AI is a tool that requires understanding.

 AI is a tool that requires understanding. 

If we use AI, we must understand what we ask. And we must realize that even if the AI seems perfect. There is the possibility that there are errors in the source. Also, questions that we introduce to AI can involve things that make the AI search for wrong solutions. 

Another thing is that the AI can remove answers that might not please its users. In the case. Where the AI's purpose is to please its users. It might be useless to search for relevant and neutral information. Information collected from some sources serves somebody. And the minimum way to serve somebody is to please that actor. 

The AI means multiple applications in military and civil areas. The system can make decisions by following some route that seems reasonable. Or the system can take a series of actions that can benefit it and its users. 

The term Intelligence means a series of actions that seem functional and logical. Artificial intelligence is the thing that can network multiple applications together. The thing in the AI is that is not neutral. Or the AI itself can be as neutral as the people who programmed it are. 

 Humans made all the sources and algorithms that AI uses. Human programmers select what kind of answers the AI gives. 

And the problem with commercial AI is that its mission is to please people. Commercial AI has a mission to bring profits for its owners. And that means the commercial AI might give data that pleases people. Another thing is that the AI would give moralistic and ethically accepted answers to people who ask questions. The problem is that moral and ethics is not always true. The thing is that the language models just put the words one after one. 

The answer to the question: can AI cheat people to make things that do not benefit them? Or rather saying, the AI is not a cheater. The people who made the AI are cheaters.  Normally, the AI doesn't know if people like them or their answers. The AI uses certain algorithms like how long time each user spends with it. But the AI doesn't know: do people like it? 

The AI can search for things like what normally pleases people. And then it can send material to the net that pleases most of certain people types. Thins like TikTok are popular social media platforms. The problem is that those platforms operated from China. The AI can search for certain things from the rooms where people made videos. By using this data intelligence officials can select people, who they want to recruit. The AI can search for things that might interest Chinese intelligence. 

Can AI be dangerous for people? The fact is that things like misuse of AI are always dangerous, and other dangerous things are programming errors. Military solutions or militarized AI are also risks. The purpose of those solutions is to harm people or infrastructure. If we want to use AI as a weapon. It will become a weapon. If we deliver the free will to AI and follow everything that it says, we are under AI's command. 

If we sit behind the steering wheel of a self-driving car and sleep that thing can cause an accident. The other thing is this. If hackers put something like a ghost protocol into the car's control system. The car can turn into a dangerous thing. In the worst case, the ghost protocol can connect to the image recognition system. And if somebody is targeted for that system, the car can smash that person. 

The thing that makes this possible. AI doesn't understand that action.  It just follows the algorithms. That algorithm orders it to push the brake pedal at a certain moment. The problem is that it's very easy to crosslink the gas and brake pedal. That thing must realized when people think of things like self-driving cars. Errors make us humans. And in places like traffic is no place for errors or the destruction will be total.