I wouldn't even mind if the training was placed in some sort of open dataset. Now Google is essentially appropriating some of your brain power to improve their proprietary AI.

I really wish we could go back to proofreading words that tripped up OCR, for a digitized corpus of books open to anyone.