I'd like to start by saying that I do not consider myself well-educated in the field of AI, I think I'm far from that, I can handle AI only as a concept, not in details.
My problem is the following: the cycle of building an AI system looks like this: you gather data, you train your model, you get a system with certain capabilities, which will be improved with the next iteration of training and so on.
People usually say it takes 10k hours of practice to get yourself to an expert level at anything. There's the investment of an employee, putting in the time of his/her life to become an efficient workforce.
So let's say companies put cameras at the worksites, in the offices, start recording computer screens, mouse/keyboard activity (if they haven't already) etc. and start training models that extract every bit of added value that comes from the employees' skillsets by training AIs with the recorded data. Doing this they will be able to (partially or fully) replace the people with cheaper automation. They will acquire techniques that are not written down in any book, that are not part of any university course.
Shouldn' t the people be compensated for their added value?
Maybe I'm not thinking straight, but I cannot rule out this scenario, and I do not necessarily believe that a well balanced redistribution of wealth will be the result of a large scale work automation process.
[link] [comments]