Artificial Intelligence Tools for Managing the Behavior of Economic Agents at the Micro Level

This article is a summary of the scientific work published by Svitlana Turlakova, Bohdan Lohvinenko

The original source can be accessed via DOI 10.33111/nfmte.2023.003

‍

Introduction

Modern enterprises are undergoing significant transformation as technological advancement accelerates within the framework of Industry 4.0. Business processes are increasingly transitioning into digital environments, where value is not only created through products and services, but also through the intensity and quality of interactions among internal teams, external partners, and broader economic systems.

In this evolving landscape, traditional approaches to managing human resources are being challenged. Enterprises must rethink their personnel strategies to respond to greater complexity, higher demands for adaptability, and the growing integration of intelligent systems. A particularly noteworthy development is the adoption of artificial intelligence (AI) tools designed to support decision-making, enhance operational responsiveness, and improve the ability to manage change effectively. These systems are not merely automation tools—they are platforms that enable reflexive management by learning from data, adapting to behavior patterns, and enhancing the quality of human capital decisions.

Importantly, AI systems contribute to enterprise stability and performance in the digital economy, without functioning as coercive or manipulative instruments. Rather, they assist in diagnosing challenges, revealing development opportunities, and enabling evidence-based human resources management. However, their rising influence also raises fundamental questions about ethics, data transparency, and the balance between technological advancement and human-centered development.

In the context of heightened uncertainty and rapid change, effective human resource management is a critical factor in building sustainable competitive advantages. The ability to analyze workforce dynamics, forecast staffing needs, personalize employee development, and foster a culture of innovation is increasingly viewed as a strategic necessity. Organizations are therefore seeking out innovative tools and technologies that allow them to better align talent strategies with business goals.

AI-powered systems are becoming integral to how enterprises plan, operate, and grow. Their applications in human resources span recruitment and onboarding, performance evaluation, career pathing, and engagement strategies. While concerns remain around job displacement and over-reliance on algorithms, the overall direction is clear: organizations are leveraging AI to improve decision quality, reduce inefficiencies, and foster a more responsive and supportive workplace environment.

The use of AI in solving personnel management tasks not only responds to immediate business needs but also opens the door to fundamentally new approaches to workforce development. It creates opportunities to enhance decision-making effectiveness, strengthen employee potential, and increase overall organizational adaptability.

Given these developments, it is essential to examine the key dimensions of AI's influence on human resource management. This includes assessing how such tools affect core HR processes, evaluating their effectiveness in supporting strategic workforce goals, and exploring how AI can be used to forecast key indicators of employee performance and enterprise development. In doing so, organizations can better position themselves to remain competitive, resilient, and forward-looking in an increasingly digital world.

Methodology

This study focuses on the implementation of AI-powered tools within cloud-based enterprise systems, particularly SAP SuccessFactors, to analyze, predict, and optimize personnel behavior. The core objective is to assess the effectiveness of AI in shaping employee decision-making and identifying the key behavioral and contextual factors that impact performance outcomes.

In the digital enterprise ecosystem, SAP SuccessFactors functions as a centralized platform where economic agents—namely, HR managers and employees—interact in real-time. This interaction supports the execution of standardized business processes and ensures alignment with organizational goals. Managers oversee task completion, provide feedback, and assess performance via the platform, while employees carry out assigned duties and engage with the system as management objects.

In this context, the management subject (MS) refers to the enterprise manager (e.g., department head), who actively uses SAP SuccessFactors to monitor and influence employee behavior. The management object (MO) is the employee, who autonomously chooses how to act within the workplace—ranging from full engagement and responsibility to partial or passive participation.

The goal of the management process is to encourage the MO to act in accordance with organizational expectations, leading to higher efficiency and strategic alignment. This goal is pursued through the manager’s use of AI-enhanced analytics and forecasting tools embedded in the SAP SuccessFactors platform.

The decision-making behavior of employees is not random but influenced by a set of reflexive factors that determine how they engage with their tasks. These include:

• Awareness, which reflects how informed the employee is about expectations and workplace dynamics;
  
• Competence, referring to the individual’s skills, experience, and confidence in task execution;
  
• Motivations and goals, representing the perceived value or utility of different behavioral choices;
  
• Digital interaction intensity, or the level of engagement with the platform and its tools.
  
  

The interaction between these factors shapes the mechanism of choice, through which the employee arrives at a decision about their behavior. AI tools integrated into SAP SuccessFactors support the MS by analyzing this complex decision space, providing insights into potential outcomes and optimal intervention strategies.

This conceptual framework is illustrated in Fig. 1, which depicts the dynamic interaction between the management subject, management object, the choice mechanism, and AI-enabled inputs. AI acts as both an analytical and reflexive tool, drawing on past behavior patterns and contextual signals to support managerial decisions.

Ultimately, this approach enables not only real-time optimization of HR processes but also the prediction and guidance of employee behavior within the digital enterprise environment.

The methodology presented in Fig. 2 outlines a step-by-step framework for managing the behavior of economic agents—namely, employees—at the micro level through the SAP SuccessFactors system. It is designed to optimize personnel behavior by leveraging AI-driven insights and reflexive decision modeling.

The process begins with data collection within SAP SuccessFactors. The system maintains a comprehensive behavioral profile for each employee from the point of hiring onward. This includes metrics on education, work experience, task completion, communication frequency, use of platform features, participation in feedback loops, training engagement, and growth trends. These metrics help quantify various behavioral dimensions such as awareness, competence, intentions, and digital interaction intensity.

Once collected, the data undergo normalization to ensure all parameters are aligned to a common scale, typically from 0 to 1. This standardization step allows the system to uniformly interpret inputs regardless of format or source. In cases where data are expressed qualitatively, fuzzy set logic is used to convert linguistic variables into numerical values, allowing the integration of subjective assessments such as motivation and engagement.

Following normalization, the reflexive characteristics of each employee are extracted. These include indicators of awareness, such as system activity and knowledge of corporate procedures; competence, measured through task efficiency and team interactions; intention, which reflects motivation and alignment with organizational goals; and digital engagement, defined by the depth and frequency of interaction with SAP SuccessFactors. Kohonen self-organizing maps are applied at this stage to classify agents into behavioral profiles, forming the foundation for forecasting likely actions in the workplace.

The next stage involves predicting employee behavior using a reflexive choice function based on decision theory. This model simulates the mental process an employee undergoes when choosing between alternatives—such as high-performance task execution versus disengagement. The function captures how external influence (from management), internal state (intentions), awareness, competence, and digital interaction jointly shape the probability of a productive decision. The outcome of this function represents the likelihood that the employee will act in line with enterprise goals.

Once these behavioral probabilities are calculated, employees are segmented into three groups based on their predicted efficiency. The first group includes employees with low motivation and a potential risk of attrition. The second group shows moderate engagement and performance. The third group represents the most effective and goal-aligned employees. This classification enables targeted intervention.

Management then designs reflexive influences tailored to each group. For those with lower engagement, the focus may be on increasing awareness through transparent communication, enhancing competence through training programs, or improving motivation through personalized incentives, recognition, and career development opportunities. Managers may also strengthen digital interaction by optimizing workload and increasing participation in digital processes. Employees already in the high-performance group typically require no intervention, though continuous support may reinforce their alignment with organizational objectives.

After these management actions are applied, the system performs a new evaluation to assess changes in employee behavior. Updated metrics are analyzed to recalculate reflexive characteristics and determine if the probability of productive behavior has increased. If goals remain unmet, the cycle returns to the grouping stage, and reflexive strategies are adjusted accordingly.

Finally, the results of this methodology are compared against predictions made by the SAP SuccessFactors AI engine. This comparative analysis highlights the relative accuracy and added value of incorporating reflexive modeling and neural network clustering into HR management. It also demonstrates the potential for combining system-generated analytics with human-led strategic planning to achieve optimal behavior outcomes in enterprise settings.

This framework allows enterprise leaders to move beyond reactive HR management and into a predictive, adaptive, and human-centered approach that aligns employee behavior with long-term organizational goals.

Results and Discussions

The Strategic Resource Management Department, consisting of 22 employees and one department head, serves as the testbed for implementing the proposed AI-driven methodology for managing employee behavior via SAP SuccessFactors. This department is dedicated to HR services for external organizations, with core responsibilities including strategic planning, talent development, performance analysis, and motivation strategies. The aim is to align employee behavior with organizational goals through data-driven decision-making and personalized personnel management.

1. Data Collection via SAP SuccessFactors

This step outlines how employee behavior data is collected within the Strategic Resource Management Department, consisting of 22 staff members. SAP SuccessFactors automatically gathers detailed information on each employee from the moment of hiring, including both static personal data and dynamic performance metrics over time. These metrics—such as productivity, motivation, and communication—are expressed numerically (mostly as percentages). The department manager uses system access to export this data for further analysis, forming the foundation for behavior management through the proposed methodology.

2. Data Normalization for Reflexive Analysis

The initial dataset retrieved from the SAP SuccessFactors system contained diverse data formats, including percentages, textual descriptions, and numerical indicators. To ensure analytical consistency and compatibility with neural network processing, a comprehensive normalization process was conducted.

This process utilized a methodology based on Zadeh’s theory of fuzzy sets, enabling the transformation of linguistic and non-numeric characteristics into quantitative values. As a result, all employee-related parameters were converted into normalized numerical values ranging from 0 to 1.

The processed dataset includes 22 agents (employees), each evaluated across multiple indicators related to awareness, competence, intentions, and digital interaction. The normalized values form the basis for subsequent clustering and forecasting procedures.

This normalized input provides a standardized foundation for neural network modeling and deeper behavioral analysis in the following stages.

3. Determination of Reflexive Characteristics

To gain a deeper understanding of employee behavior and support data-driven managerial decisions, neural network-based tools were employed—specifically, Kohonen self-organizing maps within the Deductor Academic software suite. This method enables the clustering of employees based on behavioral similarities and allows for the quantification of key reflexive characteristics essential to organizational performance.

The process begins with inputting normalized data (see Table 1) into the neural network. The Kohonen maps then illustrate how employees are grouped into clusters across four core dimensions:

• Awareness
• Competence
• Intentions
• Intensity of interaction in the digital space

Each of these dimensions is visualized separately in the Kohonen maps, offering a clear view of behavioral patterns across the employee base. The results of the clustering are presented in Figure 3.

From the neural network outputs, three clusters were formed for each direction of analysis. For each employee, we assigned the corresponding cluster values, which represent their reflexive characteristics. These quantified values, derived from the center of each cluster, are compiled in the following summary.

This quantitative insight provides a foundation for deeper behavioral modeling and predictive analytics. In the next stage, we move toward forecasting employee decision-making outcomes based on their identified reflexive traits.

4. Forecasting Behavior Using the Reflexive Choice Function

With the reflexive characteristics quantified, each employee's decision-making behavior was forecasted using a modified version of Lefebvre’s reflexive choice model. This mathematical function evaluates the probability of an employee choosing between two behavioral outcomes: effective work performance or neglect of duties. By applying the model to each individual's profile, a predicted behavioral outcome was generated. These predictions offer insight into the likelihood of productive engagement across the department, forming the basis for targeted management interventions.

5. Determining Groups for Reflexive Influence

After calculating the reflexive choice function for each employee, staff members were segmented into three behavioral groups based on the likelihood of making efficient and motivated decisions. These thresholds mirror the structure used in systems like SAP SuccessFactors: employees with function values from 0 to 0.4 form the first group, indicating low motivation and a higher risk of disengagement or resignation. The second group spans values from 0.41 to 0.69, reflecting moderate performance and mixed behavioral tendencies. The third group includes those scoring 0.7 and above, representing high-performing and goal-aligned employees who demonstrate strong commitment and intention to meet organizational objectives.

This stratification enables targeted management interventions—whether corrective or supportive—based on the behavioral disposition of each group. It also highlights which employees may require development or motivation efforts and which are driving the team’s success.

6. Building reflexive influences on groups of agents.

To improve the behavior and efficiency of moderately performing employees, targeted reflexive influences were introduced. These employees, identified as having low awareness, weak intention, and limited interaction with the digital environment, were supported through a combination of motivational and developmental strategies. These included recognition systems, bonuses, praise, and career growth opportunities to stimulate performance. Additionally, internal motivators such as professional training, open communication about company goals, a balanced workload, and mentorship were used. The measures were tailored based on the specific reflexive characteristics of each employee and aimed to positively shift their behavioral tendencies and increase their engagement and productivity.

7. Prediction of the decision-making result of economic agents after management and its comparison with the predicted values of the efficiency of economic agents by SAP SuccessFactors AI

Following the implementation of targeted managerial interventions, updated behavioral data were collected from the SAP SuccessFactors system. Analysis of this new dataset revealed an increase in the values of reflexive characteristics and corresponding improvements in the reflexive choice function for nearly all employees, with only one exception. Notably, five employees shifted to higher performance categories—two from low to medium effectiveness, and three from medium to high.

These results affirm that monitoring and adjusting reflexive characteristics allow for proactive employee behavior prediction and timely interventions to improve performance. The findings are presented in Table 4, which shows changes in the reflexive choice function before and after the applied management influence.

In parallel, the study compared forecasts derived from the proposed reflexive model with those generated by SAP SuccessFactors’ built-in AI prediction system. For consistency, the reflexive model's outputs were converted into a percentage format to match the SAP standard. The comparison, detailed in Table 5, revealed a key trend: SAP SuccessFactors tends to overrate high-performing employees and underrate those with lower performance, creating a wider-than-justified efficiency gap.

This discrepancy can lead to biased managerial decisions, especially in reward distribution or identifying underperformers. For example, SAP might downgrade employees simply due to system-flagged absence from specific project activity, without considering legitimate reasons such as business travel or health issues. These inconsistencies highlight the potential risks of relying solely on automated AI assessments in HR decision-making and underscore the added value of reflexive behavior-based modeling for a more nuanced, context-aware evaluation of employee performance.

Fig. 4 illustrates a clear divergence between the author’s reflexive choice function model and SAP SuccessFactors AI in grouping employees by predicted productivity. Notably, the reflexive model identifies a greater number of high-performing employees compared to SAP, while SAP tends to classify more individuals in the lower-performance category.

This comparison underscores a critical issue: SAP SuccessFactors may apply generalized performance thresholds that don’t adequately reflect individual differences or contextual nuances. Consequently, the performance distribution produced by the system can exaggerate the gap between low and high performers, potentially leading to skewed management decisions.

During the experiment, several limitations of SAP SuccessFactors were identified:

Despite these drawbacks, SAP SuccessFactors remains a powerful data-driven HR tool. However, its outputs should be supplemented with additional modeling or managerial judgment—such as the reflexive behavior-based approach tested in this study—to ensure balanced evaluations and more informed personnel decisions.

This combination can ultimately enhance organizational performance by aligning system intelligence with human insight.

Conclusion

In the context of rapid technological development, effective personnel management has become a critical factor in enhancing enterprise competitiveness. Integrated systems like CRM and ERP, along with artificial intelligence (AI) tools, offer substantial support in strategic decision-making and workforce optimization. AI, in particular, facilitates improved employee motivation, personalized management, and higher operational efficiency.

The study applied a proposed methodology to assess AI’s role—specifically through SAP SuccessFactors—in managing employee behavior and forecasting productivity. The findings revealed a bias in SAP SuccessFactors: the platform tends to underestimate the performance of low- and mid-level employees while overestimating high performers. This imbalance may result from the system’s limited adaptation to individual job specifics and personal traits, occasionally leading to unfair performance ratings.

Such discrepancies can result in misguided HR decisions, from bonus allocation to workforce planning, ultimately impacting the enterprise’s economic efficiency. For example, employees penalized for missed project participation may have legitimate justifications, such as business travel or personal obligations, which the system fails to recognize.

While SAP SuccessFactors excels in data collection and automation, it lacks functionality in delivering targeted managerial recommendations. Additionally, limitations include the risk of systemic bias and concerns about data privacy and unauthorized access.

Despite these challenges, AI continues to offer vast potential for improving human resource management. Future work should focus on refining AI-driven models for better personalization, minimizing classification errors, and integrating recommendation systems. Research should also compare emerging AI tools for staff monitoring to design more adaptable and accurate HR solutions in a digital environment.

‍

References

1. Vishnevsky, V., Viyetska, O., Viyetskyi, O., Vorgach, O., Harkushenko, O., Dasiv, A., Zanizdra, M., Zbarazhskaya, L., Kniaziev, S., Kravchenko, S., Lipnytsky, D., Madih, A., Mazur, Y., Nikiforova, V., Okhten, O., Sokolovskaya, O., Turlakova, S., Chekina, V., Shevtsova, G., & Shchetilova, T. (2019). Smart-promyslovist: napriamy stanovlennia, problemy i rishennia [Smart industry: Directions of development, problems and solutions]. National Academy of Sciences of Ukraine, Institute of Industrial Economics. https://iie.org.ua/wp-content/uploads/2020/04/2019-smart-promyslovist_napriamy-stanovlennia-problemy-i-rishennia_compressed-1.pdf [in Ukrainian]
2. Kolot, A., Herasymenko, O., Shevchenko, A., & Ryabokon, I. (2022). Employment in the coordinates of digital economy: current trends and foresight trajectories. Neuro-Fuzzy Modeling Techniques in Economics, 11, 78-123. http://doi.org/10.33111/nfmte.2022.078
3. Zuboff, S. (2019). Age of surveillance capitalism: The fight for a human future at the new frontier of power. PublicAffairs.
4. Logvinenko, B. (2022). Doslidzhennia instrumentiv shtuchnoho intelektu v upravlinni povedinkoiu ekonomichnykh ahentiv u tsyfrovomu prostori na pidpryiemstvakh [Study of artificial intelligence tools in the management of the behavior of economic agents in the digital space at enterprises]. The Journal of V. N. Karazin Kharkiv National University. Series: International Relations. Economics. Country Studies. Tourism, 15, 45-53. https://doi.org/10.26565/2310-9513-2022-15-05 [in Ukrainian]
5. Haan, K. (2023). How Businesses Are Using Artificial Intelligence In 2024. Forbes Advisor. https://www.forbes.com/advisor/business/software/ai-in-business/
6. Lohvinenko, B. (2022). Upravlinnia povedinkoiu ekonomichnykh ahentiv na pidpryiemstvakh iz vykorystanniam instrumentiv shtuchnoho intelektu [Managing the Behavior of Economic Agents at Enterprises with Artificial Intelligence Tools]. Ekonomika Promyshlennosti (Economy of Industry), 4(100), 78-92. http://doi.org/10.15407/econindustry2022.04.078 [in Ukrainian]
7. Traynor, S., Wellens, M. A., & Krishnamoorthy, V. (2021). An Introduction to SAP SuccessFactors Talent Modules. In SAP SuccessFactors Talent: Vol. 1 (pp. 1–6). Apress. https://doi.org/10.1007/978-1-4842-6600-7_1
8. Gartner. (2022). SAP SuccessFactors HXM Suite Enterprise IT Software Reviews. Gartner Peer Insights. https://www.gartner.com/reviews/market/cloud-hcm-suites-for-1000-employees/vendor/sap/product/sap-successfactors-hxm-suite/reviews?marketSeoName=cloud-hcm-suites-for-1000-employees&vendorSeoName=sap&productSeoName=sap-successfactors-hxm-suite
9. Zuboff, S. (2015). Big other: Surveillance Capitalism and the Prospects of an Information Civilization. Journal of Information Technology, 30(1), 75–89. https://doi.org/10.1057/jit.2015.5
10. Crawford, K. (2022). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Perspectives on Science and Christian Faith, 74(1), 61–62. https://doi.org/10.56315/pscf3-22crawford
11. Kidd, C., & Birhane, A. (2023). How AI can distort human beliefs. Science, 380(6651), 1222–1223. https://doi.org/10.1126/science.adi0248
12. Yawalkar, V. V. (2019). A study of artificial intelligence and its role in human resource management. International Journal of Research and Analytical Reviews, 6(1), 20–24. https://www.ijrar.org/papers/IJRAR19UP004.pdf
13. Cappelli, P., Tambe, P., & Yakubovich, V. (2019). Artificial Intelligence in Human Resources Management: Challenges and a Path Forward. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3263878
14. Ledro, C., Nosella, A., & Dalla Pozza, I. (2023). Integration of AI in CRM: Challenges and Guidelines. Journal of Open Innovation: Technology, Market, and Complexity, 9(4), Article 100151. https://doi.org/10.1016/j.joitmc.2023.100151
15. Turlakova, S. (2022). Kontseptualni polozhennia upravlinnia povedinkoiu ekonomichnykh ahentiv u tsyfrovomu prostori z vykorystanniam instrumentiv shtuchnoho intelektu [Conceptual provisions of management of the behavior of economic agents in the digital space using artificial intelligence tools]. Ekonomika ta pidpryiemnytstvo (Economics and Entrepreneurship), 49, 40–54. https://doi.org/10.33111/EE.2022.49.TurlakovaS [in Ukrainian]
16. Turlakova, S. (2022). Modeling the values of reflexive characteristics of agents within the management of herd behavior at the enterprises. Neuro-Fuzzy Modeling Techniques in Economics, 11, 48-77. http://doi.org/10.33111/nfmte.2022.048
17. Turlakova, S. (2020). Refleksyvnoe upravleniye stadnym povedeniyem na predpriyatyyakh: kontseptsiya, modeli i metody [Reflexive management of herd behavior at enterprises: concept, models and methods]. National Academy of Sciences of Ukraine, Institute of Industrial Economics. https://iie.org.ua/wp-content/uploads/2020/04/mono_turlakova_2020_compressed.pdf [in Russian]
18. Zadeh, L.A. (2008). Is there a need for fuzzy logic? Information Sciences, 178(13), 2751-2779. https://doi.org/10.1016/j.ins.2008.02.012
19. Romanos, G. D. (1973). Reflexive Predictions. Philosophy of Science, 40(1), 97–109. https://doi.org/10.1086/288499
20. McClelland, D. С. (2001). Chapter 27: Where Do We Stand on Assessing Competencies? Counterpoints, 166, 479–489. http://www.jstor.org/stable/42977810
21. Spencer, L. M., & Spencer, S. M. (2008). Competence at Work: Models for Superior Performance. John Wiley & Sons.
22. Robert, K., & Ola, L. (2021). Reflexive sensegiving: An open-ended process of influencing the sensemaking of others during organizational change. European Management Journal, 39(4), 476–486. https://doi.org/10.1016/j.emj.2020.10.007
23. Jamieson, M. K., Govaart, G. H., & Pownall, M. (2023). Reflexivity in quantitative research: A rationale and beginner’s guide. Social and Personality Psychology Compass, 17(4), Article e12735. https://doi.org/10.1111/spc3.12735
24. Burk, D. (2020). Algorithmic legal metrics. AoIR Selected Papers of Internet Research, 2020. https://doi.org/10.5210/spir.v2020i0.11184
25. Kohonen, T. (2001). Self-Organizing Maps. Springer-Verlag.
26. Lefebvre, V. (2001). Algebra of conscience. Springer Science & Business Media.
27. Lefebvre, V. A. (2017). Bipolar Choice in Experimental Chamber. Bipolar Disorder, 3(1). Article 115. https://doi.org/10.4172/2472-1077.1000115

‍