Why are we wrong to perceive the artificial intelligence
When we talk about the AI, the conversation inevitably turns to the category of science fiction scenarios, they say, if the AI does not take away our jobs, you kill us all. But the truth is that the AI has been around for almost 60 years, more and more penetrating into every part of our lives. Our AI can read your emotions, to solve complex geometric problems, draw pictures, like Vincent van Gogh.
Ignoring what was actually created and is currently in use, and instead focusing on the version of AI, which has not yet arrived, humanity has got a blind spot in this technology.
This blind spot distorts our understanding of AI, its usefulness and the progress that has been made in this area. It also brings us a lot of frustration when there are AI does not work as we expected. This phenomenon has a name - the effect of AI.
In effect, the two phases of AI. First, people do not see the program, which interact with both "reasonable" and therefore believe that research in the field of AI to nothing lead. But we are already surrounded by AI, and becomes more and more, so we like a frog in a pot of water, do not realize that the water is getting hotter and hotter.
AI that we have now, not like anything that draws most people in their science fiction dreams - on machines that think and act like a man, a common artificial intelligence (AIS). Instead, we have a narrow artificial intelligence (UII), which is very good in performing specific tasks such as recognizing pictures or trading. When the AIS is established, it is still beyond comprehension.
"In the early years of AI has always been a concern that the AI will never live to fulfill their promises, because all that work, by definition, do not like the AI" - says Subbarao Kambapati scientist computer science from the University of Arizona.
Carlos Gestrin, CEO Dato Seattle, which is building the AI algorithms for data analysis, says that this may be because the UII is not like human intelligence.
"As soon as something is done, it is no longer AI, - says Gestrin. - It is a matter of perception - as soon as something becomes commonplace, demystifies it no longer looks like a magical intelligence that we see in others. "
On the other hand, it also gives rise to fear of the unknown "future" AI, which seems to always be hiding around the corner. When people say that it is possible to AIS, the conversation is always accompanied by fears on the subject of what the AI can suddenly turn.
"I think that this idea is that the AI is going to break out like thunder, raises the question, well, what do we do when AI is here - says Sabine Hauert, robotics, University of Bristol. - In reality, we are working on the AI for 50 years, and improvements are miserable. "
This fear of the future humanoid AI is based on the biased feeling that this technology, which has existed for years, suddenly acquire human attributes, and is anthropomorphism. But given the way we are trying to create an AI now, it is unlikely that the future of AI are human attributes - elements such as emotion, consciousness, or even self-preservation instinct, according to Joshua Beng, a computer scientist from the University of Montreal. Because intelligent AI will have a completely different intelligence than men.
"The biggest misconception is the idea that is common in science fiction, the AI will be similar to another living being that we can imagine, animal or alien - consequently, this AI is the ego, the mind, like a man - says Beng . - However, the machine may be reasonable and without self-consciousness, the ego, and even self-preservation instinct. "
In the end, the smartest car in the world does not think like a man
Shimon Uaytson, a scientist from the University of Amsterdam, explained why people default AI assign human traits.
"We have a tendency to confer human features of any type of intelligence, because we live in a world in which people are the only example of a high level of intelligence, - says Uaytson. - We really have no idea what could be the intellect, if not one person. "
Through research in the field of AI, we find that there may be many other types of intelligence. Not every program needs to be intelligent in fact humanoid. When the technology of AI learned to do a specific task, she did not like the man, but most people will not see it as the AI. But when will OUI, he, too, will not look like a man.
"Intellect - is not the only feature of the system, - said Tomaso Poggio of MIT. - Intelligence - is one word that can refer to many things. We measure intelligence by how well a person or a computer can perform the task, including the task of teaching. For this indicator, computers are now much smarter than people in many tasks, including memorizing things, arithmetic calculations, trading on the stock exchange, the landing of aircraft. "
To put an end to the paradox - a wildly tossing between conviction that AI had not yet arrived, and that when he arrives, we all Khan - need to rethink the concept of human intelligence. We need to understand intelligence in a broader sense and understand that the machine that does the work is reasonable. The sooner this happens, the easier it will be to focus on the real risks and benefits, which, according to researchers, can bring the future of AI.
When we talk about the AI, the conversation inevitably turns to the category of science fiction scenarios, they say, if the AI does not take away our jobs, you kill us all. But the truth is that the AI has been around for almost 60 years, more and more penetrating into every part of our lives. Our AI can read your emotions, to solve complex geometric problems, draw pictures, like Vincent van Gogh.
Ignoring what was actually created and is currently in use, and instead focusing on the version of AI, which has not yet arrived, humanity has got a blind spot in this technology.
This blind spot distorts our understanding of AI, its usefulness and the progress that has been made in this area. It also brings us a lot of frustration when there are AI does not work as we expected. This phenomenon has a name - the effect of AI.
In effect, the two phases of AI. First, people do not see the program, which interact with both "reasonable" and therefore believe that research in the field of AI to nothing lead. But we are already surrounded by AI, and becomes more and more, so we like a frog in a pot of water, do not realize that the water is getting hotter and hotter.
AI that we have now, not like anything that draws most people in their science fiction dreams - on machines that think and act like a man, a common artificial intelligence (AIS). Instead, we have a narrow artificial intelligence (UII), which is very good in performing specific tasks such as recognizing pictures or trading. When the AIS is established, it is still beyond comprehension.
"In the early years of AI has always been a concern that the AI will never live to fulfill their promises, because all that work, by definition, do not like the AI" - says Subbarao Kambapati scientist computer science from the University of Arizona.
Carlos Gestrin, CEO Dato Seattle, which is building the AI algorithms for data analysis, says that this may be because the UII is not like human intelligence.
"As soon as something is done, it is no longer AI, - says Gestrin. - It is a matter of perception - as soon as something becomes commonplace, demystifies it no longer looks like a magical intelligence that we see in others. "
On the other hand, it also gives rise to fear of the unknown "future" AI, which seems to always be hiding around the corner. When people say that it is possible to AIS, the conversation is always accompanied by fears on the subject of what the AI can suddenly turn.
"I think that this idea is that the AI is going to break out like thunder, raises the question, well, what do we do when AI is here - says Sabine Hauert, robotics, University of Bristol. - In reality, we are working on the AI for 50 years, and improvements are miserable. "
This fear of the future humanoid AI is based on the biased feeling that this technology, which has existed for years, suddenly acquire human attributes, and is anthropomorphism. But given the way we are trying to create an AI now, it is unlikely that the future of AI are human attributes - elements such as emotion, consciousness, or even self-preservation instinct, according to Joshua Beng, a computer scientist from the University of Montreal. Because intelligent AI will have a completely different intelligence than men.
"The biggest misconception is the idea that is common in science fiction, the AI will be similar to another living being that we can imagine, animal or alien - consequently, this AI is the ego, the mind, like a man - says Beng . - However, the machine may be reasonable and without self-consciousness, the ego, and even self-preservation instinct. "
In the end, the smartest car in the world does not think like a man
Shimon Uaytson, a scientist from the University of Amsterdam, explained why people default AI assign human traits.
"We have a tendency to confer human features of any type of intelligence, because we live in a world in which people are the only example of a high level of intelligence, - says Uaytson. - We really have no idea what could be the intellect, if not one person. "
Through research in the field of AI, we find that there may be many other types of intelligence. Not every program needs to be intelligent in fact humanoid. When the technology of AI learned to do a specific task, she did not like the man, but most people will not see it as the AI. But when will OUI, he, too, will not look like a man.
"Intellect - is not the only feature of the system, - said Tomaso Poggio of MIT. - Intelligence - is one word that can refer to many things. We measure intelligence by how well a person or a computer can perform the task, including the task of teaching. For this indicator, computers are now much smarter than people in many tasks, including memorizing things, arithmetic calculations, trading on the stock exchange, the landing of aircraft. "
To put an end to the paradox - a wildly tossing between conviction that AI had not yet arrived, and that when he arrives, we all Khan - need to rethink the concept of human intelligence. We need to understand intelligence in a broader sense and understand that the machine that does the work is reasonable. The sooner this happens, the easier it will be to focus on the real risks and benefits, which, according to researchers, can bring the future of AI.
Комментариев нет:
Отправить комментарий