Regarding moonshot projects, I expect those to exist in light of a potential breakthrough in technique. Kind of connecting point 3 and 1 that you made.
I predict that we will see a breakthrough like we saw on transformers in 5 years from now, due to the new interest, capital (financial and human) being dedicated to this cause.
I think the best OpenAI can do is to make their product a cash cow, by reducing cost and focusing on moonshot breakthroughs to stay ahead of the curve.
Keeping in mind that it was almost 3 decades between CNNs and transformer techniques. CNNs came out in 1989 while transformers in 2017. I expect this kind of window to dramatically shorten with the renewed interest in the field.
Regarding moonshot projects, I expect those to exist in light of a potential breakthrough in technique. Kind of connecting point 3 and 1 that you made.
I predict that we will see a breakthrough like we saw on transformers in 5 years from now, due to the new interest, capital (financial and human) being dedicated to this cause.
I think the best OpenAI can do is to make their product a cash cow, by reducing cost and focusing on moonshot breakthroughs to stay ahead of the curve.
Keeping in mind that it was almost 3 decades between CNNs and transformer techniques. CNNs came out in 1989 while transformers in 2017. I expect this kind of window to dramatically shorten with the renewed interest in the field.