I don't know why everyone assumes RSI will be quadratic or exponential growth.

It might well be sublinear, but just faster than humans.

Most problems in science face severe diminishing returns.