Hacker News Reader

I don’t think throwing more computational power at the problem is the right answer to all ML problems.

Those neurons are arranged and incentivized cleverly. The structure is also very important and is necessary for the resulting intelligence.

So it's not only computational power, but also the unique structure nature found through trial and error.

I wonder how much of that cleverness will be gleaned and appropriated by those who design 3 dimensional chips.