News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
AI mimics the human brain is not that simple
AI mimics the human brain is not that simple
Some people think that man is designed according to the image of god. When it comes to real artificial intelligence, which might be our greatest invention, we have tried to do the same. A typical way of artificial intelligence is to reproduce the human brain in digital form. But top scientists say the inspiration will come from somewhere else. In fact, trying to imitate the human brain perfectly is a waste of time.
According to the team's artificial intelligence researchers, this is a tough problem. We can't simulate the human brain perfectly. Instead, we should spend time on unlocking the basic principles of intelligence.
Max Tegmark is a physicist at the Massachusetts Institute of Technology and director of the Life Institute for the future. "Too much attention to the brain is only carbon chauvinism," he said. Although scientists have yet to find the mystery, there is nothing magical about the way the brain works. "We're so obsessed with how the brain works," says Tegmark. "I think it's a lack of imagination."
History has proved his point of view. In Vitoria, a man named Cl ment Ader engineers built the first heavier than air aircraft. He mimicked the bat to build it. These machines are only chairs with large bat wings on both sides. Ader flew hundreds of meters with this almost uncontrollable device. But if he is the first person to fly successfully, why do all the people know that the Wright brothers do not know him?
Ader third version of the aircraft. Although it can keep flying, its steam engine is completely out of control.
Although the creation of artificial intelligence in our own image is not a viable approach, the group's discussions that night are still returning to biology. As Levin said, "human intelligence and consciousness are still our best examples.".
"You can get inspiration from biology, but you must never copy it mechanically," says Jan Lechun, director of research at Facebook's artificial intelligence. From an engineering point of view, tracing evolution will be very difficult." Because evolution lacks initiative, the intelligent ape is created without conscious effort or decision. On the contrary, we can walk to this day because of the random mutations of millions of years, so that we live long enough to reproduce. Maximizing or simplifying our brain's intellectual and reasoning abilities has never been a part of the problem. The human brain is extremely complex. It is full of mechanisms that can be self configured in the womb and can be self - Healing for a long time. The machines do not need these because the process of configuring the configuration is done by humans. It only needs to receive data, process and learn data.
The pioneer of artificial intelligence Jan young is speaking at the latest scientific controversy group at Pioneer Works. The speakers also included the famous physicist Max Tegmark and the host Janna Levin, an astronomer at the Barnard institute.
Lechun explained that for more traditional supervised learning methods, humans must provide thousands of examples to the system before the machine itself does meaningful work. For example, an image recognition algorithm needs to see countless apples in order to identify an apple in the photo. The second approach is reinforcement learning, in which artificial intelligence systems or neural networks - algorithms similar to those of the brain - train each other. This method is usually only for games. A chess player with artificial intelligence can play millions of games, and it takes only minutes to understand the rules of the game.
But neither of these two methods is perfect. Neither of these two methods produces an artificial intelligence that really makes them understand the world. In supervised learning, human beings are still doing all the heavy work, and those playing chess are ignorant.
"We trained the neural network in a very stupid way," says le. "It's completely different from the way humans and animals train themselves."." Babies know permanence at the age of two months. When they were half their age, they were able to intuit the workings of the physical world. But we cannot in our machine to start this unsupervised learning (if anyone can succeed, then it may be lequeux and his Facebook team, because only large companies have the resources and the structure of neural network to train high level). But in the group discussion, he shrugged: "we can't do it."."
That's why the biological basis is crucial for artificial intelligence, not the perfect reconstruction of the human brain. There are no other patterns for programmer reference. The human brain is a scientific wonder, but it is not the only answer. These researchers need to remember that there is nothing special about supercomputers in humans and in our skulls, and they shouldn't try to create something new.
"We don't really understand human thinking," says Janna Levin, an astronomer at the Barnard Institute in New York. He led a team on artificial intelligence technology and ethical futures. "We believe that through mapping we can understand human thinking, but that does not come true."
According to the team's artificial intelligence researchers, this is a tough problem. We can't simulate the human brain perfectly. Instead, we should spend time on unlocking the basic principles of intelligence.
Max Tegmark is a physicist at the Massachusetts Institute of Technology and director of the Life Institute for the future. "Too much attention to the brain is only carbon chauvinism," he said. Although scientists have yet to find the mystery, there is nothing magical about the way the brain works. "We're so obsessed with how the brain works," says Tegmark. "I think it's a lack of imagination."
History has proved his point of view. In Vitoria, a man named Cl ment Ader engineers built the first heavier than air aircraft. He mimicked the bat to build it. These machines are only chairs with large bat wings on both sides. Ader flew hundreds of meters with this almost uncontrollable device. But if he is the first person to fly successfully, why do all the people know that the Wright brothers do not know him?
Ader third version of the aircraft. Although it can keep flying, its steam engine is completely out of control.
Although the creation of artificial intelligence in our own image is not a viable approach, the group's discussions that night are still returning to biology. As Levin said, "human intelligence and consciousness are still our best examples.".
"You can get inspiration from biology, but you must never copy it mechanically," says Jan Lechun, director of research at Facebook's artificial intelligence. From an engineering point of view, tracing evolution will be very difficult." Because evolution lacks initiative, the intelligent ape is created without conscious effort or decision. On the contrary, we can walk to this day because of the random mutations of millions of years, so that we live long enough to reproduce. Maximizing or simplifying our brain's intellectual and reasoning abilities has never been a part of the problem. The human brain is extremely complex. It is full of mechanisms that can be self configured in the womb and can be self - Healing for a long time. The machines do not need these because the process of configuring the configuration is done by humans. It only needs to receive data, process and learn data.
The pioneer of artificial intelligence Jan young is speaking at the latest scientific controversy group at Pioneer Works. The speakers also included the famous physicist Max Tegmark and the host Janna Levin, an astronomer at the Barnard institute.
Lechun explained that for more traditional supervised learning methods, humans must provide thousands of examples to the system before the machine itself does meaningful work. For example, an image recognition algorithm needs to see countless apples in order to identify an apple in the photo. The second approach is reinforcement learning, in which artificial intelligence systems or neural networks - algorithms similar to those of the brain - train each other. This method is usually only for games. A chess player with artificial intelligence can play millions of games, and it takes only minutes to understand the rules of the game.
But neither of these two methods is perfect. Neither of these two methods produces an artificial intelligence that really makes them understand the world. In supervised learning, human beings are still doing all the heavy work, and those playing chess are ignorant.
"We trained the neural network in a very stupid way," says le. "It's completely different from the way humans and animals train themselves."." Babies know permanence at the age of two months. When they were half their age, they were able to intuit the workings of the physical world. But we cannot in our machine to start this unsupervised learning (if anyone can succeed, then it may be lequeux and his Facebook team, because only large companies have the resources and the structure of neural network to train high level). But in the group discussion, he shrugged: "we can't do it."."
That's why the biological basis is crucial for artificial intelligence, not the perfect reconstruction of the human brain. There are no other patterns for programmer reference. The human brain is a scientific wonder, but it is not the only answer. These researchers need to remember that there is nothing special about supercomputers in humans and in our skulls, and they shouldn't try to create something new.