Artificial intelligence to be

In connection with a recent article about the impossibility of creating an AI, I want to offer you a different view and argument on the possibility of creating an AI, while it is quite possible that you and I will witness to that.

There is no particular desire to plunge into the philosophical aspects of what can be called intelligence, especially human. In the end, most people who read this topic are more interested in technical implementation than in discourses on a spherical horse in a vacuum.

Very many, concerning at least casual topics of artificial intelligence, are embarrassed by the fact that not all human actions can be algorithmized. On the reverse side of the reasoning, it is important to note that there is a simple pattern in the head of any programmer: everything that can be described by an algorithm can be implemented programmatically.

As a result, everything that separates us and you from creating a full-fledged artificial intelligence lies in the area of ​​the simple question “What are the algorithms of actions / actions / reactions of a person?”.
')
We all know very well that a person can count, read, write, make calculations, draw logical conclusions, process information. These are the parts of his activity that we have already successfully transferred to the area of ​​machine code. Yes, yes, it was not machine code that created this in man, but man realized a part of his thinking in machine code. Existing decision-making systems even partially emulate complex logical human responses. Individual computer systems even successfully compete with a person with intelligent quizzes .

Thus, we have already realized that at least half of the “artificial intelligence” we already have. It remains to "finish" "the impossible."
What is the most difficult part of the algorithmic model of AI? Again, let’s leave long discussions about terms like “Soul, love, compassion, humanity, friendship, etc.” and simply draw a logical conclusion: when a computer can experience emotions, we can say that it is artificial intelligence. After all, emotions are considered the very "holy grail" of algorithms.

If you do not agree with this conclusion, then try to imagine a program that experiences emotions when communicating with you, or when discussing a topic that expresses your feelings towards something that protects another program or hates any a virus that is experiencing feelings - surely this program will not be worthy of being considered an AI? If you still do not agree, then I will be glad to see in the comments to this topic your definition of which, by what features, you determine whether something like human intelligence (technically, please).

So, emotions.
As you can conclude from my past topics , I am a programmer (in the present) and a psychotherapist (in the past). The topic of AI for me is also quite attractive due to the knowledge in such important areas for AI and, in turn, I myself have devoted more than one year to independent research in the field of mental algorithms.
Many researchers in the field of psychotherapy, like me, are similar in simple opinion - emotions are algorithmized. Full Most psychotherapists are well aware of this; for this reason, there are quite specific algorithms for how to get rid of certain emotions, or how to create them. People successfully treat various mental disorders associated with emotions (the same phobias and fears), and according to completely natural instructions. At one time, I even described the entire basic theory of the psyche and laid it out in an open access, you will not find anything new for yourself there, just everything is on the shelves, and most of you, probably, have already read it.

As a result, we have the simplest algorithms (arithmetic, logic). Let us add here already existing technologies for emulation of the sense organs (computer vision, speech recognition, tactile sensors). Let's dilute it with algorithms of emotions and the formation of the human psyche. We supply all this with a repository of information. We will add a “parasite program” that will make the original program less ideal (erase and block portions of memory, cause slowdowns in work, correct external factors, organs of perception, etc.) in order to more likely to imitate a person. And we can easily get artificial intelligence (to collect from "weak AI" "strong AI" ). It sounds beautiful, but it is a "spoon of honey."

Now the "barrel of tar."
To know "how it should be" does not mean "it will work now." There are still many problems to be solved before the AI ​​is implemented and presented to the public. Apart from exclusively the legal and moral aspects of the existence of AI, there are also technological barriers.

First, as it is not trivial, it is computing performance. Now it is simply not enough to implement such complex complex calculations with such volumes of information “aby where”, as a result, the first AI will be born of itself first “behind seven seals”. So we will know about this when and all.

Secondly, not everything is completely clear with separate perceptual algorithms, which are supposed to be encoded in the genome and most likely their first implementation will also be a “ black box ”. For example, there is not yet a clear idea of ​​how a person develops a taste for beauty, a sense of style, etc., although there are already certain prerequisites for solving this problem. In particular, they consist in the fact that it is not surprising that a person subconsciously searches for algorithms (patterns) and considers them “beautiful”, for example, fractals. Partial answers to this are in the mechanisms of emotions (for example, in the mechanisms of love and sympathy). Actually, mathematicians have long noticed the beautiful in the laws, or the terrible in their absence .

Thirdly, all of the above technologies have different platforms, and although individual researchers are trying to find a comprehensive solution that would simulate the brain and thereby emulate the entire AI as a whole, there is still a lot of work to be done to bring it all together in a "strong AI."

The fourth important aspect, which is not so obvious and a casual observer, but familiar to all who faced with complex AI, is the problem of the concept of time. Until now, there is no normal technological concept of how the brain perceives time and what time is in general. So, of course, the first implementations will simply push off not from seconds, but from the clock frequency (like everything else in the computer), which will be adjusted as the theory of time develops.
If you do not quite understand the meaning of time for implementing AI, then try to understand how you perceive time, how you memorize and play a melody, how you connect various events stretched on a time line, why time tends to slow down and speed up in perception. This is quite a non-trivial task.

The fifth aspect is “stuffing”. A person is not something that originated from the combination of sperm and egg, a person (in terms of intelligence) is a whole complex of knowledge / skills / preferences that were formed by his experience, and especially his social experience. It is social experience that forms almost all human emotions. And it is the social experience that will be a significant problem for AI at the initial stages, so the first “filling” will most likely also be an emulation of some person, which again can hardly be called an AI, because it “did not become like that, but they did it” . That is why the first AI will not have the usual emotions.

You can list a number of individual less significant factors, but each of them can somehow be resolved. Mankind has repeatedly set itself very ambitious tasks and successfully solved them.

I just want to make a small remark that this is only about creating artificial intelligence, and not about a “copy of a person”. We are still far from highly developed biotechnologies, and in the field of how our body functions, not less unknown than in the field of quantum physics. Creating an AI is not a game of god, it is only a creation of what a person himself would consider artificial intelligence according to individual criteria, in particular, for example, communication via a monitor screen.

Given all this, I personally do not “believe”, but I know for sure that artificial intelligence will be created. I do not know when, but it is only a matter of time.

PS:
Frequently asked question “What to read on this topic?”. Regarding the AI, it is possible to “tinker” with this list of links (more “popular resources”):
www.bbc.co.uk/search/news/artificial_intelligence
www.csail.mit.edu
www.aaai.org
slashdot.org/tag/ai
www.ted.com/search?q=ai
aima.cs.berkeley.edu/books.html
video.mit.edu and especially video.mit.edu/search/?q=marvin+minsky
And I also collected a small list of books (with a partial mental inclination) here .
If you have something to add, write in a personal or a comment.

Experience shows that the topic of AI is an excellent reason for holivar, therefore, I ask readers in advance to be more constructive in their judgments / comments.
Let's respect each other, "your cat Leopold")

Thanks for attention.