Today many global technology companies are involved in a unique race: strive literally to breathe life into artificial intelligence (AI). Machine learning systems for many has become an integral part of the business, it is not surprising if the news about AI and neural networks caught your eye almost daily. Typically, the headlines of such news are as follows: “AI defeated the man in a video game” or “AI mimics human speech”, and sometimes “the AI more effective person specifies the development of cancer.” Are we really much closer to the time when the intelligence of machines can be compared with the human, or the moment when man and machine will be able to make small talk and work together as naturally as they do among the people? How far is the car from regaining consciousness?
Premature hype
Despite the fact that all of the above “header” achievements AI is real, people like Ian Lecun, Director AI engineering at Facebook and a Professor of computer science at new York University, believe that we are too overestimate the ability of today’s AI systems and create around them excessively increased sensation.
“In fact, we are still far from building machines capable of learning basic worldview in the same way as it can make people and animals”, — commented on Lekun in an interview with The Verge.
“Yeah, can’t argue that in some particular areas of the machine have got features that go beyond human, but in the long term create a common universal artificial intelligence we have not even come close to the level of rats”.
The so called artificial intelligence is a system that does not require participation of a human operator and is capable of performing almost any task that is able to perform. However, current AI systems are specialized and can only work with a particular task, for example, to engage in speech or images or highlight specific objects in a huge data arrays, that is, to do only what they were programmed. Such specialized AI is often called “applied AI” or “niche AI”, that only once again shows their limited intellectual capabilities.
Statements of Lacuna agrees Manuel Cebrián, fellow at mit and one of the developers of Shelley, the AI algorithm capable of writing scary stories.
“AI – it’s just a great tool. But in my opinion, based on my experience with Shelley, the AI is very far from the ability to create works in the horror genre at a professional level, as it is still very far from the level of human intelligence,” notes Cebrián.
Lecun generally believed that, despite the amazing level of progress that has been achieved by researchers and developers of artificial intelligence in recent years, working with machine learning and neural networks is not entirely a development of the real artificial intelligence, about which all so dream today.
“I in no way want to belittle the achievements of our fellow engineers and researchers from DeepMind working on the same AlphaGo, however, when people interpretiruya improvement AlphaGo as significant progress in the development of General artificial intelligence is wrong. Because it is not so,” says Lecun.
Pierre Baro, CEO of Aiva Technologies, developed AI algorithm Aiva music creation, also believes that the achievements that we have made in creating synthetic intelligence, somewhat exaggerated.
“General artificial intelligence is a topic that attracts a lot of attention. In General, I certainly optimistic about how quickly the technology is evolving, but at the same time, I believe that most people simply do not understand the complexity of our own brain, not to mention how difficult it would be to create an artificial,” says Baro.
The creation of General artificial intelligence
Today people love to use the term AI for any reason, even where there can be a conversation about something else. Any news about AI can be found such terms as “machine learning” or “deep learning” and “neural networks”. Although each of these terms is to some extent connected with AI, but in fact it is not an AI as such.
Machine learning is a tool. A set of algorithms, components of the intelligent system which is trained by the absorption of huge amounts of data. Also, deep learning is a machine learning, not necessarily tied to a particular task. On the other hand, neural network is a system that simulates the brain, but again it works only in the context of those features, according to which generated machine learning algorithms.
Experts in the field of AI believe that all three above-mentioned components are the fundamental basis for creating synthetic intelligence, the ability for a human to think, that is to understand their actions and their consequences. But we are only at the beginning of this journey. No, we really have made great progress, but the current development virtually moved us from place to the creation of real intelligence. Nevertheless, it is interesting to ask the question of when we should expect this type of artificial intelligence. Is there any time frame?
According to Luke Tang, head of a startup TechCode dealing with AI, this shift to the creation of full artificial intelligence “will begin with a breakthrough in the development of unsupervised machine learning algorithms”. As soon as we come to this, “machine intelligence will exceed human”, said Tang in an interview with Futurism.
To say that it will be difficult to achieve practically nothing to say.
“Create a General artificial intelligence, a major progress in software development. Significant progress in the development of neuroscience and the development of hardware,” says Baro.
“We stand on the edge of the capabilities of Moore’s law, when transistors become so small that it is impossible to obtain physically. And new hardware environments, such as, for example, quantum computing, yet are unable to demonstrate their superiority over our ordinary hardware, even when performing common tasks”, — the expert adds.
Many agree that for dealing with the AI as a real intelligence it needs to meet five specific tasks, the first of which is the Turing test, where a machine needs to convince people that he is talking with another person and not a machine. The same Baro convinced that the present generation will witness how the AI copes with the Turing test, that is actually fooling people. However the expert believes that “this need not necessarily be exactly the General artificial intelligence, but something that to it is closer”.
The amplification of intelligence
It is impossible not to note that the emergence of General artificial intelligence will be the harbinger of the so-called technological singularity. For those who have forgotten, recall that the concept of a technological singularity tells about the time when intelligent machines will surpass human intelligence, stimulating rampant and exponential technological growth, which at a fundamental level promises to transform our lives. The author of the term and concept, is Vernor Vinge who in 1993 wrote the following:
“Soon we will be able to create intelligence that will surpass our own. When this happens, human history will reach a kind of singularity, an intellectual transition to a new level. It is impossible to escape just as it is impossible to escape from the black hole. Since then, the world will change so much that it is beyond our understanding”.
Despite the fact that this moment is what it is waiting for such people as the CEO of SoftBank, Masayoshi son and futurist ray Kurzweil, there are those (like, for example, Elon Max, Stephen Hawking, and even bill gates) who are clearly not very happy this perspective. They argue that people just don’t understand what it really means to artificial superintelligence, and we are clearly not ready for the possible consequences that can bring a technological singularity.
But why is it necessary to look at the question from that point of view? Why it is necessary to consider artificial intelligence as the sunset for all mankind, and not as his companion, friend, helper, in the end? Musk, to be fair, considering this idea, so he created the project Neuralink. Kurzweil mentioned this cooperation between man and machine when talking about that in the future inside, we will live the nanobots, repeatedly improves our capabilities.
“We must focus on the advantage that we can give the AI – enhanced intelligence, which is human intelligence, whose possibilities will be improved due to the AI,” says Baro.
Algorithms that are similar to Aiva and Shelley, already show their use in working with people. At the same time, intelligent robots, like Sophia from Hanson Robotics and Pepper from SotfBank, can easily allow us at least to imagine that truly intelligent machines are already among us. Perhaps the superintelligence with an IQ of 10 000 in the presentation, Japan’s Sona will really become the cognitive machine intelligence, to which we all aspire? If so, then wait for us remained not so much – some three decades, experts say.
“This level of AI we will reach, perhaps in 30-50 years. On the one hand, it may seem that it’s very long, but it means that many of us will have the chance to live up to this moment,” concluded Tang.
We are thirty years from the appearance of consciousness in machines. But the hype is premature
Nikolai Khizhnyak