i cant only see two things thats changed within the last 20 or so years. More TFLOPs, and some changes to back prop to cope with the nearly flat gradients. thats about it.
we have ASICs now. thats about the limits of current tech as far as brute force. so all we can do now is get larger and larger racks. which seems like a bit of a brick wall in AI. where do they think they can go from here?
is the current processing power enough to continue to create new applications of AI, or is this as good as it gets?
we have ASICs now. thats about the limits of current tech as far as brute force. so all we can do now is get larger and larger racks. which seems like a bit of a brick wall in AI. where do they think they can go from here?
is the current processing power enough to continue to create new applications of AI, or is this as good as it gets?