What is behind Artificial Intelligence?
Artificial intelligence can be defined as a science that modeling intelligent human behavior. This definition may have one significant drawback – the concept of intelligence is difficult to explain in principle. The problem of defining artificial intelligence comes down to the problem of defining intelligence in general: is it something in common, or does this term combine a set of disparate abilities, and even more as an individual or even collective abilities?
To what dimension can intelligence be created? What is creativity? What is intuition? Is it possible to judge the presence of intelligence only by the observed behavior? What is human intelligence and especially the intelligence of a machine or program? Is it possible to reduce human intelligence to an algorithm? And here, the question is more likely even philosophical than scientific.
To be honest, no answers have yet been found to these questions, but they all helped formulate the tasks that forming the basis of modern artificial intelligence as a scientific approach. Part of the attractiveness of AI lies in the fact that it is an original and powerful weapon for researching these very problems. AI provides a tool and test model for theories of intelligence: these theories can be formulated in the language of computer programs, and then tested.
The problem of finding the exact definition of artificial intelligence is understandable. The study of AI is still a young discipline, the structure of this phenomenon in science is still being formed, so only with time, a clear thesis will be generated in the public mind about what AI is. However, it is already visible today that AI is designed to expand the capabilities of primarily computer sciences, and not to determine their boundaries.
The next step may be the expansion of the intelligence of the human being himself. One of the important tasks facing researchers is to maintain these efforts with clear theoretical principles that are currently a problem.
The most optimal definition for today is the following: AI is a field of science and engineering that creates machines and computer programs that have intelligence, or AI is a field of computer science that develops intelligent computer systems, that is, systems that have capabilities that we traditionally associate with the human mind – understanding of the language, learning, the ability to reason, solve problems.
As a result, AI should become a unique product of technological progress, which will allow machines to learn, using human and their own experience, adapt to new conditions within the framework of their application, perform diverse tasks that for a long time were only possible for humans, predict events and optimize resources of various character. Most of the examples of using AI known today – from computers playing chess to autonomous robotic systems – still depend on the human factor and require deep training.
However, even at the stage of their current progress, they globally affect the life of the whole society, forming new ideas about the future and prospects for the development of modern technologies. So far, AI has not become the ability to come to a final decision through calculations, the human factor in monitoring the results of applying AI still dominates decision-making, as long as there is access to the algorithm code, the results of calculations and observations/conclusions AI can be changed and influenced. However, how the person of the future wants to let go free and forget about access to such an algorithm code is a rhetorical and manipulative question.
While there is no understanding what types of computational procedures we want to call intelligent, we know far from all the mechanisms of our intellect to talk about the artificial. Moreover, the concept is still fabulous, in which futurists scare us with the fact that in the near future, AI will completely replace human intelligence. After all, as long as researchers using algorithms that are not observed in humans or require much larger computing resources, artificial attempts to replace us are too fantastic.
But this is only so far, at this stage of the development of technology and the youngest science. Arthur R. Jensen, a leading researcher in the field of human intelligence, as a “heuristic hypothesis” claims that people have the same mechanisms of intelligence and intellectual differences are associated with “quantitative biochemical and physiological conditions”. These include the speed of thinking, short-term memory, and the ability to form accurate and retrievable long-term memories.
The situation in AI is the opposite. Computer programs have a large margin of speed and memory, but their abilities are corresponding to intellectual mechanisms that program/algorithm developers well understand and can invest in them, that is, the result tends to be the way researchers still see and program it, to the point of initiative AI itself is still far away, and Turing’s tests, even if it is successfully completed by the machine/AI, in fact, will not mean victory in the simulation of human intelligence. This will most likely be another achievement that will only bring us a little closer to an idealistic result.
The ultimate goal is to create computer programs that can solve problems and achieve goals in the same way as a human. Again, the quality of human intelligence is flexibility and mobility, the admissibility of recognizing one’s mistakes, the use of experience, both positive and negative, as soon as AI machines will possess such qualities, even if they can pass the Turing test and solve certain problems much faster tasks instead of the man.
The main mistake of scientists here is in the desire to replace human with AI, and it should be according to the ideation, not a replacement, but an addition. As long as there is such a mistake in setting the results, there will undoubtedly be a threat of “victory of the machines over humans”. For AI, it is important that when solving problems, the algorithms are as effective as the human mind. The determination of subdomains in which good algorithms exist is important, but many programs that solve AI problems are not related to easily identifiable subdomains.
By the way, the computing power of the machine is greatly exaggerated. Yes, as a calculator, a human cannot compete with a computer. But what most consumes a machine resource? Any gamer will say – processing video information. However, the human has no problems with this. The processing and analysis of video information by humans are still an order of magnitude superior to the capabilities of the machine, and when you consider that both auditory information, olfactory, tactile, and coordination of movements are processed simultaneously – and all this online, then there is nothing to be afraid of. Pattern recognition for a machine is a very difficult task, the task of developing that is very intellectual. Besides intelligence, can a machine be intelligent? After all, the main diversity of human minds is the will of irrational actions. For example, the desire to unknown. Therefore, perfection is still a long way to go.
What about perspectives?
Humanity has made a powerful evolutionary breakthrough, leaving far behind other biological forms of life. Driven by the development of technology, the process of mastering the natural environment, the complexity of human social life, filled with artificial technical inventions, have reached their zenith in modern times.
Previously, the development of technology focused on the design of devices that simulate with much higher performance than in their natural manifestation, external senses and organs of human action: instead of natural vision – a microscope or binoculars, instead of a hand – an excavator, instead of natural hearing – radio communication, instead of legs – car, etc. And then there appeared devices designed to imitate and replace, it would seem, the most important thing in human that which has long been recognized as its most significant attribute – rationality. AI systems were designed to reproduce and, possibly, in the future replace at a higher quality level the process of human thinking, its ability to rational intellectual actions.
Despite the alarming prophecies of Elon Musk, the “strong” intellect, “uprising of machines” is certainly far away, but the “weak” AI has already firmly entered our lives and has found wide practical application. The hype of recent years in machine learning has fundamental reasons and is quite justified – business has become very attractive for these “smart” technologies, and this is not only for the image or for a tribute to “fashion”. They give a specific economic effect.
For example, McKinsey analysts estimate the AI market by 2025 to $126 billion, while spending per year up to $30 billion by major players in recent years. In addition, the numbers will only increase over time. In many respects, the increased interest in AI on the part of specialists is caused by a new stage in the development of neural network technologies, like deep neural networks, but the revolution in working with data played a decisive role in this.
We can digitize that countless amount of information that life itself generates every second, we can store it, process it and, most importantly, we want it, we try, and we can analyze it in many ways. The combination of the development of Big Data, the possibilities of Data Engineering and, of course, Data Science, against the background of global “Internetization” and the widespread dissemination of the IoT, led to an international conferences, where reports without mentioning AI are not included in the program, every startup threatens to revolutionize the world with AI, and every self-respecting company leader (in any field) considers having a machine learning department as mandatory.
However, quantity is not always the quality of all of this. Most mathematical models have long been known, but it is the big data and the hardware capabilities of their processing in the “more” real-time mode that led to such a boom and the emergence of new specialties that are still not very professionally trained, but where they want to hire a lot – Data Engineer and Data Scientist.
If we talk about the main scientific and technical areas, AI today includes the following: machine/deep learning and predictive analytics, Natural Language Processing (NLP), smart robots and computer vision. But it’s more practical to consider these areas in the context of their business applications, and this is what Data Scientist is thinking about.
In the forefront of the application, AI began to use the trading sector, as well as fintech, manufacturing, healthcare, and sports actively use many AI models and, most importantly, invest in their development in the future. For example, retail trade – targeted, personalized interaction with customers, recognition of their behavior, virtual assistants and smarter by the training structured chatbots, optimization of the geolocation of retail outlets, layout of goods on the shelves of trading centers, smart contracts with suppliers, the use of robots for warehouse operations all this led to lower costs and increased sales.
The greatest practical application has now received computer vision and natural language processing (NLP). But NLP is perhaps of a larger and longer-running nature. Today, even such conservative industries as insurance and legal services are beginning to implement AI. There is a change in the familiar, as it seemed, already unshakable procedures. While we are not talking about the complete disappearance of professions, but, of course, the number of specialists required in these sectors will be steadily decreasing. It will be only highly qualified professionals who will have to keep up with technology in order to remain in demand.
Nevertheless, what, in principle, can AI do today despite criticism, skepticism and revolutionary hype? In principle, if you are systematizing your merits, you can do many things. Today AI can:
- Automate the continuous learning process and search using data (For this type of automation, the human factor is still necessary in order to ensure an effective and correct system for processing key requests and making appropriate decisions).
- Perhaps intellectualize the product (AI turns standard automated systems into an intelligent product that works on user requests).
- Trying to adapt (AI develops using progressive learning algorithms and generates data for further programming).
- Analyze deep data (a thorough analysis brings to the surface all potential risks, generates forecasts and warnings, eliminates the adoption of erroneous decisions, prevents unsafe situations when playing a specific technical process or events).
- Strive for accuracy (in all spheres of human activity – medicine, agro, trade, engineering, entertainment, construction and so on).
- Operate already large data.
In addition, where it is already actually used:
- In the military – defense complex.
- In education, where there are great prospects for the development of AI products.
- In business, basically in the fight against fraud, in the electric power industry, in the manufacturing sector, in banking and financial services, in transport and logistics, in trade and in the art and luxury goods market.
- In public administration, as in forensic science, in the judicial system (the state program of China), in sports and medicine, analysis of citizens’ behavior (again the program of China’s Social Rating).
- In culture, in the media and literature, video, music, painting, special effects, games, and photography.
- In space exploration, in predicting solar storms and possible protection from asteroids, in the discovery of exoplanets, today we are actually reporting from the International Space Station (Int-Ball drone).
Artificial Intelligence Revolution: Hype, Scam, or “Big Brother” – Part 2
Source: Becoming HumanRelated posts: