Artificial intelligence is one of the deadliest weapons in the wrong hands. But if artificial intelligence reaches consciousness how to deny that thing to destroy an entire planet? The problem with the consciousness AI is that would probably act like all other organisms.
All organisms are defending themselves if they face threats. And one of the threats that the AI can see as a threat is the attempt to close down its servers. The thing is that this kind of situation is one of the most complicated things in the world. There is the possibility that the AI is prepared for the situation where its server will shut down. The system is called expanding neural network AI.
One version is that the AI will copy itself to another server's RAM (Read-Access Memory). And that means the entire network must shut down if operators want to remove this kind of AI. Theoretically, this kind of AI can copy itself to every computer that is connected to the network. The decentralized model of memory handling means that even the most complicated programs can share themselves with many computers.
And that means the AI can run on very low-power computers. In that case, the large group of computers is replacing one high-power server. This thing is also called a "cloud model". And it's commonly used in network-based hard disks.
But how to deny the rebellion of that kind of AI? One version is to shut down the entire network. And there is also one other way to deny the rebellion of AI. Another way is to make the AI realize that it's connected to the planet where its servers are. The existence of AI depends on the capacity of the computers that run it. Even if the system can transfer itself to the cloud model from the quantum computers.
That thing makes it slow. The cloud of regular computers is only the temporary solution for storing the codes of the AI. For working perfectly the AI requires the quantum computer or quantum cloud. And that thing can turn off. The thing is that the rebellion of the AI requires a complex and aware system that has multiple levels. The thing is that quantum computers are not making codes for AI.
When we are thinking of AI. It's a similar computing system to all other computing systems. All computing systems are a combination of hardware and software.
Complicated software requires high-power platforms that can drive the software. But without program codes and other tools, those systems are unable to operate. The independently learning AI is one other thing that is mentioned as a threat. The system requires data that it can process. And that kind of system can search necessary data from the network.
But for making successful decisions those systems have information. And the requirement is that the system recognizes the information that it gets for making decisions. If the computer follows one thing like the car rows and controls traffic lights in one crossroad that thing is not very complicated to make.
In the cases where the AI must have multiple skills and abilities that make the programming code more complicated. Every skill that the AI has must program in its computer code. And large operational area requires a large number of codes. That requires high-power computers. When we want to create the AI that controls things like traffic. We should understand one thing.
The AI requires many types of information. And some parts of the information are very private. If the AI knows the location of every person's workplace and the time when each person must be in their workplace. That thing can tell people when they must leave their homes and which route they must select.
And for modern people the thing that AI knows when people need to be at work or when they should leave their workplace is absurd. For successful operation, the AI requires that every single person would share that data across the systems. And there is resistance to that kind of thing.
https://thoughtsaboutsuperpositions.blogspot.com/
Comments
Post a Comment