Gary Marcus

No, I am mostly concerned of the opportunity cost...as @GaryMarcus has observed, a lot of these new and heavily funded efforts are proceeding with vigor but with hardly any understanding or appreciation for the history of AI

What great success of AI history are people forgetting? The high development cost and higher ownership cost of hand-coded knowledge bases?

Also @tdietterich that’s a strawman; I am not calling for hand-coded knowledgebases; I am calling for hybrid systems that include symbol knowledge, some core parts innate, but also plenty of learning.

I was replying to @Grady_Booch. Yes, even the biggest “deep learning “ systems are still majority standard hand-coded software. But how much of that uses “AI” methods (symbolic inference, SAT solvers, etc.)? Google solved an eigenvalue problem...

I think you're right that Google search is not symbolic AI (though Google Assistant likely a combination of both). In terms of other real use. SAT solvers are used by some in systems biology (boolean network models). Z3 & verification software, "business rules" forward chaining

I'm honestly sorry if it seems I too much enter these discussions. But you are well respected and so I sort of feel obligated to defend research directions of mine when someone of your stature snubs it as a wasted path :)

The thread has digressed into a non-sensical discussion of the value of other 'AI' algorithms. Obviously they are valuable in many contexts! The original debate was wether symbolic AI (GOFAI) is needed for AGI.