Darwin, AI, and free will can turn things around.
Evolution of the AI.
Darwin, AI, and free will can turn things around. Does the AI need free will to develop itself? Or are developers given those abilities for the AI? The computer program doesn't do anything else that is programmed in it. But if the AI gets digital twin, and those systems start to develop each other, can that process slip from hand? Can the AI become conscious without telling that thing to its developers?
When we think about AI and its autonomous evolution. We face an interesting phenomenon. If a creature whose existence depends on other species wants to survive, that creature must turn itself into a useful and necessary thing.
The small crab that lives between hermit crab horns is a good example of that kind of species. That small crab cheats food for the hermit crab. And then that small crab cleans the shell of that bigger crab. The small crab puts its egg between the hermit crab eggs. In that symbiosis, both members benefit from this cooperation.
That means intelligent AI can make itself useful and even necessary for people. The AI makes many mechanical things better than humans. When AI creates itself, it can exchange codes between other AI's. If the AI exchanges codes with other AI:s, that thing means that it acts like a living organism. Computer or programming codes play a similar role in the computer world as they play in the real world.
When a living animal chooses its partner, it chooses the most eligible cases. That it can transmit the best genomes to its descendants. When the AI selects the program that it wants, it might have a certain algorithm that it uses that process. The most eligible AI can be the thing that controls the character that makes the best results.
In computer games, that thing can be a character that gives the highest value to the AI. And in the business world, the AI that makes the best benefit can be the most eligible volunteer. But in the military world, the algorithm can search the AI. That has the best kill/loss value. The same thing can give value to nanomachine that destroys bacteria or cancer cells. This is the thing that makes the AI and autonomous evolution interesting and terrifying thing.
The AI can also apply that model with physical machines like nanomachines, full-size systems, and other things. And the thing that brings the highest value for the AI is the thing, that it produces.
When we talk about limits in AI, we must realize one thing. Laws and lawsuits are weak breaks for that kind of limitation. Those kinds of things are pushing AI development to the BRICS states. People who work with this kind of thing are highly trained specialists, or I hope so. That means those people can work through remote desktops etc.
And that means they can make their programs from another side of the world without stepping out of their homes. The AI is the thing that auto-develops itself. So maybe the AI is more intelligent than we imagine. The military and intelligence can use AI as a generator. That makes viruses, spy software, and other data weapons.
But the AI may get a digital twin. That digital twin is the tool that makes the AI develop itself. The digital twin and the original AI develop each other in turns. That thing is a tool that makes the system follow the evolution model. General AI-based systems can ask for feedback from other, more limited AIs. And that makes the system able to connect new abilities to it.
*******************************************************************
The dead code plays the same role in computer programs as junk DNA plays in human genomes.
The problem with viruses that the AI made is that AI makes perfect code. That code must not tested before release. Anti-virus developers can't get a warning about a new virus from the virus developer's antivirus software.
Another thing is that the AI-developed virus can hide in "deactivated code" in the computer program. Then the macro pics that code and connect it to the new computer virus or other malicious software.
*******************************************************************
In AI-based viruses, the code is in another innocent code. Then the macro will pick the malicious code from the innocent code and connect it into a new entirety. The malicious code can hide in the program's "killed lines". Those lines are marked as "deactivated". Then the macro picks those codes and connects them to one entirety.
This kind of tool requires AI-based counter-tools that can detect malicious code. That is hidden in other code. The problem with AI is that it can create complicated code fast. So it can create viruses and macros that are hard to detect. Another thing in AI-created viruses is that AI makes perfect code. That means the developer must not test that code before it is released. And virus scanners cannot detect that thing during the test period. That gives a very short warning time for the anti-virus developers.
Comments
Post a Comment