Given the state of A.I. even today, which is not that advanced, we need one terabyte per day or per week for training purposes. Someone has to produce that data and combine human made data with machine generated data. Synthetic data like that will be fed into ML algorithms.
That simple process, in my calculations, will be the biggest industry on the planet by 2030. The data has to grow 1000x every three years or so. So this industry will end unemployment worldwide, for every person on the planet and also all ages.
As soon as A.I. is used for serious applications like self driving cars, then data have to be audited by humans. Human judgment is irreplaceable. These machines are statistical algorithms anyway, so even with the best data possible, they will err occasionally. So we combine two or three of them and average the result given some score, for even better accuracy.
But data is the foundation of everything in their function. The transformation of data -> information is always done by humans. Given better and better statistical algorithms, that gap will become smaller and smaller, but automation will never be able to produce information by itself.
Other serious applications include medicine and doctors. How about building houses out of epoxy and graphene walls, weighing 10 kg every house and stacking 10000 houses on top of one another? I.e. skyscrapers the size of Everest? Don't we need some serious data and A.I. architecture for the construction to stand tall?
Given the state of A.I. even today, which is not that advanced, we need one terabyte per day or per week for training purposes. Someone has to produce that data and combine human made data with machine generated data. Synthetic data like that will be fed into ML algorithms.
That simple process, in my calculations, will be the biggest industry on the planet by 2030. The data has to grow 1000x every three years or so. So this industry will end unemployment worldwide, for every person on the planet and also all ages.