Self-aware robots are indeed dangerous, but they may also push humans to make progress again.
Editor’s note: This article is from the WeChat public account “Big Data Digest” (ID: BigDataDigest), author Niu Yan Yang.
Since ancient times, mankind has hoped to create a human-like machine to do all kinds of work instead of human beings. With such an original intention, humans began a long exploration of robots.
In fact, people have been imagining and pursuing robots for more than 3,000 years. During the Western Zhou Dynasty, China’s capable craftsman, Yanshi, developed a singer who can sing and dance. This is the earliest recorded robot in China. In the late Spring and Autumn Period, Luban also built a wooden bird. According to the “Mojing”, it can fly in the air “three days.”
Until 1959, the world’s first industrial robot was born, invented by Joseph Engelberg, who is known as the “father of robots”.
In ancient and modern China and foreign countries, human beings have always been committed to the study of robots. From the industrial robots used to carry materials, parts and tools, to the intelligent robots with perception, thinking and effects, people are expecting more and more robots. High, research is getting deeper and deeper.
The latest video from Boston Dynamics shows that robots can smoothly roll, jump and other actions like athletes. Humans have made them more and more like human beings. So, in the future, can they also have consciousness like human beings?
Is there a self-aware robot like the Terminator T-800 in reality?
“Terminator” is a famous American science fiction film series. Recently, “Terminator: Dark Destiny” was launched on the line. In the film, Arnold SchwarzeneggerThe T-800, the finalist who played the role, was deeply shocked to me. “He”…has a sense of consciousness, he can produce empathy and can learn by himself.
Ano’s Terminator T-800 has turned into a loyal guardian from the villains who used to kill the protagonist. This unscrupulous robot has human feelings in dealing with John, in “life”. At the last moment of the deadline, the key person in the eyes of the lock is still John, who is hard to defend after his heart regrets. It is very moving.
Some readers are thinking that smart robots in the real world will also generate awareness in the future? If they have consciousness, then people have to think in extreme directions…
Hod Lipson, director of the Creative Machines Laboratory at Columbia University and professor of engineering, has made it clear that robots can be aware of the future.
Hod and his team have been working to create robots that are self-aware, self-replicating, and creative, and that the public is full of curiosity about the future of artificial intelligence.
As early as 2009, he boldly put forward the idea of thinking about robots on TED, which triggered a heated discussion in society. “I don’t know if we can achieve this goal in my lifetime, but we are indeed on the road to this goal,” Hod Lipson said.
On March 20 this year, Hod Lipson and his team developed a self-modeling bionic group “particle robot” to achieve a breakthrough in the migration of robots, and appeared on the cover of “Nature” magazine.
This kind of “particle robot” is no longer like the traditional robot, the overall operation is subject to some components, and it is close to the “Grey Goo” in the group, from the nano robot in the doomsday plot.
In 2004, K.Eric Drexler, a pioneer in nanotechnology, presented it in the book Engines of Creation.In the book, such robots are out of control due to their ability to self-replicate, eventually depleting the resources of the entire planet, causing the extinction of species.
At the moment, “Grey Goo” is still a crazy idea.
At the “2019 Tencent Science WE Conference” on Sunday, Hod Lipson shared his team’s ongoing research.
The following content is partly derived from Hod Lipson’s speech at the meeting.
The biggest difference between humans and robots is that we have awareness and know how to simulate the future
Does robots have feelings and consciousness, can they have emotions and thoughts? What is the nature of self-awareness and humanity? Hod said that we don’t know the specific meaning of these “fuzzy” terms.
It has always been said that when a concept is taught, it is truly understood, but Hod believes that it is only true understanding when it is built.
“Self-awareness is the ability to model yourself into future scenarios, and to what extent you can anticipate your future, the degree of self-awareness,” Hod said.
Awareness is hierarchical, and you can feel how far the future is, how true it is, and the time frame. For example, if a dog sees its next meal, he will happily wag his tail, and people can think of how they look after retirement. Self-awareness is not only predicting what will happen in the future, but also being able to make plans for the future based on predictions.
When you see this picture, do you think you can cross the pond? You will imagine yourself stepping forward, where the first step and the second step are, which means that we humans can simulate the decisions and actions in this situation, and you can determine if you can cross the pond.
Be able to meet yourself in the simulation situation in advance. This is how we learn and do things, without having to do it or watch our mistakes. This is the biggest difference between us and robots.
The critics may say usThis kind of simulator has always been like this. At first, engineers built such a simulation environment for robots, designing, programming, and testing them all in the simulation.
But human learning is not the same as machine learning. The simulation of machine learning is created by engineers. It takes a lot of time to study what is being simulated, what is important, and which is not.
Robots can also have the ability to self-simulate
Our human beings are born with the ability to simulate, and the play at the baby stage allows us to understand how muscles are controlled and even understand the physical world we are in.
Now, artificial intelligence has made great progress and can do better. Recently, the Hod team has developed a self-learning robot. We can see that it has only one “arm”. It doesn’t look like we have eyes, and the developer doesn’t tell it what it is, but let it Feel your own feelings and learn in your own imagination.
In the beginning, it squats like a baby, licking his arms, sensing his own form, and perceiving the world. After about a day, its self-image seems to have arisen, may not be perfect, but enough to do some simple tasks.
In fact, creating a self-image is not easy, just as we ourselves can’t tell how we think about ourselves.
This simple task can be achieved by programming, but the robot learned this task in its own simulation, which is an important step towards building a self-aware robot.
Blue: Self-modeling robot; Red: Conventional robot
Contrast data shows that robots capable of self-modeling learn much faster than conventional robots because they can simulate themselves without having to operate them in reality, just like we can think about the consequences of things in the brain. And then eliminate the low-implementation method, which undoubtedly saves a lot of time. This is one of the reasons why the Hod team has always wanted to make robots with self-awareness.
Hod said in an interview that he is excited about the unknown. He and the team believe that the benefits of self-aware robots far outweigh the risks, imagine our drones, driverless cars, etc. when they own Self-awareness, self-care and repair from destruction, continuous self-improvement, will save people a lot of time and effort.
“There are more and more artificial intelligence systems we build, and humans may not be able to continue to maintain these systems or program them in the future. These tasks are too big, and we need to share these tasks with the robots. They can see themselves and have their own thoughts, self-adjusting, adapting, and repairing. This is a very exciting time.”
Finally, let us recall REV-9 in Terminator: Dark Destiny, a combination of “superalloy skeleton and liquid metal” that can repair itself intact no matter how it is attacked. There will be no damage. The most deadly thing is that it can be split into two individuals and act alone.
The imagination of science fiction films is what we need to learn, but it seems to also warn us that human beings are destroying themselves in the distant future.
In our lifetime, we may not see the real intelligent robots with self-awareness and self-learning ability in reality, but technology never stops, we must think about the good side.