> The AI also considers each chip as a single artifact, rather than a collection of existing elements that need to be combined. This means that established chip design templates, the ones that no one understands but probably hide inefficiencies, are cast aside.
there should be a word for this process of making components efficiently work together, like 'optimization' for example
This is a strange distinction for the article to point out. If you want to take a more modular approach all you have to do is modify the loss function to account for that. It's entirely arbitrary.
there should be a word for this process of making components efficiently work together, like 'optimization' for example