Since economics will drive much of the future direction of AI, I think we'll begin to see more domain-specific vertical applications. There is no one size fits all and different domains require deep understanding to apply tools in the right way. Nvidia doesn't say much about their internal processes, but it's an ideal example where the product has evolved to the point that only way to develop the product is through layers of specialized tooling and I suspect they're using forms of AI that don't know about. It's use of computers to augment the humas because human can't do what needs to be done. Same for drug discovery and other fields. I can see where the real growth will be in the fields where augmentation applications do stuff humans have no ability to do.
I really enjoyed reading this. It's refreshing to hear an alternative view that is (imho) realistic and sensible and way more likely to come to fruition in the future.
The amusing thing is that the talk is about AGI but what is sold is an augmenter.
Economics and personal choices will drive things, not manifestos.
I agree on the conclusion Andy.
Understandably money follow ideologies driven by economic drivers (decrease headcount).
The author’s insights about the nature of regulated industries will drive a new wave of architectures with HCI front and center by default.
Since economics will drive much of the future direction of AI, I think we'll begin to see more domain-specific vertical applications. There is no one size fits all and different domains require deep understanding to apply tools in the right way. Nvidia doesn't say much about their internal processes, but it's an ideal example where the product has evolved to the point that only way to develop the product is through layers of specialized tooling and I suspect they're using forms of AI that don't know about. It's use of computers to augment the humas because human can't do what needs to be done. Same for drug discovery and other fields. I can see where the real growth will be in the fields where augmentation applications do stuff humans have no ability to do.
I really enjoyed reading this. It's refreshing to hear an alternative view that is (imho) realistic and sensible and way more likely to come to fruition in the future.