This bar chart (Figure 2) illustrates the number of publications per year from 2015 to 2024, showing a clear increase in activity over time. Initially, the number of publications was relatively low between 2015 and 2018, but it began to rise significantly in 2020. This increase can be attributed to the COVID-19 pandemic (Kniffin et al., 2021), which caused substantial changes in the way businesses operate and had a profound impact on employee well-being. At the same time, the Fourth Industrial Revolution, characterized by the growing adoption of artificial intelligence, led to significant transformations in employment structures and work practices, redefining the future of work (Stamate et al., 2021).

The third figure illustrates the contributions of the 10 most prolific authors in the field of artificial intelligence and employee well-being. Although the field is still in its early stages, with a maximum of two publications per author, the citation analysis reveals their significant influence within the scientific community. Particularly notable are De Cremer with 12,516 citations, followed by Bromuri with 740 citations, and Byung-Jik Kim with 339 citations (Figure 3).

The fourth figure illustrates the countries that have contributed the most to research in this field. We notice a predominance of the United States and China, followed by European countries and the United Kingdom. South Africa and Saudi Arabia are also present, and contributions from other countries such as Malaysia, Canada, Hungary, and Morocco are observed. This distribution shows a global interest in research and underscores the need to increase research efforts in other countries where contributions are reduced or non-significant, such as Taiwan, Belgium, Turkey, and Japan (Figure 4).

Keywords form the foundation of published articles and illustrate the diversity of research within the studied field (Su and Lee, 2010). Through co-occurrence analysis, it is possible to identify emerging themes and analyze the interdependence between different fields of study (Chiang et al., 2023). Our keyword co-occurrence analysis has identified four main clusters that illustrate the relationship between artificial intelligence (AI) and employee well-being. Figure 5 shows the obtained visualization. The biggest number of keywords was 10, while the lowest was five (see Table 1).

The yellow cluster focuses on artificial intelligence and its various forms such as machine learning and automation highlighting ethical considerations and employability. Artificial intelligence and machine learning have transformed our way of life and our way of working (Ortega-Bolaños et al., 2024) and will continue to bring significant changes in the years to come (Sanderson et al., 2023). Like any powerful technological tool the adoption of artificial intelligence within companies raises important ethical questions (Trotta et al., 2023). These ethical issues touch on various aspects such as transparency fairness data confidentiality and accountability (Ortega-Bolaños et al., 2024; Trotta et al., 2023)

Indeed, the success of an AI-based project first requires transparency; a transparent project is one that is clearly explained, understood by all stakeholders, and motivating (Morley et al., 2020). A project must also be fair, ensuring that all employees have equal opportunities and are not disadvantaged by the deployment of AI technologies (Ortega-Bolaños et al., 2024). Authors (Bollier, 2017; Gibbons, 2021) suggest that widespread internet access and the implementation of a universal basic income could help address issues related to fairness, particularly for employees who perceive the adoption of artificial intelligence in the workplace negatively. Moreover, the use and storage of data present significant ethical challenges. Every individual should have the right to manage and control their data, ensuring privacy and preventing misuse (Khan et al., 2021). Finally, accountability is a crucial pillar in the management of artificial intelligence. Accountability assigns responsibility to the company or individuals for the outcomes produced by AI systems (Benneh, 2023). In sum, these ethical considerations provide a framework for better managing the implementation of artificial intelligence in the future (Sanderson et al., 2023; Benneh, 2023).

Concerning the red cluster, it refers to the psychological and health concerns of employees in the workplace. Indeed, the emergence of artificial intelligence within companies has disrupted the future of work (Fisher et al., 2023). While AI can offer countless benefits to employers and employees, such as improved productivity and performance, its implementation can also have negative impacts. Employees with gaps in digital skills may find themselves at a disadvantage, creating social disparities and increasing the risk of job loss (Landsbergis et al., 2014). According to some studies, the fear of job loss and the stress related to acquiring new skills are considered a public health crisis (Case and Deaton, 2017). To address this and promote equity by creating a stress-free work environment for employees, several tools need to be put in place. These include continuous training for employees to bridge skill gaps, considering the cultural, societal, and environmental aspects by employers, and placing humans at the center of attention (Gibbons, 2021).

The green cluster highlights the role of digital transformation and COVID-19 on employee engagement, performance, and satisfaction. The digital transformation, accelerated by the COVID-19 pandemic, has revolutionized the world of work and business (Schilirò, 2021). The adoption of new technologies, such as artificial intelligence, creates job insecurity and affects employee engagement and performance (Ye et al., 2024; Hizam et al., 2023). To address this, acquiring new skills and improving existing ones among employees is a key element, enabling companies to enhance their competitiveness and equip employees with the necessary tools to support the change (Morandini et al., 2023; Li, 2022). Furthermore, artificial intelligence impacts not only performance and satisfaction but also managerial HR practices (Faye-Schjøll and Høye, 2020). The use of AI in human resources allows for anticipating employee departures and identifying factors contributing to their stress and dissatisfaction (Faye-Schjøll and Høye, 2020). AI also provides platforms that foster employee engagement and a sense of belonging. These tools create an attractive, collaborative work environment and drive business growth (Bhatia, 2018).

The blue cluster in the network visualization highlights the complex relationship between technology adoption in workplaces and its effects on employee, on autonomy and on health, revealing both opportunities and challenges. The integration of advanced technologies such as AI and automation often shifts employees from routine tasks to more complex and intellectually stimulating roles (Khogali and Mekid, 2023), which can significantly enhance job satisfaction and operational efficiency (Shwedeh et al., 2023; Shwedeh et al., 2023). However, this transition also introduces significant mental health risks, including job security stress and the pressure to quickly acquire new skills (Ghani et al., 2022) To effectively navigate these challenges, organizations must adopt a balanced approach by implementing comprehensive training programs that not only strengthen employees’ skills but also prepare them psychologically for technological changes (Morandini et al., 2023). Additionally, supporting employee well-being through accessible mental health resources, such as counseling and stress management workshops, and promoting policies that encourage work-life balance is crucial for alleviating work-related stress and preventing burnout (Bello et al., 2024). Furthermore, maintaining ethical standards in deploying new technologies by ensuring transparency, fairness, and data confidentiality is essential to build trust and foster an inclusive and supportive work environment (Potter and Doris, 2024). Collectively, these strategies can help organizations leverage technological advancements to improve productivity while simultaneously ensuring that employee autonomy and health are protected and promoted.

To conduct the word frequency analysis, we used Nvivo software. The full text of the 25 selected articles was imported into the software, and the abstracts and keywords were manually encoded into a node titled “Abstracts and Keywords.” A word frequency query was then executed, and the “word cloud” option was selected to generate a graphical representation of the 100 most frequently used words in the articles. Figure 6 illustrates the most commonly used words in the abstracts and keywords of the 25 selected articles. Words appearing in larger, bold fonts, such as “employee,” were used more frequently compared to smaller words. The term “employee” stands out as a key word in the analysis. Employees are the primary actors in the digital transformation of companies. Their level of motivation, acceptance of change, and engagement are critical factors that can significantly influence the success of a company’s digital transition (Ye et al., 2024).

To create the word cluster analysis, we also used Nvivo software. The full text of the 25 selected articles was integrated, excluding the abstracts and keywords. We then ran a “Word Frequency” query, with the results visualized as clusters to display the connections between frequently occurring words in the articles. Figure 7 illustrates the results, highlighting key associations between specific terms. For instance, the words “psychological” and “need” appeared together in a sentence, as did “career” and “collaboration.” Other notable associations include “experience” with “negative,” “employee” with “performance,” “business” with “engagement,” and “ethical” with “impact.” The pair “innovation” and “creative” is also of interest, appearing after “intelligence” and before “behavior.” Furthermore, the terms “support” and “help” often appeared together, preceded by “organizational,” while “automation” and “factors” were frequently preceded by “satisfaction.”

In order to identify the methodological approaches used in the articles, we first conducted an in-depth reading of the papers, followed by manually categorizing the articles in an Excel table according to the approaches employed (see Figure 8). This analysis allowed us to identify three main methodological approaches adopted by the authors. First, literature reviews in various forms were undertaken by 10 articles: narrative review (Khogali and Mekid, 2023; Alavi and Habel, 2021; Bankins, 2021; Chang, 2020), systematic review (Cramarenco et al., 2023; Mer and Virdi, 2023; Mera and Srivastavab, 2023; Zhang et al., 2022) and descriptive and exploratory literature review (Gorovei, 2020). Second, the quantitative approach was employed in 12 articles. For instance, some authors (Stamate et al., 2021; Jeong et al., 2024; Jiang et al., 2022; Kinowska and Sienkiewicz, 2022; Yin et al., 2024) conducted research using Structural Equation Models (SEM), while others (Nazareno and Schiff, 2021; Henkel et al., 2020; Kong et al., 2023; Xiao et al., 2023) performed regression analyses. Lastly, the mixed-methods approach was adopted by 3 articles. For example (Brougham and Haar, 2018) analyzed an open-ended question within a quantitative questionnaire, while (Konuk et al., 2023; Correia Loureiro et al., 2023) conducted semi-structured interviews followed by a quantitative study.

To create the table presenting the theories utilized in the various articles, we thoroughly read all 15 articles (from the 25 initially selected, we excluded 10 that conducted literature reviews and kept only 15 that carried out empirical studies, whether qualitative, quantitative, or mixed). We manually filled out the table, counting the number of articles in which each theory was applied, and specifying the authors of those articles. We identified 20 theoretical frameworks used across these articles, which we synthesized in Table 2. Interestingly, only the first four theoretical frameworks were discussed in two or more articles, while all other theories were employed by different authors without repetition across other studies.

Similarly, for sectoral analysis, we reviewed the 15 empirical studies and created a table showing the percentage of studies focused on various sectors. Of these, 8 conducted research in varied sectors, while 7 targeted specific ones. A bar chart was created to represent the distribution of sectors across the 7 targeted studies (see Figure 9).

In this section, we will outline the primary themes explored and the key findings reported by the authors through narrative descriptions. It’s noteworthy that while a few articles may be categorized thematically in pairs or small groups, the majority of articles introduce unique themes not revisited in subsequent works. This characteristic underscores the subject’s recent emergence and the field’s wide-open terrain, ripe for further investigations aimed at validating or challenging earlier findings. With the exception of a few unique cases, the analyzed articles can be categorized into five thematic groups. Firstly, a set of articles focuses on employees’ attitudes and perceptions regarding the opportunities and threats posed by AI, and how these perceptions impact their workplace well-being. Regarding employees’ attitudes and perceptions of the opportunities and threats posed by AI, the findings are not homogeneous. Some studies highlight that general skepticism toward job displacement from automation is prevalent, with greater awareness of STARA (Smart Technology, Artificial Intelligence, Robotics, and Automation) correlating with reduced organizational commitment and career satisfaction, as well as higher turnover intentions, depression, and cynicism among employees (Brougham and Haar, 2018). Other findings indicate that while job-replacement anxiety does not significantly impact psychological well-being (PWB), AI learning anxiety is associated with decreased PWB (Konuk et al., 2023). Conversely, some results show a positive link between the perception of AI opportunities and workplace well-being (WWB), partially mediated by informal learning in the workplace (ILW), with unemployment risk perception (URP) moderating this relationship (Xu et al., 2023). Trust in AI enhances both employee well-being and supervisor-rated productivity through improved employee-AI collaboration, particularly among those with high protean career orientations (Kong et al., 2023). Overall, general mental ability (GMA) shapes perceptions of AI’s usefulness, which affects attitudes toward the technology and ultimately influences employee satisfaction, the fulfillment of basic psychological needs, and overall psychological health (Stamate et al., 2021). Secondly, another group addresses the effects of AI and automated technologies on employee well-being, particularly emphasizing job stress and its negative impact on physical and mental health, as well as employees’ behavioral responses to this stress. Thus, the adoption of automated technologies at work consistently affects worker well-being in terms of job stress and overall health, with these effects being more pronounced in jobs at higher risk of automation (Nazareno and Schiff, 2021). Also, there is a strong association between AI adoption, heightened job stress, negative impacts on physical health, and the significant moderating effect of coaching leadership in alleviating AI-related stress (Jeong et al., 2024). Benign stress from AI adoption does not directly affect employee happiness but does so indirectly through employee engagement, which positions engagement as a key mediator between AI adoption and overall well-being (Correia Loureiro et al., 2023). Furthermore, employees with higher anxious attachment are more sensitive to AI interaction and may react by maladapting behavior by isolating themselves, experiencing insomnia or increased alcohol consumption due to their need for affiliation and feelings of loneliness (Tang et al., 2023). Providing service employees with AI tools for emotion recognition may initially raise stress levels, but also benefit well-being by raising employees’ sense of goal achievement through enhancing their ability to manage customer emotions (Henkel et al., 2020). A third group examines the contentious impact of AI adoption on job autonomy and its subsequent effects on employee well-being. The findings indicate that AI adoption promotes workplace flexibility, employee confidence, and job autonomy, all of which positively contribute to Employee Mental Health and Well-Being (EMHWB) (Shaikh et al., 2023). Conversely, the application of automated and algorithmic management practices significantly affects job autonomy, leading to indirect and potentially undesirable effects on well-being (Kinowska and Sienkiewicz, 2022). Some other articles investigate the positive influence of AI adoption on employee well-being through enhanced self-efficacy. Smart technology plays a crucial role in enhancing employee self-efficacy, which in turn has a positive influence on their overall well-being (Jiang et al., 2022). AI assistants with high intelligence not only indirectly boost employees’ AI-enabled innovation behaviors by enhancing creative self-efficacy—an effect that is amplified in organizations with greater AI readiness—but they can also have a negative impact on these behaviors through increased awareness of Smart Technology, Artificial Intelligence, Robotics, and Algorithms (STARA), particularly when organizational readiness for AI is lower (Yin et al., 2024). Lastly, a final group explores how AI contributes to tailoring workstations to meet employees’ needs, thereby improving their well-being. The evolution of the Internet-of-Things (IoT) and artificial intelligence has significantly contributed to the development of smart offices capable of understanding employees’ contexts and adapting to their needs (Zhang et al., 2022). Furthermore, job crafting mediates the positive impact of AI-enabled HR analytics on enhancing employee resilience, which, as we know, leads to employee well-being, while there is a positive moderating effect of HRM system strength on the link between AI-enabled HR analytics and job crafting (Xiao et al., 2023). The results discussed above primarily focus on the impact of AI use on employee well-being, as this constitutes the central research question of the paper. However, it is worth extending this inquiry to explore the potential effects of AI on employee productivity and performance. Consequently, in the following paragraph, we will briefly outline the key findings from the articles analyzed that address these additional dimensions, providing a more comprehensive understanding of AI’s broader organizational implications. Indeed, the question of how AI affects employee performance or productivity adds another important dimension to the discussion of well-being. For example, AI adoption promotes workplace flexibility, employee trust and autonomy, all of which contribute positively to EMHWB and, consequently, employee productivity (Shaikh et al., 2023). In addition, trust in AI positively influences supervisor-assessed productivity through improved collaboration between employees and AI (Kong et al., 2023). In terms of employee performance, smart technologies have a significant impact on company trust, self-efficacy, employee well-being and learning performance (Jiang et al., 2022). At the same time, the application of automated and algorithmic management practices significantly affects professional autonomy, with indirect and potentially undesirable effects on well-being and performance (Kinowska and Sienkiewicz, 2022).

Through an in-depth analysis of existing literature and using the Nvivo14 software, we have identified several research perspectives that can provide avenues for future studies on employee well-being and artificial intelligence. Figure 10 depicts a hierarchical diagram illustrating research perspectives identified through thematic analysis conducted with Nvivo software, we examined the “discussion of results and research perspectives” sections from all 25 articles. Major research perspectives were encoded into nodes (e.g., “asking other research questions” “adopting other methodological approaches“). Within each node, sub-nodes were created to capture more detailed suggestions. Using Nvivo’s ability to count and visualize references (articles encoded in each node), we identified similarities across the articles in terms of suggested research perspectives. Several articles mentioned the same research perspectives, such as the need for empirical studies in other countries (8 articles) or studies targeting other groups (7 articles). Once the encoding was complete, we performed a “Hierarchical Diagram” query, which visually mapped the distribution of research perspectives across the articles. The size of each rectangle in the diagram represented both the number of articles addressing a particular perspective and the volume of text devoted to it. Larger rectangles indicated greater emphasis on those perspectives. For instance, the need for posing additional research questions emerged as a key focus, as seen by the prominence of its corresponding rectangle in the diagram. This visual representation allowed for an intuitive understanding of which research perspectives were prioritized across the literature. While Nvivo does offer the ability to generate diagrams that include both nodes and sub-nodes, the result was overly complex and hard to interpret. To address this, we refined the diagram manually using Word, applying graphic elements to incorporate key sub-nodes (or sub-themes) into the corresponding rectangles. This approach helped to clarify, for example, which additional research questions were proposed by the authors or which new target groups were suggested for future empirical studies. The goal was to create a clear, concise visual summary, allowing readers to quickly grasp the main points before engaging with the detailed text that followed. In the upcoming text, we will elaborate on the various research perspectives referenced by the authors of the 25 selected articles.