9 naive questions about artificial intelligence
Technologies / / December 19, 2019
Azat Shamsutdinov
The project manager and director of digital technology Sever. AI.
What is AI?
On the definition of artificial intelligence (AI) fighting mathematicians, computer scientists, futurists, and philosophers. On the one hand, this large-scale interdisciplinary research field at the intersection of mathematics, linguistics, psychology and programming.
On the other hand, AI - a mathematical algorithm by which to establish an intelligent computer programs. They are so named because they can perform the functions that are considered to be the prerogative of man. For example, write a poem or music, keep the conversation going.
AI created for specific tasks - all such systems are highly specialized and can not perform several functions at once. If the program is designed to decipher human speech, it will never be able to play the card game.
To distinguish between weak and strong AI. When companies say they have created a product with artificial intelligence, we mean weak variants: the autopilots, voice assistants, translators. Arguments about the strong AI, the one who will be able to think and to realize itself (that is actually be on a par with the human mind), remain until scientific and philosophical discussion.
Weak AI when translated text replaces some words on the other by a certain algorithm and strong - can independently understand the meaning of the sentence. That's the main difference.
Robots - it is also the AI? A chat-bots, machine learning, neural networks?
No, we are talking about a very close and interdependent concepts, but still it is not the same thing. Let's go back to the definition: AI - scale interdisciplinary area of research (geography).
In this area there are several specialized areas of knowledge, one of which is machine learning. Along with him was isolated processing natural language text, virtual assistants, recommendation system. This is how the physical, economic and social geography.
Descend even a step below. Neural network - one of the subsections of machine learning, mathematical algorithm with automatic settings. There are four main subsections (way) machine learning: classical, with reinforcements, and the neural network ensemble methods. Imagine that this is the geography of the oceans in the global section of physical geography.
And where in this case include the robots? Chat bots, robots, Automatic translators, scanners - all this the final result and the format of the AI technology.
Can I meet AIs in everyday life?
Of course! We use it for a variety of tasks. For example, T9 predicts the word that you write - is the AI recognizes a combination of letters and offers one of the options available in its database. Home robot assistant that responds to voice commands - example of the effect of technology. Siri - it's too artificial intelligence.
AI can be trained?
Yes, he was able to learn and improve. For example, Google has created aMastering the game of Go without human knowledge AI, which is itself mastered ancient Chinese board game first, learning through mistakes and victories.
However, we should remember that self-learning - is not necessarily a sign of artificial intelligence. There are systems that are simply very well perform a specific task and are "not entered" learning function. These include robots operating in manufacturing and dealing sorting.
And to recognize emotions?
Yes, there are computer systems that can recognize emotions in facial expressions of the person during the interview. The program evaluates the position of the key points of the face (eyebrows, eyes, nose, jaw and lips) and correlates with symptoms of possible emotions, which are written in its code.
Moreover, the systems are able to depict the emotions using emoticons or Emoji. At the heart of such manipulation is very simple logic: basic emotions (happiness, joy, resentment) is easy to predict and simulate based on trigger words ( "thank you", "sorry", "insulting" and others).
And preventing the following question: no, experience of AI emotions can not. They formed a human in the process of evolution and socialization. Such a reaction system is present in animals - they experience joy, rage, anxiety, and so on. But once a person has added this spectrum ability of rational thinking.
Can an AI become more intelligent person?
On the one hand, it's pretty meaningless question, because there is no universal intelligence measurement scale. For example, we know that the heart rate of a healthy person - about 60 beats per minute. But how to measure intelligence? The number of read books, Periodic table knowledge or ability to provide answers to any questions? Can we consider intelligent cat protein and an eagle - smarter viper? How to compare the intelligence physicist and astronomer surgeon?
There is a popular test for IQ measurement (IQ) Hans Eysenck, but consider it a universal criterion, you should not. In humans, the brain works differently and "locked up" for this or that type of activity. While there is no criteria on which to navigate as an absolute figure, there will be a similar rating.
On the other hand, when we say "the machines will be smarter than human," we rather assume that they will become wiser. The mind is much broader intelligence, it is formed in the course of life and depends on various factors billion. While scientists and science fiction writers suggest the only possible (but not yet implemented) option, in which the AI become wiser man: if the technology will be implemented on the basis of DNA molecules rather than by modeling neural networks.
Can AI evaluate their actions?
Not. In order to evaluate the action, in addition to human thought process required moral attitudes, emotions and cultural norms, which change over time. Technology is (at least so far) is not available.
Is it possible to hack the AI?
Yes, probably. This program, which is managed by a man. And any program can be hacked.
is it possible that the AI will get out of control and decide to get rid of the man?
Despite the fact that we saw a lot of scenes where robots become aggressive monsters, block city life, capture confidential information, and doing other crimes, it is possible only in movie.
AI performs only those tasks that have been incorporated into it as a programmer. Independent goal-setting does not provide the technology. Artificial intelligence can be in the hands of criminals and cause harm, but we are again talking about the will of man.
Another scenario could arise only if it solves the problem of strong AI. Today it is impossible. Therefore, all the arguments on the likely enslaving humanity AI have no basis in reality.
see also🧐
- 10 skills that will never have robots
- 7 achievements of Artificial Intelligence in 2018
- Fiction and reality: robots from the movies and video games come to our world
- Why create a humanoid robot Sofia and whether it threatens humanity