From the point of view of a 19th Century farmer, machines have taken over all the jobs in the 21st Century. Jobs that occupied most people in the 19th Century now occupy small amounts of people or have disappeared. Most 21st jobs would be entirely incomprehensible for a 19th Century farmer.

There is a hard ceiling to how much data will ever need to be streamed: retinal resolution at 120 frames per second, compressed, for each eyeball on the planet. The seemingly-exponential growth of streamed data is gonna turn into a sigmoid as we get closer to this ceiling.

Too many studies that apply machine learning to science & medicine employ incorrect methodologies. Many make very basic mistakes, such as not having separate training & test sets, using the test set (not a separate validation set) for feature selection, hyperparameter tuning, etc