Pedro Domingos

Exactly. AI will require connections and processing we are not even considering, amongst most likely other things we don't know yet.

Human A.I. requires continuous data from all senses to produce the consciousness you mention. It needs to learn from mistakes, and to what a mistake is through supervision, by example. Such A.I. will eventually be a disappointment to parents, if succeeded…

Had a bunch of questions on this. Open DMs ?

How do you teach an AI gut feel?

I think emergence is key. And I believe the best example of emergence is an attractor state, where the computation of multiple elements produces a single, sequential, emergent state. Look at head direction cells for example.

Biologists don't try to solve biology. They study it. The framing of AI as something to solve is misguided. There are plenty of problems that can be solved using the methods of AI. That's a different thing.

> conscious thinking is sequential by which definition of consciousness? it is hard to say that anything is sequential if we don't still have the definition of it. In any case it sounds silly to me that evolution put a centralised bottleneck on a highly parallel system ...

The simultaneous parallel and sequential modeling reminds me of system verilog used to model parallel hardware and a sequential test bench - with virtual time.