being written. Who has time for that?