Ethical issues IN AI use in hr management: russian experience

· The observance and reproduction of key ethical norms and values that relate to a particular culture or region;

· Reliability from a social point of view - implies the ethical safety of the use of AI.

Even after numerous studies, artificial intelligence systems can be harmful. To avoid this, when using artificial intelligence, you need to constantly monitor the system. Research Motivation: One of the promising developments in an HR sphere is the use of artificial intelligence that allows to optimise business processes. Therefore, in order to increase the effectiveness of businesses is directly linked to efficiency of HR management.

Research Question: How do HR managers in Russian companies understand the ethical problems when using AI systems and what measures are taken to solve them?

Research Gap: There are no studies on the example of Russian companies.

Hypothesis:

· HR managers of Russian companies face the problem of discrimination when using AI systems in HR business processes AI raises a number of ethical issues.

· HR managers of Russian companies when using AI systems in HR business processes face the problem of opacity of AI systems.

· HR managers of Russian companies when using AI systems in HR business processes face the problem of data confidentiality.

· HR managers of Russian companies believe that their specialty will replace AI in the future.

1. Theoretical foundation

1.1 Tasks in HR management

At the time of the advent of the digital economy, the world is witnessing trends in the changing role of human beings in the activities of organizations (Vasileva & Pestryakov, 2018). For Russia, as well as worldwide, human resources management improvement is one of the most pressing challenges of the economy's digitization. Researchers predict that there will be a high demand for highly specialized skills in the coming years, a shift of more staff to remote jobs, and a widespread adaptation of new information technologies for business processes (Rabota.ru report, 2019). For this reason, heads of organizations need to be aware of modern information and communication technologies (hereinafter referred to as ICT) and how they can be used to manage people.

Table 1. Human resource management includes many tasks, from recruiting to personnel policy

The main purpose of effective human resources management is to maximize the potential of employees for the benefit of the company and to create the necessary conditions to ensure the employees satisfaction. At the moment, the traditional principles of building human resource management systems are strongly influenced by the ever-growing market for information products and technologies. One promising technology that can change the face of personnel management is artificial intelligence (hereinafter AI).

1.2 Sources of artificial intelligence

There are many definitions of artificial intelligence. For example, the new Oxford Dictionary defines artificial intelligence as a computer system capable of performing tasks that normally require human intervention, such as speech recognition, translation, and decision-making (OED Online 2019). The Russian national strategy for the development of artificial intelligence adopted in 2019 and valid until 2030 defines artificial intelligence (AI) as "a set of technological solutions that mimic human cognitive functions (including self-learning and finding solutions without a predetermined algorithm) and allows you to achieve results at least comparable to the results of human intellectual activity" [Decree of the President of the Russian Federation No. 490]. The complex of technological solutions includes machine learning methods, processes and services for data processing and decision-making.

Doherty's definition emphasizes that artificial intelligence is not needed to replace human beings, but to expand human capabilities (Doherty, 2019). John McCarthy argues that the main purpose of artificial intelligence is to simulate human cognitive functions for model research to understand the nature of human intelligence and consciousness (Dushkin, 2019). By summarizing all these definitions, artificial intelligence is a software system for solving various tasks with the help of anthroposcale intelligence functioning on an automated basis.

1.3 History of artificial intelligence

In this part of the study, we will talk about the origins of artificial intelligence. The development of artificial intelligence was influenced by such factors as, for example, inventions in the field of electronics and engineering. Some early milestones include problem solving work, which included basic work in learning, knowledge representation, and inference, as well as demonstration programs in language understanding, translation, theorem proving, associative memory, and knowledge-based systems (Buchanan, 2006).

Since 1980, there has been a notable breakthrough in operations management research, including various processes that aim to improve efficiency and effectiveness in operations management. It is now common to hear that AI problems still remain unsolved, even though there is a lot of investment in the AI field to try to find solutions to these problems. The lack of development and progress is influenced by a number of reasons, for example, some solutions to problems that were formulated back in the 1980s, are designed for specific situations, or machines did not have enough data to accurately suggest solutions to problems.

In 1980, the American Artificial Intelligence Association was founded to offer services to artificial intelligence communities not only within the country, but also abroad. It is also necessary to remember that in the 1980s the world changed enormously. Enterprises began to enter international markets, which added a number of problems that were more difficult to solve compared to previous problems. Therefore, all the solutions developed in the 1980s had to be modified to solve really serious problems. In recent years, many conferences on this topic have also been held. These conferences provided the authors with an incentive to explore this area and identify progress, trends, and areas for future work (Meziane et al. 2000; Proudlove et al., 1998).

Today, AAAI also offers many different conferences to all interested communities in the field of artificial intelligence, such as the National Conference on Artificial Intelligence, Conference on Innovative Applications of Artificial Intelligence and Artificial Intelligence and Interactive Digital Entertainment Conference.

Artificial intelligence has a long history, some equate the history of artificial intelligence with the history of mankind. People didn't understand how the AI understood the world around them, much less how it controlled it. Historically, the first questions of artificial intelligence, which are related to the processes of thinking, were investigated by the science of philosophy.

The main questions that the ancient philosophers pondered were as follows:

1) Can formal rules be used to draw correct conclusions?

2) How does an ideal object like a thought come to be born in a physical object like a brain?

3) What is the origin of knowledge? (Gorbachevskaya, Krasnov, 2015).

Aristotle formulated the principles that guide the rational part of thinking. Raimundus Lullius argued that a mechanical device can conduct useful reasoning. Thomas Hobbes said that human reasoning and numerical calculations are similar.

Carnap in his book The Logical Structure of the World, defined a computational procedure for extracting knowledge from the results of elementary experiments. This can be considered the first theory of thinking as a computational process (Carnap, 2009).

Formal logic originated in ancient Greece (Lohin, 2001). The founder of formal logic is considered George Boole, he developed the logic of statements (Ivanov, 1998). Gottlob Frege is the creator of first-order logic, and Alfred Tarski is the first to show how to relate logical objects to real-world objects. Mathematicians have developed concepts such as logic, the theory of computation, and the third largest theory of probability. Gerolamo Cardano formulated the definition of probability, and Thomas Bayes proposed a rule for updating probabilities based on new facts. Scientific advances in mathematics and philosophy led to the creation of the first computing devices.

The nature of thinking has been studied for many years, but little has been achieved in practice. The reason is that the scientific research that allowed us to study physics and astronomy was not effective enough in the study of artificial intelligence systems. Artificial intelligence is a universal scientific field, because it can solve problems in any field of human activity. Alan Turing evaluated the effectiveness of intellectualization using a developed test. The test assumes that an object with artificial intelligence behaves like a person, and as a result, their behavior will be impossible to distinguish.

Thanks to the research of such scientists as Kohonen, Grossberg, and Anderson, a theoretical foundation was formed, on the basis of which it became possible to construct powerful multilayer networks. However, the problem was their training (Gorbachevskaya, Krasnov, 2015).

1.4 Types of artificial intelligence

When using an AI system, you can't say that they work reliably, because the indicators can be inaccurate, even if the program is functioning properly. Therefore, if you need to solve a critical problem, it is very important to duplicate the use of AI systems by other systems.

Thus, the future of intelligent control lies in the combination of traditional control with the potential capabilities of systems based on the use of artificial neural networks (Gorbachevskaya, Krasnov, 2015).

Modern researchers have identified two concepts in the definition of artificial intelligence: strong and narrow artificial intelligence. Narrow artificial intelligence is considered a narrow program based on clear algorithms designed to solve specific problems. Many researchers, defining artificial intelligence, describe narrow artificial intelligence, emphasizing that artificial intelligence is helping people solve problems. It is worth noting that these are simple algorithms that a specialist can always understand. It is quite simple to understand how AI works and on the basis of what decisions are made. Strong artificial intelligence, in turn, is called neural network algorithms, which are trained by themselves and we do not always know how and why this algorithm makes decisions.

The PwC report (2017) identifies three types of artificial intelligence. The first type, Assisted AI systems, support the human in decision- making or taking actions. Assisted AI systems demonstrate mechanical intelligence. The AI performs routine, repetitive tasks. Employees who use such systems are responsible for making their own decisions. These auxiliary AIS are usually applied to existing processes. For example, AI technologies that screen documents or transcribe speech into text faster and more efficiently than a human. Such applications can help companies decipher customer calls in large sizes, which helps to better identify customer needs, as well as evaluate employee productivity and efficiency (Microsoft 2019).

In the report (PwC 2017), the second type of AI is Augmented AI systems. Such systems complement the human decision-making process, and they can already learn from their interactions with humans and the environment. Thanks to analytical intelligence, AI can learn from data and process information to solve problems. In such situations, decisions are made jointly by humans and AI. These augmented artificial intelligence artifacts open up new opportunities for companies. For example, in the medical field, Topol (2019) documents all AI applications, which allows you to quickly scan patient data and provide doctors with an accurate interpretation of the data for accurate diagnosis.

The last third type is autonomous AI systems that can adapt themselves to different situations and thus make decisions independently, without the help and participation of humans. The human thus delegates decision-making to the AI. Autonomous artificial intelligence systems already show intuitive and empathic intelligence. Thanks to intuitive intelligence, AI can act creatively and more effectively in new situations, and thanks to empathic intelligence, AI can understand people's emotions and respond appropriately and influence people. For autonomous AI to work without human intervention, AI requires enhanced intuitive intelligence and more advanced empathic intelligence to cope with new situations and interact effectively with people. Autonomous AI applications are very often used in the service industry, such as chatbots, which provide fast direct customer support (Huang, Rust, 2018).

1.5 Using artificial intelligence systems in HR management

The use of multiple artificial intelligence technologies in an organization has a positive impact on a company's competitiveness and on the productivity of employees. Businesses built with a digital core outperform those with a traditional operating model. Digitally driven organizations are more profitable than their industry competitors (McKinsey).

There are 4 phases of AI implementation (Khoreva, Vaiman.2021). The first phase is the Assistant. The assistant sorts the data and helps you make decisions. In this case, the AI is only used to determine the choice that the employee is most likely to make and will offer this choice as a starting point when the employee is faced with multiple decisions. Phase 2 is the Monitor. The monitor sets up an artificial intelligence system to provide real-time feedback. In this case, Al can be trained to accurately predict what the user's decision will be in this situation. If the user is going to make a choice that does not match their history, the system can flag this discrepancy. The next phase 3 is Coaching. Modern employees want to receive regular feedback about their work. Artificial intelligence capabilities can easily generate feedback for employees, allowing them to look at their work and reflect on variations and mistakes. A monthly summary analyzing data gathered from their past behaviour can help them better understand their decision-making patterns and practices. The last 4 phase is the Teammate. The partner companies will develop a unified network of people and machines, in which both will contribute. As AI improves through interaction with individual users, organizations that fully integrate AI will organically develop a community of experts.

Artificial intelligence technologies have a significant impact on HRM (Human Resource Management), reducing the burden on employees from routine work and helping them to achieve their goals more quickly. Artificial intelligence technologies are now helping to solve several HR-related problems. One such problem is recruitment and selection. Most HR managers spend their time attracting candidates, reviewing resumes, conducting interviews, and advising candidates on various issues. However, more than 72% of companies have difficulty finding candidates with the necessary skills (Durcevic, 2020). Statistics show that companies spend an average of 14 to 63 days closing their vacancies, with each vacancy costing the company approximately $500. USA (Coyle, 2020). In summary, the selection of new staff members is a routine and costly task.

The use of bots for staff selection can help to reduce costs and reduce the time needed to find new staff. One of the advantages of using bots to communicate with candidates is the ability to customize the interactive assistant's speech style according to the norms accepted in the culture of the organization (Altemeyer, 2019). For example, if an organization has a democratic culture and creative atmosphere, the bot can be programmed to have an informal and playful conversation with candidates and staff. This approach allows for closer contact with the candidates and gives the company an advantage in the fight for the necessary new employees (Gikupulous, 2019). Artificial intelligence technologies are also used to conduct video and routine interviews. One of the methods used in staff selection is the use of face and emotion recognition algorithms. The algorithms process the video and determine whether the person is looking at the camera or looking away and reading the text. The system analyzes the position of the eyes and the direction in which the applicant looks. It can also make notes if it is suggested that a person is potentially deceiving. Examples of such HR systems are HireVue, Skillaz.

In addition, artificial intelligence technologies can help to improve employees skills. Recently, the Udemy Online Education Platform conducted independent research that revealed that 84% of surveyed users felt that they lacked certain skills needed for their profession (Skills Gap Report, 2018). It is worth noting that this figure is increasing every year. The lack of technical skills, leadership and management skills, interpersonal skills and other so-called soft skills were the main points of concern. According to Russian company executives, current employees and applicants have on average better communication skills, creativity and flexibility than employers need. However, staff and applicants often lack sufficient knowledge of the subject area, as well as the necessary leadership and organizational skills (The Global Skills Gapinthe 21^st Century: report QS, 2020).

According to the Delloite study, more than 75% of employers in Russia and the world prefer to train current employees in the necessary skills rather than to seek new candidates for these tasks (Deloitte, Trends in human resources management in Russia, 2019). At the same time, more than 40 per cent of Russian employers are forced to hire new employees because of the slow pace of training of current employees. In addition, expenditures in the global learning and development sector in 2017 amounted to more than US$ 200 billion (Bersin, 2019).

However, about half of these funds were spent without the expected return. Despite the current negative statistics, staff training is considered one of the best ways to invest in the organization's current and long-term success (Cherer, 2017). Artificial intelligence technologies used in voice assistants can be useful in employee training. For example, a voice assistant can listen to calls made by a sales department and then can suggest ways to improve sales performance. The Voice Assistant can also listen to managers' conversations with subordinates and give them advice on how to develop communication and management skills. It is worth noting that, while the voice assistant is a clear example of how AI technologies can be useful for HR development, there are legal and ethical limitations to the use of this technology. However, there are other options for training staff using AI technologies. The AI-solutions can track the performance of individual company employees by comparing them to the most successful professionals in the industry, and then, based on the analysis of collected data, AI algorithms can develop a personalized training program for employees (Nigel & Sheri, 2018). Chat bots can also be used to train company employees. Chat bots help new employees to adapt much more quickly in the company, as bots are available 24 hours a day and can answer many questions about working in the company. One of the most important areas in which bots can help employers is increasing the speed at which new information technologies are introduced and mastered. Chat bots are intuitive and easy to use tools and greatly facilitate the transition to new technologies. An example of a chat bot for learning is WalkMe, which helps employees learn new technologies. The bot has no software interface and uses a chat interface in natural language. The system uses a combination of artificial intelligence and analytics to predict user behavior and offer step-by-step help in mastering new technologies. Using WalkMe helps users perform tasks without any training, even if they move between different software tools.

As stated earlier, the main challenge in managing human resources is to enable the company to derive maximum benefit from the skills of the employees. This requires not only the loyalty of the company's employees, but also their high level of engagement. The engaged employees are interested in the company success and are ready to invest their efforts and time to achieve common goals. Gallup found that a high level of employee involvement leads to a reduction in staff turnover, absenteeism and other misconduct by staff members (Practicum Group, accessed 11.04.2020). A study by Aon Hewitt found a correlation between staff engagement and company profit (Practicum Group, accessed 11.04.2020). It was found that increasing employee engagement by 1% increases company profits by $20 million, increasing employee engagement by 5% and 10% increases company profits by $100 million and $200 million respectively. According to Gallup, only 15% of employees worldwide are engaged, and in Eastern Europe only 10% of employees are engaged (Jouany & Makipaa, 2020). The results of the studies show that the management of engagement holds great promise for the organizations for further growth and requires new approaches to address existing personnel management challenges.

In order to increase the level of engagement of employees, as well as the efficiency of their work, various methods of analysis and control employees engagement are used. The concept of employee's mood analysis is quite simple: large amounts of textual information need to be analysed and divided into categories by engagement or relation. Understanding the mood of employees throughout the company, as well as by division or location, helps to create better working conditions for employees. This is possible by combining natural language processing and machine learning. Algorithms work with unstructured information obtained from employee conversations, emails and other incoming information.

In addition, machine learning is used to teach artificial intelligence words that are related to current or future problems (Gulliford & Dixon , 2019). One example of a platform that uses AI technology to analyze the mood of employees is Ultimate Software's Xander AI solution. The goal of system developers is to combine analytics with emotional intelligence, which helps to get a fairly accurate picture of what is happening in an organization. The system receives data from comments on the company's intranet, as well as feedback from staff interviews. The system is trained to the extent that it is able to draw conclusions about the mood of both individual employees and all company personnel. In addition, AI technologies can facilitate interaction between staff. While e-mail is one of the most used tools for staff interaction, there is no substitute for personal communication. Verbal communication has always been, and will continue to be, the fastest and most effective way to interact, but at present one of the participants in the interaction can be an AI, not a person. Many people already use voice applications on mobile devices. These technologies can also be useful for HRM. In addition to the above, some companies use artificial intelligence technology to track the arrival of employees at work. This helps the company not only to track time and analyze attendance records, but also to ensure the security of the office.

1.6 The pace of introduction of artificial intelligence

In 2019, Gartner published a study that reported that only 17% of organizations use AI-based solutions to perform HR functions, with another 30% expected to do so by 2022 (Baker, 2020). The artificial intelligence market in Russia is growing rapidly. Companies have mastered the first, pilot versions using AI and are now looking for new ways to use and develop the technology. For example, in the HR sector, this is the use of voice assistants (chatbots) and other things that are mentioned in our work. Researchers believe that in 2020, artificial intelligence has entered a new era.

Artificial intelligence will bring greater success to those industries where the integration of AI is economically justified, for example, in the field of HR. Based on the statistics provided by Farid, voice and speech recognition remains the best application, based on total combined revenue of $ 38.8 billion from 2018 to 2025 and annual revenue of $ 8.8 billion in 2025, key factors that contribute to success include the widespread adoption of digital assistants (Nigmatullin, 2020).

After the pandemic, IDC research company made a forecast for IDC FutureScape: Worldwide Digital Transformation 2021, in which it described that investment in artificial intelligence is growing and will continue to grow annually at a rate of 15.5% (CAGR) until 2023 (www.idc.com ).

Artificial intelligence technologies are becoming the key to solving a variety of problems. Since 2020, the Russian Government has been paying close attention to the development of artificial intelligence. On the official website of the government of the Russian Federation, you can see the national program, the total budget of which from 2019 to 2024 will reach 1.8 trillion rubles. The number of organizations that are starting to use AI has increased by 19% over the past two years.

AI technology has great potential. According to a study conducted by PricewaterhouseCoopers, artificial intelligence can add more than $ 15.7 trillion to global GDP by 2030, or 14% to its current level (Rao, Verweij, 2017). At the beginning of the 21st century, research centers began to be created around the world that are engaged in the development of AI technology. Now, many digital companies, such as Google, Apple, and Amazon, are investing hundreds of millions of dollars in AI development. Even the President of the Russian Federation at the All-Russian Open Lesson said that the country that will become a leader in the creation of artificial intelligence will be "the ruler of the world" (Ria Novosti).

The largest number of investments in this area falls on the United States ($511,089 million), followed by China ($451,201. 4 million), followed by Japan ($168,644. 9 million) and Germany ($118,158. 5 million). Russia ranks ninth in this list with a large gap ($42,270. 9 million). This is due to the fact that in Russia the total number of mutual funds at the end of 2017 was 1504 with a total amount of assets of 3036.2 billion rubles (CENTRAL BANK OF THE RUSSIAN FEDERATION). If we consider the structure of R & D investments, the share of public financing in Russia will be 68.8%, while the share of the private sector will account for only 27.2% of the total amount (Gorodnikova, Gokhberg, Ditkovsky, 2019).

Russian companies do not have enough funds, so Russian companies conduct their own internal research, and they prefer not to disclose the data obtained about the company, so as not to reveal the sources of their advantage to competitors. Fact that the Russian AI market is just beginning its development. And today it is significantly behind the foreign markets, but due to the serious mood, it is significantly ahead of them in terms of growth rates.

Russia should make the patenting service for companies developing in the field of artificial intelligence more affordable in financial terms. But the problem of accessibility has recently begun to be solved. For example, the Skolkovo Innovation Center offers reimbursement of costs to companies that have received the status of participants of the Skolkovo Foundation in the amount of 75% of the total amount of this service.

Another reason that Russia does not take a leading position in the field of AI development is the lack of qualified specialists in the field of artificial intelligence and big data. Therefore, good specialists in the market are dismantled immediately, offering them an unrealistic salary. Now the number of investments in the field of artificial intelligence development exceeds hundreds of billions of dollars in all countries of the world.

If decision makers see transparency in the results of artificial intelligence, they have more confidence in the final decisions that AI systems make. It is also important to add human control at the beginning of the work when making decisions, then over time this will lead to the recognition of AI. Artificial intelligence is increasingly penetrating our lives, becoming a normal phenomenon and gradually replacing humans in a number of professional fields (Akulinin, Adamov, 2019).

AI technology has great potential. According to a study conducted by PricewaterhouseCoopers, artificial intelligence can add more than $ 15.7 trillion to global GDP by 2030, or 14% to its current level ().

Thus, that companies face various barriers to the introduction of AI technologies into HRM, such as seeking financing. It is sometimes difficult to quantify the benefits of implementing AI in HRM, and to do so it is necessary to test the validity of AI investments and try to calculate the possible results before introducing AI technologies. Higher priority should be given to projects that address critical issues such as improved decision-making, faster staff performance or more effective management. ROI can be used to quantify investments in the AI. Management needs to see the link between AI applications and the business results that they generate. There is a need to assess the relationship between the results of using AI solutions to manage human resources with HR metrics that affect financial performance. A good example of the effectiveness of such evaluation is the AI system for searching candidates implemented by IBM. Their decision contributed to a significant increase in the number of candidates for the company. This application allows for faster processing of a large number of candidates' profiles, which has increased the speed of recruitment as well as the quality of candidates. As a result, IBM reduced HRM spending by $107 million in 2017 (Guenoli & Feinzig 2020).

1.7 Ethical issues in the use of artificial intelligence

Ethical issues are another challenge in the adoption of AI technologies. For example, in the field of education, when using AI, there are ethical risks when applied arbitrarily, so there is a discriminatory potential in them.

1. Violation of the free will of a person and their potential for self-development (when using technologies that take a decision and make a choice of employees or select students).

2. Replacing teachers with AI.

In the context of the rapid development of digital technologies, especially in education, the integral component of educational institutions is increasingly developing in the direction of decentralization (Coppin, 2004). This all leads to the fact that people begin to move away and abandon strict educational standards. For example, people choose short courses, self-study, or training courses, replacing them with full-fledged training that should last a whole semester, a module, or several years. It is very important to combine online and offline platforms when launching courses, because artificial intelligence technologies are mainly used in this direction. Every year (also in the last year, the COVID-19 pandemic contributed to this), the obligation to actually attend classes loses its uniqueness. In the process of evolution and during this period, people have rethought many things and began to switch to online courses. With the transition to online learning, the attitude to education also changes, teachers can not also influence and control specific children and students, through online learning, the teacher can influence the group as a whole. Consultants who have had jobs are replaced by online consultants, and they in turn are bots. Many studies indicate that AI will soon replace tutors and teachers. According to a PwC survey, 58% of managers and technology experts believe that by 2022, AI will be able to replace tutors in the future (Rahwan, 2018). This can all end badly, because the main point of a teacher is not only to teach children, but also to instill the right values in the process.

3. Gamification of education.

The gamification of education is the fusion of education and entertainment. Game components are included in the educational process thanks to AI, virtual reality and the game environment. For example, a student takes a test accompanied by virtual effects. Surprisingly, this technology has very good indicators, students learn the material easier and faster, and approach learning with pleasure, so this technology of education has support on the market.

But from an ethical point of view, this approach to education is completely inappropriate. First, it is hardly the norm that with the help of such a learning technology, the boundaries between entertainment and learning are blurred in children. Aspects of gamification are characterized by a high potential for tolerance formation (Feil-Seifer, Mataric, 2010). Such methods of mastering the material can not replace the conscious processes of finding solutions, in which the student consciously makes efforts. Also, this technology can be dangerous because over time, the student will not be able to distinguish the game component from the content and will most likely become addicted to games.

The most terrible thing for any person is the psychological consequences. The introduction of virtual assistants will greatly affect a person, to the point that a person will forget how to communicate with real people and solve life problems in real time.

4. Advisory programs.

Such programs can also have an impact on the quality of education. These programs are necessary in order to generate personal training tactics, help you choose a profession, and advise you on which qualifications are best to acquire (which is the most popular). In our time, many people have come across tests that help to choose the most suitable profession for each person, through analysis. AI programs perform these functions and are already located in various career guidance centers. According to experts, in the future, people will be trained on specially selected training cards, which will be specially designed for each person. This method can help determine a person's innate talents.

If you look from an ethical point of view, such programs can demotivate the test takers. For example, the test taker has been playing sports all his life and he was doing great, but the test showed him that he has more mathematical abilities, not sports, and a person can give up his hobby in favor of solving a machine. Due to the high degree of trust in the expert level of AI programs, test takers may perceive the program's recommendations relatively uncritically (Hakli, Makela, 2019). In fact, the question of the ratio of innate abilities and their development in the course of training has not yet been solved from a scientific point of view. After testing, many people can give up the professions they dreamed of working in and choose professions on the advice of the program. Accordingly, for all test takers, this test may end in different ways, and for some with psychological trauma. The problem is that the student may stop developing in the direction in which he has developed for many years and start moving in another direction. But it is not possible to check the effectiveness of the program's forecasts at this point in time. When implementing advisory programs, there is a risk of discrimination. For example, the system may recommend work by gender or by dividing people into races.

5. A set of interactive educational programs.

This technology implies that when preparing a student, the level of requirements for him must correspond to the level of his capabilities. This means that the system can adapt to the level of training of any student and offers him tasks that he can perform and further material for study.

Training in such programs is successful, as research shows. The programs lead the students themselves into some confusion about their true success in the learning process (Bankins, Formisa, 2019). However, there are ethical problems here. Even despite the successful development of a particular program, the applicant may not find confirmation in his further training or in his career. This circumstance gives rise to ethical disputes about whether the systems of individual assessment of the success of training mislead those who are trained according to them (Danaher, 2019).

In the field of health care, there are also several ethical problems.

1. Automated collection of data on the state of human health and calculation of the probability of occurrence of diseases in patients.

It is assumed that forecasts will be introduced, which will be formed by artificial intelligence, these forecasts will calculate the probability of a particular disease in the patient. The forecasts will be stored in electronic charts and it is assumed that this method of data collection will facilitate the work of doctors. AI systems in this situation are even worth using, because doctors are human and they are also affected by the psychological aspect. For example, doctors usually use standard treatment options for patients, recall cases from their practice when someone was treated and the person was cured. At this time, the AI programs know all the cases, both with a bad and a good outcome, but due to the fact that the machines are not affected by the psychological factor, they select the treatment that really suits the patient at the moment.

But there are ethical concerns about this technology. Many believe that if the AI assessment data on the patient's condition is stored in an electronic medical record, then employers, banks, and acquaintances may be aware of human problems and this is already a violation of the personal zones of human life. For example, if an employer finds out that an employee begins to develop a terrible disease, he can fire him or terminate his contract with him, because of this, the person will lose his job. Everyone knows that in the course of treatment, the patient's medical record is not only with the doctor, but also with the administrators. If a familiar person receives information about the patient's medical history, he can learn about all genetic diseases, infections, even about abortions, because of this, the patient may begin to have undesirable consequences. There can be many options for information leakage and eventually it can get into open access, and in the future it will be transmitted to the place of work, if a person decides to take a loan from a bank. In this situation, we can distinguish discrimination, because if the AI program diagnoses a person with a stroke within 6 years , he may lose the deal or he will not be hired because of this ifnormation, because everyone hires candidates in the long term. Too accurate AI predictions can lead to serious ethical dilemmas. If we consider the relationship between people, it can be noted that if a girl in the prognosis of the disease is revealed that during pregnancy she will have a pathology, this may affect their future relationship with a young man or husband. The situation is made worse because AI forecasts can make a fundamentally erroneous forecast. Consequently, patients will suffer because of discrimination and because of false information.