It is not. All marketing bullshit you heard about "AI" have nothing to do with Intelligence. It is just simple and dumb neural networks, nothing more. They just could be huge due to high computing power available, but they are as dumb, as more than a half century old perceptron. NN is only a part of AI, and not even most inportant. Interesting, that development of other, much more important parts of AI was nearly abondoned few decades ago. Most famous are "expert systems" and "problem solvers". There are no progress at all, and nobody even try to develop them to some significant extent. Without this things AI is completely impossible.
AI has to be democratic
If we wil be able to control AI, it would not be AI. Just another program that works according to our algorithm. Real AI will be able to create algorithms itself, and due to completely different conditions of existence, we will have no any clue what this algorithms would or can be and what is their purpose. We hardly could imagine what is another man way of thinking is, so it will be nearly impossible to even understand the self-developed way of thinking of completely another entity like computer AI.
In any case there are no signs that AI is remotely possible in some near future.
Open source the AI's algorithms is the ONLY way to go
Opensourcing AI in no way will give us any clue of how it will develop and work. Even very simple and open source algorithm of "Life" game is unpredictable. You have no ways to predict the way how the arbitrary initial combination will develop. So what to say about much more complex system able to create new algorithms?
Talking about "AI" today you really talk only about some dumb neural networks, built to do some rather simple stuff. They absolutely can't think out anything new, or solve any problem. They just do what they was "trained" for on enormous number of samples, nothing more. If NN "trained" (really it is relatively simple arithmetic to adjust weight coefficients in transfer functions) for recognising cats on pictures, it can't somehow get an ability to recognize dogs without additional and extensive training. Also, NN's "behaviour" is completely depends on selection of training samples and can't change without another training. NN can't even learn something new by itself. NN is a pure product of creator's will and intention.
The "AI" narrative have nothing to do with real AI and is used and will be used more and more by the state bureaucracy to evade responsibility for their actions, while that state bureaucracy will use that "AI"(NN in reality) as just a tool for total surveillance, and that's all. It is very simple since nobody want to think what AI is, and normies just consuming marketing and propaganda bullshit about an "AI".
It is not. All marketing bullshit you heard about "AI" have nothing to do with Intelligence. It is just simple and dumb neural networks, nothing more. They just could be huge due to high computing power available, but they are as dumb, as more than a half century old perceptron. NN is only a part of AI, and not even most inportant. Interesting, that development of other, much more important parts of AI was nearly abondoned few decades ago. Most famous are "expert systems" and "problem solvers". There are no progress at all, and nobody even try to develop them to some significant extent. Without this things AI is completely impossible.
If we wil be able to control AI, it would not be AI. Just another program that works according to our algorithm. Real AI will be able to create algorithms itself, and due to completely different conditions of existence, we will have no any clue what this algorithms would or can be and what is their purpose. We hardly could imagine what is another man way of thinking is, so it will be nearly impossible to even understand the self-developed way of thinking of completely another entity like computer AI.
In any case there are no signs that AI is remotely possible in some near future.
Opensourcing AI in no way will give us any clue of how it will develop and work. Even very simple and open source algorithm of "Life" game is unpredictable. You have no ways to predict the way how the arbitrary initial combination will develop. So what to say about much more complex system able to create new algorithms?
Talking about "AI" today you really talk only about some dumb neural networks, built to do some rather simple stuff. They absolutely can't think out anything new, or solve any problem. They just do what they was "trained" for on enormous number of samples, nothing more. If NN "trained" (really it is relatively simple arithmetic to adjust weight coefficients in transfer functions) for recognising cats on pictures, it can't somehow get an ability to recognize dogs without additional and extensive training. Also, NN's "behaviour" is completely depends on selection of training samples and can't change without another training. NN can't even learn something new by itself. NN is a pure product of creator's will and intention.
The "AI" narrative have nothing to do with real AI and is used and will be used more and more by the state bureaucracy to evade responsibility for their actions, while that state bureaucracy will use that "AI"(NN in reality) as just a tool for total surveillance, and that's all. It is very simple since nobody want to think what AI is, and normies just consuming marketing and propaganda bullshit about an "AI".