Many people are thinking incorrectly about AI.
They think new jobs in future will involve just humans using AI to constantly make something.
That is mostly incorrect.
AI is good because AI produces songs faster than anyone else, movies faster than anyone else, books faster than anyone else, research papers faster than anyone else, arguments faster than anyone else.
So their conclusion is that AI will produce a lot, and humans will just control the quality, keep what is good product and throw away what is bad product.
And right now, that is true. Right now, when using AI, you produce as much as possible, then keep what is good. That is the basic strategy and the only possible strategy right now, along remembering and saving best prompts.
However, in future, AI will have ability to control own quality.
In other words, one AI will constantly produce lots of stuff, while other AI will serve as quality control to only keep the good.
Right now, AI has no good sense of what is good. So humans are needed as workers to control quality of what is produced and to keep giving commands to produce more.
This is called quantity to quality. AI produces great quantity of different things, and humans select what is best.
But future AI will be built in a way to know what is best. In other words, it will produce great quantity, and then AI will select what is best.
This new AI doesnt need humans after start. It just keeps producing, and separating good from bad.
But this new AI only requires 1 thing: knowledge of what is good. So in this case, human would have to correctly instruct AI of what is "good", and AI would just keep producing and selecting the good among produced.
In this sense, AI would compare outputs to see which output matches "good" more.
This new AI would be far more powerful, because AI is faster at choosing good than human is, so this AI would produce more than AI which needs human to control quality.
