Run, leather bastards! We analyze myths and horror stories about AI and robots

Most authors of articles and posts about artificial intelligence look at AI from two opposite angles: some talk about AI as a new miracle technology that can solve almost any problem of humanity. Others paint a grim prospect of the imminent enslavement of humans by the sinister Artificial Intelligence and its robotic servants. In general, the topic is controversial, and we have long wanted to bring a little clarity and adequacy to it. First, let’s look at more or less general myths and horror stories about AI, and then we’ll move on to technological misconceptions and figure out together what AI is really capable of and what can be classified as fables.

Myths and horror stories about AI

1. AI will take away people's jobs and professions.

   This is probably the most common myth about AI and the one that scares most people the most. And they can be understood: the fact that very soon (allegedly) a technology may appear that will take away your job and profession should really cause concern.

   In reality, everything is quite complicated, but not at all as gloomy as you might think. The introduction of AI technologies and automation can, and inevitably will, affect the economy, leaving many people without work. First of all, of course, workers who perform the most basic functions that are relatively easy to automate will be laid off.

   But this does not mean that very soon all the functions of people will be taken over by AI, and only those who work directly on maintaining the functionality of the AI ​​itself will remain in demand. And even then, only until the future Skynet gains strength for complete autonomy, or turns people into living batteries connected to virtual reality.

   In fact, past industrial revolutions, although they led to major changes in the labor market, putting many people out of work, caused an increase in the number of jobs rather than a reduction in them. Therefore, real facts and statistics do not particularly support the gloomy forecasts; rather, on the contrary, we have every reason to expect that the development of new technologies will spur economic growth even more and lead to the emergence of a whole range of new professions.

2. Super-smart AI will instantly eliminate many jobs that do not require very high qualifications.

   As mentioned above, AI, like any new technology in principle, will gradually put some people out of work. So those whose job it is to perform simple functions do have some cause for concern.

   However, this is nothing unusual. New technologies have always led to changes in economic models, leaving entire professions behind. For example, in the first decades of telephone communication, subscribers were switched to each other by operators, who were no longer needed after the advent of automatic switching. Technologies like voice messaging and email have greatly reduced the need for office secretaries. And so on.

   Today, chatbots and artificial intelligence-based virtual assistants are beginning to seriously threaten the employment of many people working in call centers, for example.

   There is only one key lesson here - the development of technology leads to the elimination of the need only for those people who perform the most basic functions. To prevent a new brainy chatbot from depriving you of your job, you need to learn and develop complex skills that require thinking and creativity, that is, what still remains the exclusive prerogative of homo sapiens (and even then, not all).

3. Universal AI will be able to do any job.

   Another myth, again related to employment, tells us that quite soon some kind of universal AI will appear that will be able to do almost any job equally well.

   In reality, if specialized AI technologies really soon replace people in some positions, the emergence of a universal supercomputer that can perform a variety of functions with equal efficiency, from technical support for buyers of sex dolls to managing a nuclear arsenal, cannot be expected in the coming decades . Why?

4. AI will quickly surpass any human in terms of intelligence.

   Because intelligence is not something linear and unambiguous. Yes, in some areas (such as computing speed or memory capacity) computers are already noticeably superior to humans, while in others - such as creativity, strategic thinking and emotional intelligence - they are still not even close to humans, and barely Will the situation change radically in the near future?

   “We project onto AI what we would do ourselves. But the smartest computer on Earth is still the human brain, and it is very unlikely that we will be able to build AI that is more versatile and powerful than our brains. AI does not have consciousness, and therefore is unlikely to become smarter than us. AI cannot exist without people,” says David McCall, vice president of innovation at data center operator QTS.

5. The introduction of AI will lead to the enslavement of humanity by robots, or the complete destruction of people.

   This, in fact, is the basic scenario and the end result of the development of artificial intelligence technologies according to those who consider AI a threat. And opinions that AI can still be a threat are spread not only by people very far from this area, but also by such well-known and authoritative personalities in the world of technology and business as Elon Musk (who, as is known, has repeatedly stated that he believes artificial intelligence and robots are a real danger).

   In reality, although no one can accurately predict the distant future, there is no doubt that in order to potentially bring such a scenario to life, humanity must go a very long way, not only managing to create AI that is conscious and surpasses human intelligence, but also ensuring that the future robotic overlord has every opportunity to potentially seize power.

AI: technical misconceptions

Now let’s look a little deeper: let’s go through the myths associated with the technologies underlying AI, and misconceptions about the real capabilities of artificial intelligence today and in the near future.

• Machine learning based on neural networks means that AI can learn just like humans.

  In fact, neural networks are quite a powerful and functional technology, but they do not come close to the complexity of the human brain. And they are not able to learn in the same way as a person does.

  To simplify a little, all deep learning AI based on so-called “artificial neural networks” can currently do is emulate the way biological neurons in our brains can recognize and remember patterns. This ability is already widely used, and allows solving many different problems, from improving the quality of translation and electronic speech recognition, to identifying financial fraud and efficient and safe computer control of a car.

  But artificial neural networks are not even close to the human brain. Each person's brain contains more than 200 billion neurons, each of which connects to more than 10 thousand other neurons through synapses, each of which contains up to 1 thousand molecular switches. In addition, there are about a hundred different neurotransmitters that influence the way neurons communicate. It is believed that one human brain contains more switches (the active components of each process) than all the computers, routers and Internet connections on Earth combined.

• Machine learning-based AI can learn autonomously.

  There is an opinion that an AI project based on machine learning can learn on its own, and only requires a basic set of direction and strategy from a person. Actually this is not true. For a system to work effectively, it usually requires a team of experienced specialists who identify a problem for AI, prepare the data on which the system learns, set standards for datasets, and maintain the software in working order, updating and improving it.

• “Cognitive AI” technologies are capable of understanding and solving various problems.

  According to another common misconception, cognitive AI based on neural networks is already a universal solution that can be used to solve various kinds of problems. In reality, current AI systems fail to solve problems they were not designed to solve. Just as they are not capable of making decisions in cases where the answer is not one hundred percent obvious.

• AI is capable of being 100% objective.

  All AI technologies are in one way or another based on data and the rules for their interpretation set by the system developers. And since all people are biased in their own way, AI will always be so. At the moment, there is no technology that truly works to completely eliminate any bias. Although work in this direction is underway.

• The AI ​​algorithms are closed, and it is impossible to understand how it makes decisions.

  Since AI was initially positioned as something mysterious and closed, as a technology that generates results without the user being initiated into the algorithm of its work, many people see AI systems as such. However, as time goes on, more and more systems based on AI technologies with transparent algorithms and open source code appear.

Instead of an epilogue: Polarization or how trends are created

As a conclusion, I would like to reflect a little on the topic cui prodest (who benefits): why they talk so much and so persistently about the dangers of AI, although it would seem that the reality is obvious - modern AI and the systems based on it are certainly a promising technology that will find widespread use. application in many areas and will leave a considerable number of people without work. But the level of development of AI technologies is still too far from causing serious concern, and nothing foretells dramatic revolutionary changes in the near future. So why are tech industry insiders like Elon Musk talking about the dangers? Here's a guess (you can argue with it in the comments): such experienced businessmen who know how to market products and ideas are using polarization to attract attention and investment to AI-based technologies. The way polarization works is to create a “hot” and discussed topic by broadcasting powerful polarizing messages to the masses. In our case, these are warnings about the dangers of AI on the one hand and advertising of AI platforms from numerous IT companies and startups on the other. At the same time, both sides present AI as something very functional and powerful enough to cause concern. Whereas in reality, artificial intelligence is nothing more than the next round of software evolution. Do you agree? We express our opinions in the comments.