New robots learn things by watching videos.

Do we create puppies that can serve us? 

The biggest question is this. "What does the word "we"? Do the word "we" mean regular people? Does that word mean some company leaders? Or, does that mean military high command? When we say that robots take our jobs, we forget to ask, what kind of jobs do robots take? Are they wanted jobs, well-paid jobs, or what they are? We all know that war is an inhumane thing. 

But still, we have armies. We all know that the mission of high command is to defend the country. Everybody wants to be company leaders and members of military high-commanders. But nobody wants to be a worker. Nobody wants to be private. And that is the problem. The leader sends henchmen to work and then sits on a chair in the office. And that happens in the rain or the sunshine. 

When we say that robots take the jobs of medical doctors we don't think about people who live in remote areas in places like Australia. When happens something the doctor can come there in hours. The robot that has the medical kit can recognize the injury and do the job in minutes. If it can recognize and react to the injury. The same robot can act as a gardener and surgeon and its mission depends on program modules. And that robot can also act as a military operator. 

(ScitechDaily, Robot That Watched Surgical Videos Now Operates With Human-Level Skill)

"By utilizing imitation learning, researchers have enabled a surgical robot to execute complex procedures as proficiently as human surgeons." (ScitechDaily, Robot That Watched Surgical Videos Now Operates With Human-Level Skill)

So, we cannot always do good things. We know that we must defend our countries. And robots are the ultimate tools also for those things. But when we think about robots we face things that the same robots that protect a country can also attack to other countries. 

The robot surgeon performed its operation like a human surgeon. That thing is not a very big surprise. But the very big surprise is how the machine learns those skills. That robot just watched videos. The breakthrough in machine learning is the robot that learns like humans. The thing that made this robot interesting is that it just watched videos. And then that system could make surgeons like humans. 

When we discuss the use of robots in health care those robots can leave room for social interactions between patients and nursing staff. When we think about human-shaped robots that can use videos to learn things, we face one thing. The robot can use any film to learn things. And in some ideas, the military robots can use things like action movies to learn new skills. The human-shaped robot can act as a cleaner, but the same system can act as a robot warrior. This means when we want to create puppies, we can create one of the biggest threats in the world. 

That kind of technology can be a new tool for robot science, which is one of the fastest-advancing technologies in the world. Learning machines can learn things faster than any human can do. Another thing that we must realize is that computers can scale their skills to all systems, that are connected to the network. That means that if one robot looks at the film and copies those skills into itself, it can transfer those skills to all similar robots in the network. And that is the biggest thing in AI-based technology. 

...or, do we create Frankenstein? 

The learning machine that learns from videos can be the revolution in technology in all its sectors from everyday warehouse workers to military drones and robots. The human-shaped robot that can learn like a human is the tool, that can change the world forever. We can say that this kind of system is the modern Frankenstein. If we want to use those robots as weapons it's our business. 

Of course, we can use them for more productive things than killing other people. We can use learning robots for making moon stations, we can use them to close damaged nuclear reactors. We could turn our gardener robots into surgeons that could save our lives in remote areas. But then everything that we want to see is negative things. Otherwise, we see what happens when some other nation attacks on us. The problem is that everybody wants to be generals in the army. Nobody wants to be in a pothole at the front. Robots are one answer to that problem. 

Even if robots cannot make surgical operations they can work as cleaners, warehouse workers, and soldiers. The man-shaped robots can turn all aircraft into drones. The same robots can also operate as infantrymen. And that is the most advanced, promising, and frightening scenario in the world. Learning drones can beat man in air combat. The fact is that the same robots that go to radioactive places and maybe operate for years in the deep sea and outer planets can be the biggest opportunity and threat for humans. 

The learning robots can be the things that travel to Mars and Jupiter a long time before humans. The thing that people must realize when researchers make new robots is that the robot is not human. Skills that robot has are like modules. If we think of things like neural networks or deep learning networks. Robots can also act as workstations in those systems. In that model, all participants of the deep-learning neural network can watch different movies. Then those systems share their data over the network. And that means: that humans cannot respond to that kind of thing. 

https://scitechdaily.com/robot-that-watched-surgical-videos-now-operates-with-human-level-skill/

ChatGPT passes the U.S. medical licensing examination.

 

The biggest difference between machine learning and human is that human uses fuzzy logic. The machine uses precise logic for its actions. When a machine operates, it uses precisely described action chains. The problem is that no pre-programmed action chain has a match for a real-life situation. 

On the streets are some surprises. Like suddenly coming cars are everyday situation. Nobody can predict all possible situations that robots can face in real life. And that makes the system hard to program. 

Precise logic means that programmers must describe every single thing that requires notice or reactions to the machine. The machine cannot operate. If something is missing from its code. 

When the learning system will face that kind of situation it says "unknown situation". And then ask for orders from the human operators. 

The reason. Why ChatGPT passed the U.S. medical examination, but the AI cannot recognize U.S. marines that closed the base because the type of information that AI must handle in the examination is limited. 

In the examination, the AI uses a precise and limited type of information. The AI must just handle letters with precise ASCII values. AI is best in that kinds of missions. 

And when marines closed the target. The AI must handle uncertain types of information. The difference between human and machine learning is that the coders must precisely describe the missions to the AI. 

Sometimes that the person takes a box or some unexpected item like red paper. And cover the head by using that item is enough to cheat the AI. The reason for that is simple. Nobody ever described a person whose head is covered with paper to AI. And that means the AI goes into a trap. 

In the test, marines used trees and boxes. And the AI didn't recognize them. Nobody even describes the case. The two men walk to the target area and keep the box over their heads. Nobody describes. The situation for the AI is that the box with four feet are closing it. 

The AI is the support system. Not the main operator. But the systems advancing. And that kind of test is needed for the R&D work of that system. The same thing that made it easy to cheat the AI is making it hard to make robots that can go shopping for people. 

There are so many variables that making perfect descriptions for every possible situation is hard. Programmers are hard to describe all possible situations, and blocks to the computer. 

But why the AI that mission is to launch an alarm if some unauthorized person is in house works? Sometimes regular alarm systems make alarms. When the burglary alarm launches that system sends information to the guard, who checks the situation from the screen. 

The fact is that the AI uses precise logic. If the alarm system is on, the AI simply knows that nobody should move in some area. And then if something moves in some room, it sends information to the human operators. The AI can detect cases like possible fights in public transportation. And then make an alarm to human operators who make decisions on how to act in that case.

https://www.unilad.com/news/us-marines-defeat-pentagon-ai-robot-cardboard-boxes-175338-20230121

https://scitechdaily.com/the-rise-of-artificial-intelligence-chatgpts-stunning-results-on-the-us-medical-licensing-exam/

https://shorttextsofoldscholars.blogspot.com/

Can computers steal your memories?

Can computers steal your memories?

The computer has not own will

In this text, I will use the words "computer wants" and at the beginning, I must say, that the computer itself doesn't want anything. If the computer would kidnap some people, it makes that thing, because the program inside it would translate it's codes that it must trap somebody, and that means that the computer itself doesn't make mistakes.

Every mistake, what computers make are made by programmers. So if the computer would be programmed to require some kind of identification, while a person would go through some gate, it would close everybody who has not that mark outside. Same way if the computer, what is connected to the robot is programmed to take the fingerprint from every person on the street, the robot would follow the orders and make that thing. Robots ever apply for orders, what they get.

They do exactly, what the programmer would do, without caring about the things like how much trouble the order would cause. And this thing makes computers very effective for things as military actions. In that world, the actor must do exactly what superior would say, and they must not apply anything. But if the opponent would break the algorithm, it can predict every movement, what the opponent will do.

This makes difficult to make programs for autonomous drones

And this is the thing, that makes the creation of the autonomous drone very difficult. When the opponent would learn, how the drone moves, when it's under attack, shooting down of that thing would be quite easy if the opponent would know, what direction the drone would turn.  Have you seen the movie "Tron"? In that movie, the hero would be trapped in the computer memory. But could that thing be possible in real life? Could computer copy our mind or thoughts in its memory, and turn them to part of it? And why it would make that thing?

The thing is that if the high-power radio wave would travel through the human brains, it would start to oscillate with the same amplitude with the electricity of the nervous system if the frequency of the radio wave is the same with the electricity, what travels in the nervous system, and that electromagnetic radiation can be detected by using normal radio receiver. The system can also use highly targeted radio ray, what can go through the pineal gland, and that thing can be used to record the electric phenomenon of the pineal gland. This organ has been connected with abstract thinking and imagination.

What if computers steal our imagination and ability for abstract thinking?

So the computer steals our imagination. And this is why computers would do that thing, and record the electric phenomenon of the nervous system. They need our productivity. The human being is the only organism on our planet, which has the ability to abstract thinking. And productive thinking is the ability to transform the abstraction to the image on the paper, or ability to create the plans and construction orders from the mind. And this is the thing, why computers would need humans.

Even the most intelligent and advanced artificial intelligence is hopeless if it cannot produce new things. And the thing is that artificial intelligence cannot create anything new. Computers are playing chess better than most of the people, but do you know, why chess is used in the tests of artificial intelligence? In the real world, chess has quite a limited gaming area and strict rules. This makes the programming the chess program quite easy if we want to compare that process with the process, where the computer must drive the car in traffic.

Why computer is better chess-player than we are?

The thing that makes artificial intelligence so effective chess-player in the world is the capacity to calculate things very sharply. It can calculate tens or even millions of ways to move the buttons on the chessboard, and it can make gaming models by using those calculations. In this case, the actions of artificial intelligence base the idea, that there are multiple models of the game stored in the memory of the computer, and it would select the best, what is possible in that case.

Why a computer cannot operate that way in everyday life. While the car is driving in traffic, in that environment it is strict and clear orders, but there are many things, which are called "surprising variables". So if we are programming the computer to follow traffic lights we must remember, that some people might have a red or green skirt, and that's why the system is extremely hard to program. And traffic is a really good example of the environment, where the reality is away from stimulations, models, and theories.

But if we would want that robot would make the same things, what we are doing in everyday life, that kind of thing would be too hard to make, because for that thing is needed so many models, that we cannot create them all, and how many programmers would even know them all? What kind of things would a person do, while going to shop? When we would think the case, that computers would want to make some kind of designs, that thing would turn really difficult.

What if computers could benefit the brain waves during the creation process?

So what if we want to make a computer or artificial intelligence, what designs the building, we must understand that the reason, why architects are hired is the please the customer. So how a computer can please its customer? The answer is simple, the brainwaves are telling, does something please the person, who is sitting in the front of the screen, and then the computer can start the creative process.

That process is based on the methodology, where the computer would use simple forms like triangle, square and circle at the beginning of the process. And then the computer would put more and more details in the image. During that process, the system would follow the EEG of the customer, and then it can select the forms and things, that would please the customer. In that case, the chair would have the radio-transmitter behind the head of the person, and the radio receiver would observe the changes of the EEG, what would be moved to that system.

Of course, the system would need data about the formulas and other rules for buildings of that area, but the solution might satisfy every customer on Earth because the EEG would uncover if we like or dislike something. But this kind of thing is the vision of tomorrow. And maybe someday the artificial intelligence would create buildings and act in development missions.

Image:

https://www.houseofbots.com/images/news/2704/cover.png

How to teach robots to follow spoken orders?

How to teach robots to follow spoken orders? 

The idea of machine learning is the same as the people way to learn things, and that's why computer game-style platforms are very interesting things to teach robots. The action or learning process is similar, what is used to teaching dogs. The robot must connect spoken words to some actions, what the programmer can do with it. 

And this makes possible to make robots, which are understanding spoken language. This method is suitable for every kind of robot, and the idea is that the words are connected to recorded actions, so when the programmers are teaching the robot to open the door, they might say "open the door" and then they would transfer to use the virtual workspace, where they can use data gloves and probably Virtual Reality glasses, and make the action by using the virtual movements by using character, what is similar what is seen in the computer games. After the system has been successfully recorded the action, that means that the command "open the door" would activate this kind of movement series. 

The machine learning means that the character, what is the thing that is controlled by a computer would say some words, then the character is moving and acting as the programmer would want. And then those words are connected to the actions, what the programmer has made for the character. 

Moving of the character can be done by using the game platform, and then the virtual character would be used to test the effectiveness of the movements, and then they can transfer to the robot, which will connect those movements and other actions to the words, what the person says. This kind of technology can also be used with combat robots. And those robots can follow the orders, what the commander says without excuse. 

This kind of robot would be very interesting in the many dangerous actions. And thing what makes character dangerous is that some game character would get the physical form. When we are thinking about computer games the programmer, who will make commands to those systems could use the computer games to selecting the most suitable series of movements, and then the commander must just select the most suitable movements and connect the voice command to that series of movements. 

When we are thinking about the actions of robots every kind of action is a series of movements. That means the system would just connect the movements to the command, and that makes that kind of robot very dangerous. The robot itself would not care, what kind of commands it would get, and if the voice recognition or other safety system accepts the command, that allows the operator to give the command "sweep that area", and then the robot will do everything, what includes in that action. So this kind of thing can be used in the Pentagon's terrifying new robot army. 

Image.

https://upload.wikimedia.org/wikipedia/en/5/50/Cylon.jpg

How to teach character in computer games?

How to teach character in computer games?

When we are thinking about machine learning, computer games would be the right place to test and innovate that kind of thing. Computer games are one of the biggest parts of the world of computers, and when we are thinking about artificial intelligence, what would be an effective tool for making those games more realistic, interesting and difficult, we must think the method, what the programmer would use when that person would make the character learning things, what are important in the game. The thing is that the learning of the character would happen by a very effective method, where the movements of the most successful player would be recorded, and then that data would transfer to the automatized characters. That means that the system might copy the master player to the virtual characters.

And then somebody would ask, how the master player would be selected. The answer is simple, in every scenario of the computer game are the character, what is controlled by a human, and the character, what is controlled by the computer against each other. Then the master would be selected by a very easy method. If the character, what is controlled by the computer will win, that means that the computer is master, but if the human player is winning, the character, what is controlled by a human is master. And the code of the game would order the system to use the movements of the master.

This means that the movements, that are recorded from the human users are transferred to the character, which is lost in the battle. When we are thinking about this way to increase the skills of the computer-controlled characters, we might think, that when the player would step in the virtual game board, and start to play, the artificial intelligence would record the things, what the player makes, and then transfers those actions to the characters, what are coming after fallen characters.

That means that the computer-controlled figures, what are coming later would have more complicated movements, than the characters what the gamer faces in the first moment. In this scenario, the gamer would use more complicated movements every time, when that person would be advanced in the game, and the idea is that the gamer would teach the solution, how to fight against the player.

This would make possible to create harder and harder computer games. The machine learning is one of the key elements in computing, and by using this kind of method, computer games would become very difficult. When the gamer would play the game, the movements and actions would be recorded, and transfer to the next characters, until the computer would win. This is one version of using machine learning.

https://upload.wikimedia.org/wikipedia/en/thumb/e/e4/Tetris_DOS_1986.png/250px-Tetris_DOS_1986.png