Even though the ‘attacks’ of the aliens and the zombies were very famous things to talk about the last few years, we all know they are not going to happen. But amongst all villains in the movies, there is the one that we can definitely fear- the sentient robots. Even if we are not sure when, but the arrival of the sentient robots is certain. Having a world where conscious machines exist is something we need to think about, a lot! We have to question even our existence.
Recently, a research of artificial intelligence is having a revolution. Humans can be surpassed in games like chess and Go by the AI systems. The systems can easily recognize faces and even drive safer than we do. However, according to many researchers, conscious machines that have self-awareness and can catch or have feelings are far away in the future. A generalized intelligence should be created first, a machine with features and capacities to learn and to feel. That is when the AI can get to the point of having a conscious machine.
Other researchers believe that the era of these machines is very near.
Justin Hart is a scientist at the University of Texas who stated that many people have expectations for the AI, where the self-awareness will be the ‘last game’. He and his colleagues are creating machines that have basic minds. They created a robot that moves like an infant to learn parts of its body. The other robot they’ve created, talks continuously about things it looks at, and if you hit it, it will cry.
It cannot be said that robots have wide experience or that they are as proud as the humans are. Still, they can perform many things and conditions which are almost identical to the human mind. One of these qualities is the empathy.
Beside the cool side of the creation of these machines, the scientists are designing them due to their ability to fix many issues in the learning systems of the machines. They can be very powerful, but still cannot be seen through. The relation with the input to the output is the only way these machine work by. The systems of the AI don’t go deeper with the logic, they only memorize these connections.
Speaking of humans, we can definitely say that their thoughts are very hard to read. We, the humans, constantly observe ourselves and put our thoughts in different perspectives. That basically represents the main purpose of human’s conscious minds. On the other hand, robots are very easy to read. You can easily approach their actions and what and how they did an action.
Selmer Bringsjord states that a very good result will be delivered if they could catch some parts of the conscious structures of the humans. Selmer Brinsjord is a researcher at the Rensselaer Polytechnic Institute in New York. Even though we are somehow concerned and feel threatened by the sentient robots, maybe we can look at their creation like something that will be of a big help to us.
Know yourself, robot
The AI systems are developed the most in the self-driving cars. These cars know where they should turn left or right and when to stop. They are making these actions based on the laser readings and radars that are constant. Later they translate them into algorithms. But the rest of the driving is supposed to be done by an actual driver, because a consciousness is required.
The neuroscientist Paul Verchure says that the self-reading cars have to know what the other car tends to do.
To demonstrate this action, two scientists created a robot with characteristics of a crazy driver. The robot is very small and moves on the algorithm using its logic. They set another robot that followed the first one wherever it started, so its path is not clear. That way it had to clear up its path and define the moves of the second robot, by using logic.
The two scientists Hod Lipson and Kyung-Joong Kim also produced a strategy which is intercepted. They followed Darwinian evolution. They fundamentally created a not so perfect duplicate to the actor’s brain. Even though the duplicate wasn’t perfect, it was good enough to show off its next actions, states Lipson.
The team, moreover, developed another robot that could understand its own body. This robot has four legs and is of the size of a tarantula. When you switch it on, the robot is unconscious about its features and the parts of its body, but it capable of learning. When it sees different moves, it learns how to perform them. Just like a newborn baby.
It needs only four days of observation to learn that it has four legs, and realizes how to collide with them so it can start moving. When the scientist switches off one of its motors, the four legged robot immediately knows that now it only has three legs and its moves are not so effective.
Lipson says that the self-awareness of this robot is very low.
Initiative is another human characteristic that the researchers want to add to the AI. They use the game Go. The machines take part in solving problems, which is the hardest function.
The neuroscientist Ryota Kanai had a discussion for the journal “Trends in Cognitive Science” about his startup Araya and how he tries to develop an inborn motivation in the machines. He demonstrated this action in a virtual landscape where a car has to climb a hill. The agent starts to move the moment it’s given a command.
Then, after the command was given, the agent starts to move and solves the problem of how to pass through the hill, without having any given instructions.
Kanai states that they did not give any instructions about the environment to the agent. The agent itself makes all the exploration of the terrain and intuitively chooses the right path.
What you have to do is motivate the robots not to quit, but observe and move forward. Joscha Bach, who is a Harvard’s researcher of AI, says that these machines are stubborn just like the humans. When given poisoned mushrooms to the virtual robots they decided to eat them instead of avoiding them.
In order to develop self-awareness and motivation, the most important function that needs to be created is the feature of paying attention or focusing. Selectiveness and attention was important in AI research and also for the creator of GO, Google DeepMind.
The computer science professor, Stanley Franklin describes the consciousness as an attention filter. Franklin and his colleagues are building an AI system that will be able to know what thing to concentrate on the most. They called this AI system LIDA.
Its first version was created to work as a server for the Navy of the U.S. It was mainly reading e-mails, and was focused on the job specifics and bureaucracy of the government.
Ever since then, the team was using this system to develop minds of animals, mostly behavioral ones. For instance, LIDA is very likely to react the same way as humans do when they have their attention caught by someone else. That delivers humanlike responses to same situations by LIDA.
Many other researchers have built similar models with similar characteristics. One of them is Pentti Haikonen’s robot called XCR-1. Haikonen says that his creation has basic emotions.
The system of this robot works by learning how to recognize different relations, just like human’s brains do. Haikonen interprets exercises on his robot by showing it an object and telling the color of it. The robot learns how to recognize it. The auditory and the vision modules of the robot then collide.
Haikonen explains that the reactions of the robot are the same as the ones of the humans in certain situations.
The things can get interesting if you show a certain green object to the robot and tell it is blue. If the auditory module is more dominant, then the robot will only accept the information that it heard.
Heikonen also added some features like ‘pain’. He demonstrated this capability by tapping the robot. The robot then shouted “Me hurt.”
According to Haikonen, some people may feel bad about this, but he personally is indifferent.
A New Type
Researchers are continuously developing ideas and concepts about the creation and ‘birth’ of the new species with humanlike characteristics. The enormous process will give us the time to adapt and accept the fact that one day we are going to be joined by this new species which will be advanced just like us.
Susan Schneider focused her studies on the implications of AI. She explains that the reason why she and her colleagues see the non-human animals the same way as the humans is that they recognize the consciousness in them. She raises the question of how we will treat the conscious machines once they come. Making them work for us will be a sign of slavery.
These machines will react to different situations and objects the same way as humans.
Kurt Gray and Daniel Wenger have even exposed their desire to see the robots creating their own religion as they will have many social interactions.
Lipson states that these robots won’t be heartless, but instead, they will most likely have poetry and music as parts of their lives, which won’t be understandable to us.
Source > futurism.com