Ogirko, Ihor; Shapenko, Liudmyla; Zolotar, Olha (2025). AI AND DATA ANALYTICS IN PUBLIC ADMINISTRATION: TOWARD RESPONSIBLE AND SECURE IMPLEMENTATION. Social and Human Sciences. Polish-Ukrainian scientific journal (https://issn2391-4165.webnode.com.ua/), 01 (45).

 AI AND DATA ANALYTICS IN PUBLIC ADMINISTRATION: TOWARD RESPONSIBLE AND SECURE IMPLEMENTATION

 УДК 378:37+ 340.1+342.951(007.5)                                       

Ogirko, Ihor, Professor, Doctor of Physical and Mathematical Sciences, Institute of Printing and Media Technologies, Lviv Polytechnic National University (Ukraine, Lviv), ORCID: https://orcid.org/0000-0003-1651-3612 ;

Shapenko, Liudmyla, Ph.D (Law), Associate Professor, Kyiv National Aviation Institute (Ukraine, Kyiv), ORCID: https://orcid.org/0000-0001-7351-641X ;

Zolotar, Olha, Doctor of Law Science,  head of the scientific laboratory, Scientific Research Institute of Informatics and Law at the National Academy of Legal Sciences of Ukraine (Ukraine, Kyiv), ORCID: https://orcid.org/0000-0002-2461-6745

 SUMMERY

        The article examines the peculiarities of artificial intelligence (AI) implementation in the conditions of digital transformation of society and the development of new technologies as one of the of the key tools for improving the efficiency of management, decision-making and decision-making and process automation. The author emphasizes the importance of AI integration integration to optimize the quality and efficiency of management decisions. The study identifies identifies the main areas of AI application, in particular: big data analysis for trend forecasting, process automation, implementation of intelligent decision support systems, and the use of chatbots. The research also includes analysis of the specifics of AI implementation, taking into account the state of digital transformation in the country and the current legislative framework. It is noted that an important role is played by is the development of a legal framework for regulating the use of AI in interaction with the latest technologies. At the same time, the article analyzes the implementation of AI and and the problems of artificial intelligence development are investigated. It is emphasized that effective implementation of AI in Ukraine is possible only if it is in line with international standards.

The practical possibilities of artificial intelligence that can be effectively used in modern society are investigated. that can be effectively used in modern society. The main areas in which artificial intelligence is used are analyzed in which artificial intelligence is used are analyzed. It is emphasized that the issue of using of artificial intelligence technologies in organizational development should be considered in two ways: on the one hand, the organization should be prepared for the introduction of technological innovations, and on the other hand, AI-related solutions must meet the requirements of organizational development of organizational development and meet its needs.

The article describes the concept of using AI to solve problems and improve organizational development processes, the main purpose of which is to propose the use of certain technologies and methods. Among the obstacles to accelerating the formation of Society 5.0 in Ukraine are Russia's armed aggression against Ukraine, which is qualified as a war of aggression and has large-scale global consequences, imperfect government institutions, insufficient efficiency of lawmaking, technological lag, low qualification of human resources, and weak support for innovative changes in society.

For society to accept innovative changes, it is necessary to pay attention to the development of education and science, and to support the formation of digital entrepreneurship. Today, the organization of governance is largely dependent on the information economy, which is developing thanks to digital technologies. The characteristics of the new information environment include the increasing role of information and knowledge in society, which are becoming key factors in economic development. Thus, with the advent of technologies such as the telephone, computer, Internet, and artificial intelligence, marketing has changed significantly.

The study analyzes the legal issues of digitalization in the environmental sphere in relation to the relevant processes in the economic sphere. The study finds that the digitalization of environmental protection requires solving institutional, infrastructural, ecosystem and automation problems, with the legal factor playing a crucial role. The factors of digitalization should be taken into account not only in strategic, forecasting and program documents, but also in legal documents, in particular in regulations.

At the same time, while emphasizing the benefits of digitalization, we should not neglect its possible negative consequences, especially in the field of ecology.

Keywords: artificial intelligence, AI, administrative processes, development problems, technologies, resources, management, development, data mining, automation.

 

 

Problem statement. In the modern context of technological transformation, artificial intelligence acts as a tool that can significantly increase the efficiency of management processes. The relevance of this study lies in the need to determine the essence and specifics of the integration of artificial intelligence into the public administration system, as well as to analyze its potential to ensure the effectiveness of management decisions.

Analysis of recent research and publications. Artificial intelligence issues are considered in scientific research by specialists in various fields, including management, law, economics, and other fields. The essence of artificial intelligence, its mechanisms, characteristic features and classification are studied by the following Ukrainian scientists [1; 2; 3; 4; 5; 6; 7; 8; 9; 10; 11], how A. Kolesnikova, O. Karapetian, A. Pohorelenko, T. Yarovoho, L. Zhyvtsovoiu, O. Skitsko, P. Skladannyi, R. Shyrshov, M. Humeniuk, M. Vorokhob, R. Kvasnytska, N. Skopenko, S. Tsyhanov, O. Petriv, Yu. Karpenko, V. Vitlinskyi, D. Pchelianskyi, S. Voinova, N. Reznikova. Among foreign scientists who pay considerable attention to the study of AI problems, it should be noted Work А. Vitforda, L. Anastasopulosa, S. Asha, T. Shanafelta, B. Virtsa, Kh. Esha, Kh. Mekhra, S. Lina, and others. Despite the significant developments of scientists in the context of this problem, the key topics of serious scientific discussions remain the study of a number of challenges related to the implementation of AI in the field of management.

The purpose of the article. The main idea of the article is to analyze the problems and challenges associated with the development and application of artificial intelligence technologies in managerial activities.

Presentation of the main material.

Artificial intelligence is a branch of computer science that deals with the formalization of problems and tasks that mimic actions that are inherent in humans. Nowadays, AI is an integral part of human life: it helps to study, work, and solve extremely complex problems in various fields. Modern machines are able to recognize speech, have technical vision, which allows you to accurately determine a person's age, gender, emotions, identify objects, and more. AI is widely used in various sectors of modern society, including finance, healthcare, education, transport, weapons and military equipment. It is used for medical diagnostics, e-commerce, remote control, and Earth sensing.

Today, AI is able to influence the information space in the largest forms. Therefore, its implementation and use must be socially oriented and meet the interests of human security, as well as the protection of his personal space.

In today's society, one of the biggest threats to AI is its impact on security. This prompted the international organization UNESCO to develop a standard of ethics for artificial intelligence, which was adopted by 193 member states of this organization. This approach is aimed at preventing the potentially negative impact of AI on the processes taking place in the world community. The UNESCO Ethics Standard defines the basic principles, including: protection of human rights, freedoms and dignity; promoting peace, justice and inclusion; conservation of diversity and protection of ecosystems.

AI technologies are developing and implementing extremely quickly. There are more than 2000 IT companies in the country engaged in software development and specializing in the development of AI. At the same time, Ukraine has made significant progress in implementing an open data access policy.

As of 2023, according to the Open Data Index, Ukraine ranks 31st in the world [1; 2; 3; 4]. Despite the fact that Ukraine has not yet achieved a leading position in the field of artificial intelligence, it has every chance to do so. Systemic support for the development of artificial intelligence, which provides for an increase in public and private funding for research, the creation of a favorable investment climate, and support for innovative startups, can contribute to a significant strengthening of Ukraine's position in the global AI market and positive changes in this area [2]. The National Academy of Sciences of Ukraine, the Ministry of Digital Transformation of Ukraine, the Ministry of Education and Science of Ukraine, the Ministry of Strategic Industries of Ukraine and other government organizations of Ukraine are involved in the process of creating the National Strategy for the Development of Artificial Intelligence in Ukraine [3; 4; 5; 6]. In Ukraine, practical experience in the use and implementation of artificial intelligence is intensively accumulated in order to improve the quality of services and solve current problems [1; 2; 3; 4]. Thus, together with citizens and the Kyiv City State Administration, artificial intelligence developers are working on the creation of urban digital services based on artificial intelligence technologies in order to develop smart infrastructure in the capital. The Kyiv Artificial Intelligence project team conducts research on ethics and discrimination in the development of AI-based services in various fields, as well as initiates discussions on the protection of digital rights of Ukrainians. The Ministry of Digital Transformation of Ukraine is developing an educational series on artificial intelligence, which explains the key characteristics and problems associated with the creation and application of automated systems. Some of the lectures are aimed at a younger audience, while online courses are addressed to industry specialists. In particular, the programs pay attention to the issues of observance of human rights in the process of development of modern technologies. Within the framework of the educational program, it is also worth noting the announcement of further cooperation between the Ministry of Digital Transformation of Ukraine and Google Ukraine. The companies agreed to create courses on the use of AI for business development, public service and personal branding [2; 3; 4; 5; 6; 7]. However, there is a lack of appropriate legal standards governing the development and implementation of such technologies in both the public and private sectors.

Artificial intelligence is a branch of computer science that is concerned with creating machines capable of performing tasks that would normally require the use of human intelligence. This includes tasks such as language comprehension, pattern recognition, learning, and planning. Of particular interest is the section on learning models, where the term "machine learning" (ML) is the main concept. Machine learning is a subfield of artificial intelligence, which focuses on developing algorithms capable of learning from data without the need for additional programming. Machine learning systems analyze large amounts of structured and unstructured data to identify patterns and improve their performance. In machine learning, models use statistical methods to make predictions or make decisions based on the analyzed data. The areas of application of machine learning cover a wide range of fields: from automatic translation of languages and recommendation systems to pattern recognition and autonomous driving. Thus, machine learning is one of the subfields of artificial intelligence. Trained models are capable of making predictions. One of the areas of machine learning is deep learning. These concepts are similar and are often used interchangeably, although there are some conceptual differences between them.

Deep learning is a subset of machine learning that is based on the use of complex models called neural networks. These models are able to learn and improve independently by analyzing large amounts of data (Big Data).

The concept of neural networks or neural networks is key to further research. Neural networks are a class of machine learning algorithms that mimic the structure of the brain using layers of interconnected nodes, or "neurons". They are the basis for deep learning, which allows systems to learn and analyze on very complex data.

The Data Science & Intelligent Systems program includes the following areas:

1.                      Machine learning and artificial intelligence.

2.                      Data analytics.

3.                      Internet of Things.

4.                      Smart solutions and systems.

To obtain a qualitative forecast, you need a large amount of data and a data markup mechanism that provides structuring, categorization, and annotation of data. First, the amount of data matters. Big Data is a term describing vast, diverse, and fast-changing amounts of data collected from a variety of sources, including social media, sensors, mobile devices, or e-commerce, and can be used to analyze and solve complex problems in areas such as business, medicine, science, government, and technology. Big Data is understood as data that has a large number of records (millions, billions, or more). With the help of artificial intelligence algorithms, the system analyzes data on aid recipients and identifies potential violations and misuse of resources.

However, the use of such systems carries significant risks. In particular, the creation of profiles of vulnerable populations can lead to discrimination and human rights violations. A change in the political situation or aggravation of social problems can lead to the fact that individuals will be on the lists of persons subject to prosecution due to unlawful data processing. In addition, there is a risk of violating European data protection standards and non-discrimination. Similar problems arise in other areas, in particular in the field of healthcare, lending, employment, where automated decision-making systems can create additional threats. For example, the Ministry of Justice of Ukraine uses artificial intelligence to assess the risk of recurrence of criminals, and the National Agency for the Prevention of Corruption (NACP) uses automated verification of declarations. Although these systems are designed to increase the efficiency of state bodies, they can also lead to human rights violations and increased social inequality [7; 8; 9; 10].

Thus, banks can use the personal profile of customers solely for the purpose of concluding agreements on opening accounts, granting loans or purchasing securities, that is, the list of potential risks is exhaustive and understandable. At the same time, providing the state with profiles of vulnerable and marginalized groups of the population is much more dangerous, since it is these groups that most often receive social assistance. In the event of a change in political power or aggravation of social problems, it is likely that individuals may be included in the lists of persons who are threatened with persecution or repression. Therefore, the majority of human rights defenders actively oppose not only state projects in this direction, but also against the introduction of such systems in the private sector. In addition, sooner or later the issue of compliance with European standards will arise, especially in the field of personal data protection and prevention of discrimination based on social status. These challenges require special attention to prevent the negative consequences of automation of vulnerable areas.

The new case management tool allows you to create an electronic space for interaction between recipients of social services, their providers, and case managers. Thanks to this, the person who seeks help receives an individual approach and timely support. Artificial intelligence significantly speeds up the processing of requests, which allows you to quickly respond to people's needs.

Currently, the Ministry of Justice of Ukraine uses the Cassandra software developed on the basis of AI algorithms in test mode. This system determines a person's tendency to repeat crimes (recidivism) using a questionnaire containing 97 questions. The results of the analysis are integrated into the pre-trial report, improving the validity of court decisions. Other machine learning algorithms of the system are used to analyze large amounts of data, compare them with information from different sources and identify signs of corruption offenses.

AI is also actively used to monitor political advertising on the Internet, which allows you to detect manipulation and disinformation. In many countries, police use AI to track down criminals and identify suspects. In addition, AI is effectively used to search for missing people [8]. For example, facial recognition technologies have successfully helped to find dozens of people who disappeared during military conflicts and natural disasters.

Despite the active use of AI in management and the attempts of professional circles to create legislative regulation of its application, many problems remain unresolved. Given that artificial intelligence is an integral part of information technology, two main groups of risks can be distinguished: socio-economic and technological. Technological risks are associated with the peculiarities of the use of information and communication technologies (ICT) and software. Artificial intelligence and machine learning are interrelated, but not identical, areas of information technology (IT) development: machine learning is one of the key sub-fields of AI. Scientists use AI to analyze complex scientific data, conduct simulations, and solve complex problems in various fields, including physics, biology, and economics. With the help of AI, SMM specialists (Social Media Marketing) can analyze large amounts of data from social networks, which allows them to identify current trends, predict audience behavior, and develop effective strategies for marketing campaigns.

Automated tools greatly simplify the process of scheduling and publishing content, as well as provide fruitful interaction with the audience through chatbots and personalized messages. With automated tools, the processes of scheduling, publishing content, and personalized interaction with the audience become more efficient.

In the field of programming, AI is used to automate routine tasks such as writing code, testing applications, and detecting bugs. AI contributes to the creation of intelligent programs that can learn and adapt to changing conditions.

Data Science is an interdisciplinary field of knowledge that is formed at the intersection of three main areas: computer science, domain expertise, and mathematics with statistics. Its structure can be explained using a Venn diagram demonstrating the effective interconnection of these areas. Data Science integrates programming tools, deep domain knowledge, and mathematical modeling techniques, allowing you to work with data to analyze, predict, and support decision-making.

Computer science encompasses programming, modeling, DevOps, and other technologies. Domain expertise is an examination in any subject area where it is advisable to analyze data. It can be medicine, advertising, education, sociology, economy, industry, etc. Good Data Scientists are usually specialists in a particular domain area. Data Science is based on mathematics and statistics, so people with an academic or professional background often move into data science, using their knowledge to analyze data.

Modern AI is the ability of machines and programs to analyze information, process it, and form conclusions that serve as the basis for decision-making. The main feature of AI devices is their ability to constantly learn, quickly process information, accumulate knowledge and apply it effectively. This means that such systems can acquire the ability to perceive and analyze the world around them, similar to how the human brain does. The potential of AI application is too wide, and today it is already used in many areas, such as medicine, finance, industry, trade, as well as in everyday human life. By using AI modules, businesses can achieve significant revenue growth and improved core performance indicators.

The Internet of Things (IoT) is a system that connects devices connected to the Internet, ranging from smartphones and TVs to household appliances such as refrigerators. In fact, any device connected to the Internet that does not belong to traditional desktop or laptop computers is considered part of the Internet of Things. smart speakers, etc. That is why machine learning methods form the basis for the effective functioning of many IoT devices. Machine learning (ML) is  a set of methods in the field of artificial intelligence, as well as algorithms used to create systems capable of learning from their own experience. In the process of learning, such systems process and analyze large arrays of input data, identifying patterns that allow them to perform complex tasks. Today, machine learning is mainly defined as the direction of development of AI technologies, aimed at creating flexible algorithms capable of self-learning and adaptation based on the data obtained. As a subset of artificial intelligence, machine learning allows computers to "learn" in a similar way to the human brain. Therefore, this technology is widely used in various fields, including healthcare, urban planning, financial analysis, and other fields,  requiring the processing of large amounts of data.

For example, Netflix uses machine learning to analyze data and make recommendations based on user preferences. In turn, Tesla applies this technology to ensure the operation of its self-driving cars. Thanks to machine learning, systems are able to identify patterns in large data sets and make informed decisions based on this analysis. Machine learning is a subset of AI – one of its components. Although these concepts are different, they remain closely interrelated.

At the same time, machine learning is one of the areas of AI that provides the ability of a computer system to learn from data and make decisions based on the results of this learning. In addition to machine learning, the concept of AI includes such areas as deep learning, robotics, Natural Language Processing (NLP) and other areas. Unlike conventional programmed algorithms, models machine learning systems improve their efficiency through the accumulation of experience. Their accuracy depends on the volume and quality of the available data. In the field of IT, machine learning is considered as a fundamental area of Data Science. For some time, the terms ML and Data Science have even been used interchangeably. The Data Scientist profession typically involves knowledge of machine learning, including proficiency in relevant algorithms, programming languages, and the basics of mathematical analysis, and statistics. In general, machine learning has become a massive technology that finds application in many areas: recommendation engines in social networks, photo processing algorithms in smartphones, image and natural language recognition applications, financial fraud monitoring – all this works thanks to machine learning models.

However, the difference between AI and ML can be very significant. AI, ML, Big Data, and the Internet of Things are key technologies that have the potential to radically change lives and the world. In particular, machine learning and the Internet of Things have a wide range of applications.

Data Science uses machine analysis methods to understand the large amounts of data at the disposal of businesses. Today, most companies accumulate significant amounts of data, but often do not know how to use it effectively. That is why they involve Data Science specialists and analysts to optimize processes and gain useful insights.

Closely related to data analysis, the direction of Data Mining. Data Mining allows you to identify hidden patterns and patterns in large data sets.

A particularly important process in data analysis, machine learning, and decision-making is Data Cleaning. This is the main stage that determines the accuracy, reliability and applied value of the data used to obtain the results.

Data Exploration is the first important step in exploratory data analysis, in which existing datasets are examined to identify patterns, anomalies, or natural structures. The main goal of Data Exploration is to form a clear story about this data, as well as a deeper understanding of its structure and features before further analysis or modeling.

Feature engineering is a key step in data analysis, which involves the process of creating new characteristics (features or features) that may contain new types of variables or derived data. This improves the quality of analytical models and allows you to gain deeper insights from the data. For example, if we have a dataset with indicators of wages and education levels of a certain population group, you can create a new variable,  which combines these two indicators. The new variable will allow you to display the dependencies between indicators in more detail and is used to build analytical models. It is the creation of such effective variables that is the main task of Feature Engineering.

Predictive modeling is used to create models based on available data to predict future events or outcomes. The main goal of predictive modeling is to select modeling methods, clean up data, build and adjust the model, and evaluate its performance based on test data.  It is used in data science [4; 5; 6; 7; 8; 9], machine learning and statistics to solve problems where the key is to predict the values of the target variable.

Using Machine Learning and Data Science, researchers can predict the structure of a given protein, filter out 99% of ten thousand possible combinations, and focus on the hundred most likely cases. It is important to adequately present the data. Personal understanding of data is one thing, while the ability to interpret and pass it on to others is another, more difficult task. This process of data transformation and refinement is very important and costly.

Also, do not confuse the concepts of Data science and Machine learning. These tools overlap in many ways, but they are still different and have their own tasks. AI is much broader, covering many areas outside of ML. In a broad sense, AI includes all technologies and methods of imitating human thinking using machines, including machine and deep learning models, neural networks,  large language models, generative AI, etc. The implementation of full-fledged artificial intelligence goes far beyond ML. This requires not only the construction of data processing algorithms, but also the machine logic of making judgments and making independent decisions to achieve a certain goal.

Artificial intelligence (AI) is a variety of technological and scientific solutions. Artifical narrow Intelligence (ANI) is an algorithm that specializes in one specific task or area of expertise. This type of intelligence can independently develop or surpass human intelligence only in the process of performing a specialized task. Examples of ANI include chess supercomputers, search algorithms, self-driving car control systems, etc. Machine learning algorithms belong specifically to the ANI category.

Artifical general intelligence (AGI) is an AI system that is capable of solving any intellectual tasks that a human can solve. This type of AI operates with logic and uses categories of abstract thinking, which allows it to extrapolate its experience to a wide range of tasks and situations, which ANI algorithms cannot achieve[11].

Artificial super-intelligence (ASI) is the highest level of AI that surpasses human abilities in all aspects, including creativity and social skills.

In addition, AI has become the subject of philosophical reflections and discussions. At the same time, the field of practical application of AI tools for the needs of business, education, medicine, and other industries is developing. This is the main difference between ML and AI: the concept of Machine Learning indicates practical technologies and areas of application of algorithms in work tasks. Machine learning is one of the areas of AI where algorithms allow computers to draw conclusions based on data without following strictly defined rules. That is, computer systems are able to detect patterns in complex and multi-parametric tasks (which the human brain is not able to solve) and find more accurate answers. This leads to correct forecasting. Neural networks that simulate the work of the human brain using artificial neurons self-learn based on previous experience and make fewer and fewer mistakes with each subsequent use [12].

Neural networks are a type of machine learning, not a separate tool. The main goal of machine learning is to partially or fully automate decision-making processes in complex analytical problems. Therefore, first of all, machine learning is aimed at providing the most accurate predictions based on the input data, so that business owners, marketers, and other specialists can make informed decisions. decisions in their work. In the process of learning, a computer system can predict the result, remember it, reproduce it if necessary, choose the best of several options. Currently, machine learning covers a wide range of applications from financial institutions, restaurants and gas stations to production.    Machine learning models are developed to solve practical problems by identifying patterns in large data sets. These are computer vision tools, advertising algorithms, mobile photo processing tools, content generation, and others. At the same time, the main goal of AI technologies is much broader, being to emulate human intelligence and behavior in various scenarios. ML is just one way to achieve this goal, but it's not the only one.

Machine learning is based on three main components that provide the most effective interaction with the customer:

1.    Дані: це основна інформація, яку зазвичай отримують від клієнта. До них входять будь-які вибірки даних, з якими система повинна працювати під час навчання.

2.    Ознаки: це частина роботи, що проводиться в тісній співпраці з клієнтом. Визначення ключових бізнес-потреб і спільне вирішення, які характеристики та властивості повинна відстежувати система в результаті навчання.

1.                      Algorithm: this is the choice of a method for solving a business problem. This task is solved without the participation of the client, using the knowledge and experience of the company's employees.

This approach allows you to implement machine learning in various areas of business and technology as efficiently as possible, ensuring the accuracy of forecasting and the selection of the best solutions.

Data – the more data is uploaded to the system, the better, faster and more accurate it will work. The data itself directly depends on the task  facing the computer system. For example, to recommend products to buyers that may be of interest to them,  they need a  history of their previous purchases. To predict changes in prices in the market, you need a price history. The most difficult and at the same time the most voluminous part of the work is the collection of this data. There are two methods of data collection: manual and automatic. The manual method is much slower, but more accurate. The automatic method is faster, but can lead to more errors.  Qualitative data sampling is very important, as it affects the accuracy of the prediction obtained as a result. It is also necessary not to limit data collection to the limits of human thinking, but to provide the most scattered information, since the system can detect relationships where a person does not notice them.

Features are  properties, metrics, data characteristics. Properties include those data characteristics on which the initial result of analysis or modeling directly depends. Since the correct data characteristics directly affect the result, the process of selecting them usually takes much longer than the machine learning process itself.  it is the avoidance of limitations in the set of data characteristics so as not to distort machine perception by personal assumptions. And with it, the end result [3; 4; 5; 6; 7].

Algorithm is a system of sequential operations to solve a certain problem. In other words, it is a method of solving a problem. For each specific task, you can choose a separate sophisticated algorithm. The speed and accuracy of the input processing result directly depend on the chosen method. There are cases when even perfectly written algorithms do not help solve business problems. For example, if you need to increase the number of cross-sales on the site and you are sure that for this it is enough to improve the algorithm for recommending products. However, this does not take into account the fact that customers come through direct search links and ignore the tips for buying other products shown on the site. Therefore, before starting work, it is important to determine the real cause of the client's problem. If it is a technical problem, it is solved by the company's specialists.

According to the presence of a teacher, machine learning can be divided into three types: Supervised Learning, Unsupervised Learning, and Reinforcement Learning.

The main types of machine learning:

-          Supervised Learning is used when it is necessary to train a system to recognize objects or signals based on labeled data, where the model receives pairs of input data with correct answers (labels) to adjust its parameters or build a model that can predict responses for new data.

-          Unsupervised Learning is based on the principle of "this thing is the same as the others", which consists in identifying similarities and hidden structures in unknown data without labels or categories. Algorithms in this type of learning can detect differences, find anomalies, and recognize unusual patterns without using labels that indicate correct answers.

-          Reinforcement Learning consists in the interaction of the agent's model with the environment, where he acts in order to maximize the rewards received for his actions. The difference between this approach is that it does not have unambiguous correct solutions, since the results depend on the variable reward. Reinforcement Learning is used where the computer system is faced with the task of effectively completing tasks in an external environment, where there are many possible options for action, such as in computer games, trading operations and unmanned vehicle management.

According to the use of algorithms,  there are two main types: deterministic and probabilistic.

A deterministic algorithm is a computational process that always produces the same result under the same inputs and execution conditions. Determinism ensures reliability and predictability of results, which is important in many fields, including programming, science, and engineering, where critical accuracy and stability of task execution matter.

A probabilistic algorithm is a computational process that uses elements of randomness or probability distributions to solve problems. These elements can change their result under the same input conditions, even with different runs. Such algorithms are used in statistics, machine learning, finance, and other fields to simulate complex phenomena and solve problems where randomness or uncertainty is important.

Classical learning is a well-known and well-studied learning algorithm that was developed more than 50 years ago, mainly for statistical bureaus. These algorithms are mainly applied to work with data such as classification, clustering, regression, and other analytical tasks. They are used for forecasting, customer segmentation, and other aspects of data analysis and processing.

Neural networks and deep learning represent a modern approach to machine learning, applied in cases where recognition or generation of images and videos, complex control or decision-making algorithms, machine translation, and similar tasks are required. Yes, multiple approaches can be combined, in which case multiple independent machine learning models are combined together as ensembles of models to achieve better results.

Therefore, ML technologies can be categorized as artificial intelligence, and in many contexts, these concepts are used interchangeably. However, we will focus on their differences and determine how AI differs from ML. Full-fledged artificial intelligence is able to solve a wide range of tasks that can be solved by humans. AI is able to reproduce the patterns of thinking that humans are used to calling common sense, extrapolate the experience gained to a wide range of new tasks, identify abstract thinking and creativity, and take into account the social and ethical context when making decisions. ML has a narrow application related to data processing to solve specific problems. The benefits of combining machine learning and the Internet of Things are numerous. One of the main advantages of machine learning is its ability to process huge amounts of data and extract insights from it, even if it is not known exactly what trends to look for.  which helps to reveal the essence of the problem. Machine learning algorithms typically use one or more data libraries in order to speed up the development process. By definition, machine learning is an integral part of artificial intelligence, they are different but closely related.

As a result of transformational processes in the institutional economy of Society 5.0, there is an intellectualization of the economy with signs of innovation and digitalization at all levels. In the context of globalization, the substitution of labor for knowledge is a radical change of the XXI century, characterized by the use of the latter in the practical processing of knowledge resources.

In this regard, in the digital Society 5.0, the source of value is knowledge, not labor. Therefore, when implementing digitalization in the environmental sphere, an extremely important aspect is to take into account side effects, such as cyber incidents and cyberattacks, as well as to ensure effective cybersecurity and protection of the created digital space from cybercrime and possible abuse of digital technologies.

In the context of the rapid development of artificial intelligence, it is particularly important to establish a legal framework that can ensure a balance between technological innovation, ethical responsibility and the protection of fundamental human rights. In Ukraine, as in many other countries, a systematic legislative framework that comprehensively covers the regulation of the development, implementation and monitoring of AI systems has not yet been created. Existing regulations, particularly in the area of personal data protection, are only partially capable of addressing the specifics of algorithmic decision-making, the absence of human control, or the functioning of self-learning systems.

One of the key legal challenges is determining responsibility for the consequences of decisions made by AI systems, especially in cases of autonomous operation. There is a need to introduce principles of transparency (explainability) and reproducibility of decisions, as well as to develop standards for mandatory auditing of algorithms used in the public sector or in vulnerable social environments, such as healthcare, the judiciary, and social protection. The legislative initiative should provide for the creation of an independent body for ethical monitoring of AI use (e.g., an Algorithm Ethics Commission), similar to the institutions proposed under the EU AI Act.

In the law enforcement sphere, there are difficulties in qualifying violations committed through AI, especially when it comes to so-called ‘digital harm’ that has no direct material expression but violates privacy, the right to a fair trial, or equality. For example, discrimination caused by opaque algorithms for selecting job candidates or granting loans requires the development of legal protection mechanisms at the level of individual court appeals, administrative appeals, or the introduction of ‘algorithmic ombudsman’ models.

On the other hand, legal regulation of AI is impossible without taking ethical dimensions into account. That is why there is a growing need to create ethical frameworks based on universally recognised principles: respect for personal autonomy, fairness, inclusion, avoidance of harm, integrity and accountability. International standards play an important role here, in particular the UNESCO Recommendation on the Ethics of Artificial Intelligence (2021), which can be implemented in national policies through the development of ethical codes for developers, suppliers and users of AI systems.

Particular attention should be paid to the problem of ‘digital inequality’ caused by unequal access to data, algorithms and opportunities to use AI in the public and private sectors. If these imbalances are not addressed through legal and political means, a situation may arise where the digital elite gains advantages in knowledge, power mechanisms and resources, which would contradict the principles of democracy and social justice.

Thus, the formation of a national and cross-border legal regime for AI regulation requires a systematic approach that integrates institutional, ethical, regulatory and educational components. The decisive factor in the effectiveness of such a regime should be not only the number of laws, but also the quality of institutional accountability, the existence of mechanisms to monitor compliance with ethical standards and human rights, and the constant updating of the regulatory framework in line with the pace of technological progress. Only under such conditions will artificial intelligence serve as an instrument of public good rather than a source of uncontrolled risks.

Conclusions.

AI is one of the priority areas in the field of digital transformation of public administration, so it is necessary to quickly make decisions that will allow you to adapt to the rapidly changing processes in the field of AI and influence the evolution of technologies. To ensure the effective implementation of AI technologies and obtain the most positive results, it is necessary to constantly develop and update complexes of optimal regulatory measures. It is also important to focus not only on the development of laws, programs, strategies, but also to take into account the problems that hinder and inhibit the development of AI. The interaction of computer science and AI determines a new level of development of modern society, turning it into a technologically oriented and intellectually rich space. The application of these two fields in various fields, from public administration and health care to economics and education, highlights their importance in addressing today's challenges and challenges. Social informatics, which is based on the analysis of social processes and the use of information technologies, interacts with artificial intelligence, which provides the ability to efficiently process and interpret large amounts of data. This cooperation is manifested in the implementation of e-governance, the improvement of medical technologies, the creation of intelligent economic management systems and the improvement of educational practices. At the same time, an important aspect of the application of these technologies is to take into account the ethical and social challenges that may arise as a result of automation and the increase in the processing of personal data. The balanced development of social informatics and artificial intelligence requires careful consideration and protection of the rights and interests of citizens. In general, this interaction is formed around the idea of using advanced technologies to improve the quality of life, management efficiency and solve global problems. The development of these industries can determine not only the technological, but also the socio-cultural picture of the future, in which information technologies and intelligent systems become a necessary part of a sustainable and modern society. The need to introduce digitalization in the regulation of environmental relations can also be considered in the context of individual industries.

REFERENCES: 

1.      Nikolskyi Yu. V., Pasichnyk V.V., Shcherbyna Yu.M. Systemy shtuchnoho intelektu : navch. posib. Lviv : Mahnoliia 2006, 2010. 280 s.

2.      2. Spirin O. M. Dyferentsiiovanyi pidkhid u vyvchenni osnov shtuchnoho intelektu v kursi informatyky fizyko-matematychnoho fakultetu vyshchoho pedahohichnoho zakladu : dys. … kand. ped. nauk : 13.00.02. Kyiv, 2001. 227 s.

3.      Spirin, O. M. Pochatky shtuchnoho intelektu: navch. posib. dlia stud. fiz.-mat. spets-tei vyshchykh ped. navch. zakl-iv. Zhytomyr : Vyd-vo ZhDU, 2004. 172 s

4.      Sukhoruchkina O. M. Informatsiine zabezpechennia intelektualizovanykh robototekhnichnykh kompleksiv. Stan ta perspektyvy rozvytku informatyky v Ukraini. K.: Naukova dumka, 2010. S. 547-561.Швирков О. І. Проблема штучного інтелекту і людиновимірність штучних інтелектуальних систем : дис. … канд. філос. наук : 09.00.09. Житомир, 2006. 170 с.

5.      Shevchenko A. I. Svitovi tendentsii ta praktychni dosiahnennia u problemi shtuchnoho intelektu. Stan ta perspektyvy rozvytku informatyky v Ukraini. K.: Naukova dumka, 2010. S. 561-572.

6.      Karpenko Yu. V. Etychni pryntsypy zastosuvannia shtuchnoho intelektu u publichnomu upravlinni. Visnyk Natsionalnoi akademii derzhavnoho upravlinnia pry Prezydentovi Ukrainy. Ser. «Derzhavne upravlinnia». 2019. № 4(95). S. 93-97.

7.      Stratehiia rozvytku shtuchnoho intelektu v Ukraini : monohrafiia / A. I. Shevchenka ta in. : za zah. red. A. I. Shevchenka. Kyiv : IPShI, 2023. 305 s.

8.      Shevchenko А. І. ABOUT GENERATION OF ARTIFICIAL INTELLIGENCE. Штучний інтелект, 2016. № 1. С. 7-15.

9.        García-Martínez, I., Fernández-Batanero, J. M., Fernández-Cerero, J., & León, S. P. (2023). Analysing the Impact of Artificial Intelligence and Computational Sciences on Student Performance: Systematic Review and Meta-analysis. Journal of New Approaches in Educational Research, 12(1), 171-197. DOI: https://doi.org/10.7821/naer.2023.1.1240.

10.    Kostenko Oleksii, Shapenko Liudmyla, & Ogirko Ihor. (2024). METAVERSE: DIGITAL IDENTITY, ELECTRONIC JURISDICTION, DIGITAL RESERVATION, POST-QUANTUM DIGITAL CIVILIZATION, AND PHILOSOPHY OF IMMERSIVE ENVIRONMENTS. In Oleksii Kostenko & Yuriy Yekhanurov (Eds.), DIGITAL TRANSFORMATION IN UKRAINE: AI, METAVERSE, AND SOCIETY 5.0 (pp. 20-28). SciFormat Publishing Books. https://doi.org/10.69635/978-1-0690482-1-9-ch3

11.    Machine Learning and Artificial Intelligence: Analyzing Their Key Differences

https://wezom.com.ua/ua/blog/mashinne-navchannya-ta-shtuchniy-intelekt-rozbirajemo-yih-klyuchovi-vidminnosti

12.    Mashynne navchannia ta suchasni instrumenty DATA Science : metodychni rekomendatsii / uklad. : I. V. Skorniakova, O. V. Kuzminova, N.V. Popyk. – K. : NAU, 2023. – 32s. https://er.nau.edu.ua/bitstream/NAU/60673/4/%D0%9A%D1%83%D1%80%D1%81%D0%BE%D0%B2%D0%B0_%D0%9C%D0%B0%D1%88%D0%B8%D0%BD%D0%BD%D0%B5%20%D0%BD%D0%B0%D0%B2%D1%87%D0%B0%D0%BD%D0%BD%D1%8F%20%D1%82%D0%B0%20%D1%81%D1%83%D1%87%D0%B0%D1%81%D0%BD%D1%96%20%D1%96%D0%BD%D1%81%D1%82%D1%80%D1%83%D0%BC%D0%B5%D0%BD%D1%82%D0%B8%20Data%20Science.pdf

Комментарии

Популярные сообщения из этого блога

Vorotin, Valerii; Vorotina, Nataliia; Prodanyk, Vasyl (2025). THE HYBRID TRAJECTORY OF PUBLIC ADMINISTRATION IN UKRAINE: A MODEL OF RESILIENCE, REFORM, AND EUROPEAN INTEGRATION. Social and Human Sciences. Polish-Ukrainian scientific journal (https://issn2391-4165.webnode.com.ua/), 01 (45).