What is behind Artificial Intelligence?
Artificial intelligence can be defined as a science that models 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 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 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.
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. Today I repeat once again, 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 speed of thinking, short-term memory, and the ability to form accurate and retrievable long-term memories.
The situation in AI, as I said, is the opposite. Computer programs have a large margin of speed and memory, but their abilities are corresponding to intellectual mechanisms that developers well understand and can invest in them. The result tends to be the way researchers still see and program it. True 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 man. I think not soon.
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 the human minds is the will to irrational actions. For example, the desire to unknown. Therefore, perfection is still a long way to go.
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.
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 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, as 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 of criticism, skepticism and revolutionary hype? In principle, if you are systematizing your merits, you can do many things. Today AI can:
In addition, where it is already actually used:
My opinion, at this stage, taking into account the available results, of course there is no threat so far.
All this is fantastic in the style of the Terminator and the fight against Skynet. However, if there is possible the biological synthesis (synthesis of the human mind and artificial intelligence), the danger of using the results obtained in malicious intent is possible.
The danger is possible if AI is created on a biological basis, that is, not on modeling neural networks, but on growing DNA-based neural networks with simultaneous programming. However, the suspense is always scary. Whenever humanity was on the verge of new discoveries, innovative developments, or technical revolutions, people were afraid: what would bring these radical changes?
Therefore, it was in the era of the transition from horses to cars, and at the dawn of the development of electricity, and during the development of the World Wide Web. Some saw changes in perspective, others saw a threat. What are alarmists talking about today, which include well-known scientists and businessmen (Elon Musk, Hawking):
Nevertheless, the more real threat today is the practical implementation of the theory of ‘’Big Brother’’ by the state and special services using AI technology. Moreover, here it is already worth fearing today!!!
The expression about Big Brother is well known, both to lovers of social dystopias and to fans of literature in general. The phrase “Big Brother is watching you” gained fame after the release of the novel by the famous British writer J. Orwell “1984”, which continued the theme of the “faithful” revolution, which began in his work “Animal Farm”, which was an allegory for the October Revolution of 1917. In modern society, the term “Big Brother” is used to denote totalitarianism, anti-democracy and surveillance. The theory is that intelligence agencies of all developed countries organized a mechanism for total surveillance of citizens, including surveillance of Internet users. First of all, this concerns the work of American intelligence agencies, which are likely to process more data than leading technology companies.
In September 2017, Dawn Meyerriecks, deputy director of the Central Intelligence Agency for Technology Development, said her agency was working with 137 AI projects, many of which were related to Silicon Valley companies. This is an impressive figure. Apparently, the CIA is going to make AI technology the main tool for working with information, which means significant cash injections in this area. US intelligence agencies are already working with a gigantic amount of data, which, in fact, includes the entire Internet. The CIA is currently working on creating predictive algorithms with AI elements that could find non-obvious causal relationships in disparate data sets. Such systems should alert intelligence analysts to important events that slip out of sight of conventional tools. Decisions made on the basis of machine analysis will be used to make political and military (operational) decisions.
It should be noted that now the special services, delivering daily reports to the country’s leadership, are not able to assess the situation. Only tracking social networks requires gigantic resources, not to mention the analysis of satellite images, statements by local media and news messages in various social media and chat rooms. Therefore, special services will be forced to use developments in the field of AI. The beneficiaries have already become large technology companies that can be as data collectors, such as Alphabet Inc. (GOOGL) and Facebook Inc. (FB), and vendors of flexible AI platforms to perform various operational tasks, such as IBM Watson from International Business Machines Corp. (IBM).
The Chinese Government, which has already done a lot to implement the policy of the theory of the ``big brother ‘’. So there are AI threats to humans, but again, these same threats come from the human and the state itself, as a form of organization of control of human society. Another more real threat than the Rise of the Machines may be the same human factor, or rather, a Human interest in the easy money associated with the hype around AI.
Unfortunately, the hysteria around AI generates numerous pseudo-revolutionary startups that can only wash away an investor’s wallet and not create a revolutionary product.
Under the cover of AI technology and its development, ordinary routine processes are hidden that have nothing to do with AI. For example, former Engineer.ai employees told the Wall Street Journal that the company was tricking investors and users into claiming to use artificial intelligence to develop applications.
In fact, this work was done by cheap programmers (https://www.wsj.com/articles/ai-startup-boom-raises-questions-of-exaggerated-tech-savvy-11565775004). In 2018, Engineer.ai raised $ 29.5 million. Among its investors: Deepcore Inc., a subsidiary of the Japanese conglomerate SoftBank, as well as Zurich venture company Lakestar (the first investor in Facebook and Airbnb) and Singaporean company Jungle Ventures. A company becomes more attractive to investors when it claims to use AI in its work.
Since the operation of this technology is difficult to track, experts cannot always determine whether it is really used in creating a product. However, the former and current employees of the company, as well as the documentation that came to the Wall Street Journal journalists, saying that Engineer.ai does not use AI to build application code — this is done by engineers from India. And there are many such cases around the world. The reason for all this is hype and money. According to PitchBook, venture capital firms almost doubled funding for AI startups in 2018 compared to 2017. To attract the attention of investors, it is enough to have only “ai” in the domain name of the company.
The story repeats itself with blockchain. Earlier, a report by London-based venture capital firm MMC Ventures showed that technology companies, even in Europe, which call themselves AI, startups, do not actually use artificial intelligence in their products. In total, there are 2,830 such companies — about 40% of all startups. Some novice technology developers use the fancy phrase “artificial intelligence” to draw attention to themselves and their products in order to get more funding. According to MMC, companies claiming to work on AI solutions attract an average of 15–50% more investment.
At the same time, startups themselves do not always declare the use of artificial intelligence. Therefore, the AI sector is a potential bubble that, due to hype, can burst, harming real market participants, not fakes. Another threat is that AI could potentially be used to wreck the same person. Distinguishing truth from fake is becoming increasingly difficult. Artificial intelligence masters the natural languages of human culture as fraud and propaganda.
In the summer of 2016, ZeroFOX, an information security company, discovered a new kind of Twitter bot called SNAP_R. It tricked users clicking on links, redirecting them to questionable sites. It served as an automated phishing system that analyzed the behavior patterns of users of a social network and found out their interests and needs.
At that moment, when an unsuspecting user flips through the news feed, the bot throws him some kind of entry like “Archaeologists have discovered the grave of Alexander the Great in the United States — for details, click on the link’’. SNAP_R did not pursue any malicious purpose, since it was only a working concept.
But the very fact of its existence warns us once again how careful we should be in the world of fake information, which is already being played by AI (https://www.blackhat.com/docs/us-16/materials/us-16-Seymour-Tully-Weaponizing-Data-Science-For-Social-Engineering-Automated-E2E-Spear-Phishing-On-Twitter-wp.pdf). At the same time, two researchers, thanks to SNAP_R, built a neural network that can learn from the analysis of large amounts of data.
For example, it learned to recognize images by analyzing thousands of other images. It was able to recognize spoken language by learning from a database of conversation records with technical support. And, of course, can already generate phishing messages by analyzing Twitter and Reddit posts and known cases of online attacks. The mathematical powers of AI are used today everywhere in many areas — from speech recognition to text translation. The same power can perfectly work to deceive thousands of Internet users. I think it will be strange if the technology is not used for fraudulent purposes. Everything indicates it already possible today.
Many technology experts have serious questions about the AI that Deepfakes generates — fabricated visual content that is very similar to real. (https://www.nytimes.com/2018/03/04/technology/fake-videos-deepfakes.html?module=inline) In addition, many other examples that have far from positive uses. Therefore, the dilemma of the relationship between AI and humanity depends on humanity itself and the goals for which AI will be used.
A big plus of a human in this matter is the fact that he can manage the whole process, yet. The main thing is to contribute to the creation of robots/algorithms that will bring only benefit to human existence, and not harm. An army of technology experts should start their work today and create all the necessary precautions. There are no serious studies on this issue; therefore, it is necessary to promote the emergence of special research institutes for the study of machine intelligence and life in the future. Is it possible to completely control all the processes of AI development, only time will tell.
Create your free account to unlock your custom reading experience.