Unpacking the Black Box: How AI Capability Enhances Human Resource Functions in China’s Healthcare Sector

Abstract

Artificial intelligence (AI) is transforming organizational functions across sectors; however, its application to human resource management (HRM) within healthcare remains underexplored. This study aims to unpack the black-box nature of AI capability’s impact on HR functions within China’s healthcare sector, a domain undergoing rapid digital transformation, driven by national innovation policies. Grounded in resource-based theory, the study conceptualizes AI capability as a multidimensional construct encompassing tangible resources, human resources, and organizational intangibles. Using a structural equation modeling approach (PLS-SEM), the analysis draws on survey data from 331 professionals across five hospitals in three Chinese cities. The results demonstrate a strong, positive, and statistically significant relationship between AI capability and HR functions, accounting for 75.2% of the explained variance. These findings indicate that AI capability enhances HR performance through smarter recruitment, personalized training, and data-driven talent management. By empirically illuminating the mechanisms linking AI capability to HR outcomes, the study contributes to theoretical development and offers actionable insights for healthcare administrators and policymakers. It positions AI not merely as a technological tool but as a strategic resource to address talent shortages and improve equity in workforce distribution. This work helps to clarify a previously opaque area of AI application in healthcare HRM.

1. Introduction

After more than 60 years of evolution, artificial intelligence (AI) has become ubiquitous in the past decade [1,2]. Today, with the advent of ChatGPT-style generative AI systems, there has been a phenomenal upward trend in the development of, and investment in, AI.

In recent years, the State Council of China and central departments have provided support and paid great attention to the prosperous development of a new generation of artificial intelligence technologies [3]. In July 2021, the Ministry of Industry and Information Technology issued the “Three-Year Action Plan for the Development of New Data Centers”, aiming to enhance national digital infrastructure through greener, more efficient, and regionally balanced computing systems. As a direct result, by early 2024, China had established eight national data hub regions and over ten data center clusters, significantly increasing nationwide computing power. According to official data, the total computation capacity surpassed 200 EFLOP/s by the end of 2023—laying the foundation for large-scale AI applications across sectors, including healthcare [4]. Currently, many AI products have begun to be applied in the field of human resource management.

Moreover, the application of AI developed by China covers many sectors such as finance, medicine, education, industry, gaming, and law. Generative AI has more applications in consumer-side scenarios, such as gaming, law, education, e-commerce, etc., while maturity is low in business-side scenarios such as healthcare, finance, and industry. Artificial intelligence has been announced by many as the next source of business value [5]; therefore, it is very important for companies to research and study the AI capabilities. At the same time, the emergence of AI has also had an impact on human resource management (HRM). In the past 20 years, the use of AI technology in human resource management has been an emerging trend [6]. The focus of initial research was more on presenting and explaining the concept of AI. Most articles studied the impact of AI in the general field of human resources or focused on the participation and contribution of AI in the field of medicine. Due to differences between different sectors, there is currently a lack of research on the impact of AI on human resource management in the healthcare sector. Therefore, the innovation of this study is to examine the impact of AI capabilities on human resource functions in China’s healthcare sector. On the one hand, it is hoped that this study will help to more clearly understand the particularities of the human resource sector in China’s healthcare sector. On the other hand, it is hoped that the study will demonstrate that AI has a positive and significant impact on human resource functions in the healthcare sector and can help both academics and executives, providing them with new ideas to better face the arrival of the artificial intelligence era.

Recent international evidence shows that AI adoption trajectories are bifurcating across sectors and skill profiles, with sophisticated prediction, natural language, and pattern recognition systems diffusing more rapidly where high-quality data assets, complementary digital infrastructure, and absorptive human capital already exist [7,8,9].

At the same time, empirical studies reveal heterogeneous productivity realization. While firm-level gains in innovation output and process efficiency materialize in early-adopting organizations, macro-level productivity lags can persist because complementary organizational capabilities, governance routines, and reskilling investments diffuse more slowly (the so-called “AI productivity J-curve”) [8,10,11].

These dynamics sharpen strategic questions for HRM in knowledge-intensive and service domains, including healthcare, where data richness coexists with stringent ethical, privacy, and reliability constraints on algorithmic deployment [12,13].

Moreover, research on AI capability underscores that performance benefits emerge not merely from acquiring advanced models but from orchestrating tangible (data pipelines, scalable compute), human (domain experts, hybrid “translator” roles), and intangible (governance, change agility, risk tolerance) resource bundles into path-dependent, hard-to-imitate configurations [14,15,16].

Concurrently, labor market studies document asymmetric task reallocation. Routine, codifiable tasks are increasingly automated, while demand rises for non-routine analytical, creative, and empathetic tasks that are complementary to AI systems [17,18,19]. This “task polarization” interacts with organizational job design and talent development strategies, elevating the salience of adaptive learning and continuous upskilling in sustaining equitable workforce outcomes [17,20]. In parallel, sectoral analyses caution that without deliberate capability building, automation can amplify skill-biased inequality and erode middle-skill roles, thereby challenging sustainability and decent work objectives embedded in international development agendas [7,21].

From a human resource (HR) perspective, the net organizational value of AI deployment is partially mediated by employee productivity and engagement outcomes. Structural equation modeling evidence indicates significant direct and indirect performance effects of AI through enhanced employee productivity, while parallel hospitality and services research highlights contingent risks to engagement, commitment, and well-being when job insecurity perceptions or technostress are unaddressed [14,22,23].

In the healthcare domain, Zheng et al. [24] observed that the implementation of AI tools in clinical training contexts was associated with marked performance improvements among medical residents, most notably among women. The proposed mechanism centers on real-time assistance across clinical, diagnostic, and follow-up processes, which appears to lessen cognitive load and refine decision-making. These observations reinforce the proposition that AI can operate as a valuable complementary instrument in settings where both accuracy and immediacy are essential.

Thus, strategic HRM in healthcare must integrate capability maturation, ethical governance, and proactive psychological contract management to convert algorithmic predictions into sustained, inclusive human capital advantages [12,25].

Finally, emerging governance scholarship emphasizes responsible AI frameworks—encompassing transparency, data lineage, security, and inclusive design—as dynamic complements that reduce adoption frictions, mitigate alienational contract risks, and support the iterative learning loops required for safe clinical and administrative scaling [15,26,27,28].

This synthesized perspective justifies deepening the present study’s focus on how AI capability shapes human resource functions in China’s healthcare sector and clarifies the mechanisms through which value capture and equitable workforce outcomes can co-evolve.

China’s healthcare system faces a set of entrenched human resource pressures that heighten the relevance of developing AI-enabled HR capabilities. Coverage has broadened. However, shortages of qualified clinicians and allied professionals persist, and talent remains unevenly distributed—most notably between well-resourced eastern urban centers and under-served western or rural areas. High workload intensity, administrative and documentation burdens, and constrained advancement pathways contribute to burnout and rising turnover in critical skill groups. At the same time, legacy organizational structures in public hospitals and patchy, siloed digital infrastructures slow the smooth incorporation of intelligent tools into core HR processes. These conditions complicate effective recruitment, targeted training, and agile talent deployment, and they argue for a capability-oriented approach rather than ad hoc technology adoption. Strategically cultivating multidimensional AI capability—spanning data quality, technical and hybrid skills, governance, and change readiness—offers a pathway to mitigate staffing gaps, promote more equitable service provision, and strengthen institutional resilience as digital transformation deepens [29].

Despite accelerating national policy support for AI and a burgeoning general HRM literature, three interrelated gaps remain unresolved for China’s healthcare sector: (i) a theoretical gap—limited empirical disentangling of how a multidimensional, third-order AI capability (spanning tangible, human, and intangible resources) translates into HR functional performance; (ii) an empirical context gap—scarce sector-specific, organization-level evidence from Chinese hospitals where talent shortages, equity concerns, and stringent data governance heighten the need for nuanced capability measurement; and (iii) a practical gap—insufficient guidance for hospital administrators and policymakers on which complementary resource bundles (data pipelines, hybrid skills, change and governance routines) most strongly associate with enhanced recruitment, training personalization, and talent deployment. Addressing these gaps, the present study operationalizes AI capability as a higher-order composite and tests its effect on HR functions, thereby illuminating internal value creation mechanisms and informing capability-building strategies.

This study offers two key contributions. First, it provides empirical validation of AI capability effects using real-world data from Chinese hospitals—a sector and national context where such evidence remains limited. As noted by Böhmer and Schinnenburg [30], empirical investigations into AI-driven HRM remain scarce, especially in sector-specific and public service environments. Second, by confirming the structure and predictive validity of the third-order AI capability model originally proposed by Mikalef and Gupta [5], the study strengthens its generalizability and measurement robustness across sectors and geographies.

The main research question is whether AI capabilities are having a positive impact on human resource functions in China’s healthcare sector. To address the research question, Partial Least Squares Structural Equation Modeling (PLS-SEM) is applied to investigate the hypothesis of the AI capability and human resource functions model. The research findings reveal that AI capability has a positive and significant influence on human resource functions. The findings of this study have important implications for academics, executives, human resource professionals, and policymakers in China’s healthcare sector, who can leverage artificial intelligence technologies to optimize and enhance the organizational performance of human resources.

In the following section, this study conducts a review of the existing literature (Section 2), including theoretical background, AI, human resources, and the structure of human resources in China’s healthcare sector. In this way, hypotheses are formulated based on this theoretical support. Next, the research methods and database are presented (Section 3). The Section 4 contains the research results. Finally, the discussion and conclusions of the study, its limitations, and future directions for research will be presented.

2. Related Work

2.1. Artificial Intelligence and Its Capabilities

The history of AI research dates back to the original Turing Test of 1950, which is still considered a benchmark for determining whether an artificial system possesses intelligence or not. When a person interacts with both a human and a machine and is unable to distinguish between the human and the machine, the machine is deemed intelligent [31]. Subsequently, John McCarthy, one of the pioneers of artificial intelligence, first defined the term AI in 1955, arguing that the goal of AI is to develop machines that behave intelligently [32].

The development of artificial intelligence can be roughly divided into three stages. The first stage (1950s to 1980s) saw the emergence of programmable digital computers based on abstract mathematical reasoning and symbolism, which developed rapidly. However, due to the difficulty in formalizing many concepts and insufficient processing power, AI was met with skepticism and criticism. The second stage (1980s to 1990s) was characterized by the rapid development of expert systems and significant advances in mathematical models; however, it faced challenges due to limitations in the knowledge acquisition and reasoning capabilities of these systems, high development costs, and opposition to symbolic processing models, leading to another period of stagnation. The third stage (early 21st century to present) has been a prosperous era for AI, benefiting from increased computational performance and the rapid spread of Internet technology; it has been applied in numerous fields. In particular, in 2015, Google’s AlphaGo program defeated the world champion in the board game Go, marking a renaissance of artificial neural networks in the form of deep learning [31]. The emergence of ChatGPT in late 2022 further demonstrated the new possibilities and challenges that AI can bring to various industries.

After more than 60 years of research, AI is now a complex discipline that spans multiple fields and is a branch of computer science. Defining AI from the perspective of computer science as a discipline aimed at modeling human cognitive functions, including perception, reasoning, learning, and decision-making capabilities. Mahalakshmi et al. [33] emphasizes that AI is not limited to a single field but is widely used in industry and everyday life. Currently, there are numerous definitions of AI in the academic community; scholars have not reached a consensus and there is no universally accepted definition [34]. Scholars have defined AI from their own research perspectives, but with minor differences, considering it as a computer system with human-like intelligence [35].

In conclusion, regardless of whether a unified definition of AI can be reached, it is undeniable that the development of AI in recent years has provided new technological means to endow machines with intelligence, enabling them to think like humans and perform some tedious tasks in place of humans. Looking back at the changes of the past few years, AI, as a rapidly developing field, has opened the door to innovation, bringing great potential and opportunities to all areas of life.

Building upon definitional plurality, contemporary research reframes AI less as a monolithic technology and more as a layered capability stack wherein prediction engines interface with judgment, data stewardship, and organizational routines [7,14]. Under a resource-based and dynamic capabilities lens, firms derive sustained advantages by sensing emerging algorithmic opportunities, seizing them through cross-functional integration, and continually transforming socio-technical processes to embed learning feedback loops [15,27].

Empirical mediation models demonstrate that AI’s performance contribution is partially channeled through employee productivity, highlighting the socio-technical complementarity between machine outputs and human interpretive, relational, and ethical competencies [14]. Parallel sectoral syntheses in hospitality and services categorize employee outcomes into attitudinal (engagement, commitment, well-being, satisfaction, burnout) and behavioral (performance, creativity) clusters, revealing mixed valence effects contingent on surveillance intensity, training adequacy, and job insecurity perceptions [22,23].

Labor economics and innovation studies further contextualize capability development within macro patterns of routine task displacement and polarization, where automation augments high-skill analytical roles while pressuring mid-skill routine occupations and leaving many low-skill service tasks resilient yet quality-constrained [17,18,19,20]. These structural shifts amplify strategic HR imperatives around reskilling, task redesign, and equitable talent pipelines in order to prevent widening wage dispersion and to align with inclusive growth goals [7,21].

At the micro level, psychological contract scholarship indicates that AI adoption can weaken the positive linkage between relational/transactional contracts and engagement, motivating the delineation of an emerging “alienational” contract type characterized by intermittency and algorithmically mediated interactions [12]. Managing this risk requires transparent communication, participatory implementation, and capability-aligned training to mitigate the technostress and job insecurity drivers of disengagement and turnover intentions [22,23,25].

Generative AI introduces an additional layer. User continuance and recommendation intentions depend on perceived information quality, system reliability, privacy concerns, and perceived innovativeness—factors that, when positive, reinforce organizational learning loops and broaden the effective scope of AI capability deployment [11,13]. Robust governance mechanisms (traceability, accountability, adaptive strategic flexibility) serve as intangible complements that stabilize trust, accelerate diffusion across functional domains, and align technical affordances with evolving HR workflows [26,27].

Recent bibliometric mapping of the rapidly expanding ChatGPT literature shows a still-fragmented collaboration structure (multiple small author sub-networks), strong concentration in a handful of institutions (e.g., Stanford, Harvard) and countries (the United States is leading, with high betweenness roles for the United Kingdom, India, and Spain), and three dominant thematic clusters—data processing applications, user behavioral intention studies, and differential diagnosis use cases—underscoring both multidisciplinary breadth and nascent coordination gaps in scholarly knowledge production [28]. These structural patterns provide a useful context for interpreting rigorous post-adoption behavioral models. The continuance/recommendation framework integrating expectation confirmation, information/system quality, privacy concerns, and perceived innovativeness achieves high explanatory power, highlighting how informational and experiential quality, moderated by risk perceptions, convert early experimentation into sustained, advocacy-oriented usage trajectories for generative AI systems [11].

Collectively, this corpus positions AI capability as an emergent, composite phenomenon whose organizational efficacy depends on multi-level alignment: (a) macro labor market transition management; (b) meso-level orchestration of data, talent, and governance resources; and (c) micro-level preservation of engagement and psychological contract quality in augmented task ecosystems [12,14]. This integrated framing underpins the subsequent hypotheses by clarifying why heterogeneity in AI capability maturity is expected to map onto differential human resource function performance outcomes.

The use and assistance of AI in organizational settings have inspired scholars to investigate AI capabilities. Like other emerging technologies and information systems, it is widely recognized that the adoption and exploitation of these technological innovations is an effort that the entire company must undertake and requires significant attention in several key areas [36,37]. Based on this perspective of leveraging emerging technologies as a key organizational capability, companies need to develop an information technology (IT) capability in order to effectively utilize new and emerging technologies [37]. It can be said that AI, like other IT functions, received considerable attention in early studies in the literature [38,39].

In today’s business environment, AI is not only a technological trend but also a strategic resource that has a profound impact on companies’ innovation capabilities and market competitiveness. Mikalef and Gupta [5], in their pioneering study, proposed a comprehensive conceptual framework for AI capabilities, based on Resource-Based Theory (RBT), and developed corresponding tools for their measurement. They consider AI capabilities to be a company’s ability to select, coordinate, and utilize its AI-specific resources, a capability that encompasses both technical depth and organizational breadth [5]. The core of AI capabilities lies in its multidimensional structure, which includes tangible resources (such as high-quality data, advanced technology, and infrastructure), human skills (covering technical expertise and business acumen), and intangibles (like internal coordination mechanisms, change capacity, and risk acceptance). The introduction of this concept provides a theoretical basis and practical guidance for companies to build and enhance their AI capabilities.

However, due to the lack of systematic studies on AI capabilities, there is currently considerable confusion about how to precisely define these capabilities. However, what we have learned so far is that the application of AI technology can not only unleash employees’ innovative potential by automating repetitive tasks but also expand human cognitive abilities and problem-solving through augmented intelligence.

The complexity of AI capabilities indicates that they go much further. To investigate the impact of AI on organizations’ human resources, we need to explore its dynamic capabilities. To date, AI has been widely recognized as a fundamental source for transforming production modes and strengthening learning processes, thereby improving and expanding upon existing decision-support systems [5,40]. AI should be considered a business capability, not just a technological advancement [41]. Capabilities can be defined as complex skills and accumulated knowledge that are utilized through organizational processes, enabling the company to coordinate activities and leverage its assets [42]. Therefore, data constitute a valuable asset, the prolonged and cumulative use of which can contribute to the creation of an AI ecosystem within the company [43].

Despite the fact that in two extensive global surveys of senior executives, the lack of AI capabilities was identified as the main challenge for its implementation [44], we can still take advantage of the combination of humans and machines. When AI provides tentative solutions, it is possible to create concrete and valuable outcomes through human interpretation, argumentation, and action [45,46,47]. We believe that in the future, AI capabilities will continue to surprise everyone and achieve breakthroughs in existing perceptions.

The difference between AI technology and applications compared to other technological advancements lies in its ability to imitate cognitive functions, perform tasks in a manner similar to humans, and possess learning and self-correction capabilities [48]. Mikalef and Gupta [5] define AI capability as a firm’s ability to organize, integrate, and effectively utilize its AI-based resources. The organizations capable of managing AI-specific resources have the ability to leverage various AI technologies and create value [49,50]. Grant [51] classifies capability resources into tangible resources, human skills, and intangible resources. A company’s capability is the result of the joint work of the resource team. Coordinating the development of existing resources with the development of resources and capabilities to obtain a future competitive advantage is a delicate task [51].

The introduction of the concept of AI capability aims to better determine what resources organizations should acquire and develop in order to benefit from AI technologies [5]. Although there is great enthusiasm for this technology, many organizations face challenges in obtaining performance benefits from its use [52,53]. It is important that, in addition to the technology itself, we also focus on the resources that organizations need to develop. These complementary organizational resources are necessary to establish firm-specific and hard-to-imitate AI capabilities [54].

2.2. The Application of an AI Capability in Human Resources Within China’s Healthcare Sector

With the continuous advancement of AI technology, its application in the field of human resource management has become a topic of great academic interest. In 2017, Google launched its recruitment tool Google Hire, designed to help companies track candidate information and efficiently coordinate the interview process. In 2020, the MIT Media Lab developed an online trainer that, based on the personal characteristics of employees, offers immersive and personalized training experiences. On the other hand, IBM’s Watson system, a well-known global software technology development company, is capable of comprehensively evaluating the current performance of employees and their potential for future development.

Some of the effects of AI on organizations include shifts in power; reassignment of decision-making responsibilities; cost reduction and service improvement; and changes in staffing and downsizing [55]. AI has played a significant role in business management decision-making. While many companies are still waiting to see how AI develops, some have already incorporated AI into their daily management decision-making processes.

AI can position and develop the company’s internal employees by building talent quality models, identifying and selecting key fields, and other technologies [56]. The operation and development of businesses should change with current data [57]. The rapid development of AI is transforming traditional organizational structures and labor relations, giving rise to innovative management models and concepts that are driving organizational management towards intelligent transformation.

In China, the application of AI technology in the field of human resources in the healthcare sector is gradually increasing and covers various aspects, from improving the efficiency of medical services to optimizing human resource management. The development of AI technology has enabled companies to improve their recruitment, training, and development procedures, as well as performance management. In the past, hiring managers were required to spend hours sorting through applications and resumes to find qualified candidates [58]. Today, artificial intelligence is used to optimize performance management, analyzing data from various sources such as electronic health records, patient satisfaction surveys, and employee feedback [59].

Moreover, AI-based recruitment tools in Chinese healthcare institutions can automatically perform multiple tasks, including resume filtering, professional title evaluation, and even conducting initial interviews with candidates [60]. The application of AI technology in the healthcare field can compensate for shortages in human resources and improve medical accuracy [61].

Although equity in the distribution of health human resources has improved in the health departments of Chinese provinces, equity between regions has encountered a bottleneck; the distribution of health human resources remains relatively scarce in the western regions [62]. By leveraging the development and progress of digitalization in the health sector, it is possible to achieve the optimization and efficient utilization of medical resources. Establishing a national database of medical talents could allow for the unified management of medical personnel information, enabling transparency and information exchange. This would promote the mobility and rational distribution of talent, thus addressing the lack of equity in the distribution of human resources.

In summary, the capabilities of artificial intelligence have a profound impact on human resource functions in China. The health sector, in particular, needs to overcome challenges in human resource functions such as talent shortages, brain drain, lack of training, and inadequate performance evaluation. Artificial intelligence can enhance the efficiency of human resource management in the health sector through smart recruitment, personalized training, data-based performance evaluation, health monitoring, and the optimization of resource allocation. This will improve the work experience and professional development of medical staff.

With this background, we state the following hypothesis:

H1: 

An AI capability positively influences human resource functions in the health sector in China.

According to the above ideas, Appendix A, Figure A1 represents the general conceptual framework of AI capability and human resource functions.

3. Materials and Methods

In this section, we focus on the most relevant methodological aspects of this research. We start with the research design and the appropriateness of the study methods used, followed by a justification of the study population and the selected sample, data collection, measurement of variables, and data treatment.

A quantitative analysis was conducted based on a representative sample of healthcare centers and the confirmation of research hypotheses through the PLS-SEM technique. We tested the research hypotheses with PLS-SEM—structural equation modeling (SEM), estimated with partial least squares (PLS). In social sciences and business administration, the scientific community has commonly accepted the dominant logic of quantitative analyses based on Multiple Regression Analysis (MRA) and structural equation models. For its part, the PLS estimation method is particularly interesting in the early stages of theory development [63], or when researchers include scales that have been tested and validated in previous works, or due to a relatively small sample size [64,65], or finally, when the models are very complex [66]. Thus, the choice of PLS was justified by the sample size, the research objectives, and the complexity of the model.

The data were processed using SmartPLS 4.1 software [67]. For generating the descriptive statistics of the sample, SPSS, version 23, was utilized.

3.1. Sampling and Data Collection

One of the first decisions in any research involves specifying and delimiting the population to be analyzed based on the problem and the main objectives of the research [68]. Focusing the research interest on the AI capability and the human resource functions operating in China’s health sector, the selected unit of analysis is the individual, that is, the healthcare professionals in China. The target population, which constitutes the universe to be studied, is made up of healthcare professionals in China.

To determine the size of the population, it is necessary to refer to figures from the Chinese health system. According to the 2022 China Health and Wellness Development Statistical Bulletin [69], as of the end of 2022, the total number of medical and health institutions nationwide was 1,032,918, of which 36,976 were hospitals, 979,768 were primary health care institutions, and 12,436 were public health professional institutions. Among the hospitals, there were 11,746 public hospitals and 25,230 private hospitals. Hospitals were divided by the number of beds: 21,904 hospitals with fewer than 100 beds, 5483 hospitals with 100 to 199 beds, 5174 hospitals with 200 to 499 beds, 2190 hospitals with 500 to 799 beds, and 2225 hospitals with 800 or more beds. The total number of health personnel in the country was 14.411 million. Of the total health personnel, 11.658 million were health technicians. Among the health technicians, there were 4.435 million doctors (assistants) and 5.224 million registered nurses.

The survey was targeted at human resource managers, financial managers, project managers, and other health professionals from both public and non-public hospitals in China. The study survey was implemented via the Chinese professional questionnaire creation platform Wenjuanxing https://www.wjx.cn/. Prior to collecting survey data, we enlisted the assistance of a hospital director to disseminate our online survey via WeChat (a Chinese messaging app). The survey outlined the study’s purpose and assured the confidentiality of respondents’ information, stating that the data would be used solely for research purposes and requesting their informed consent for data collection.

Fieldwork was conducted between January and April 2024. A total of 331 valid cases were obtained, representing a response rate of 10.58% relative to the total number of potential users to whom the survey was sent. The sampling error for the 331 responding workers was 5.39% (95% confidence level, p = q = 0.5), which ensured sufficient representation of the entire population of health workers in China. The power value (0.82) was obtained with G*Power 3.1.9.7 [70] using a post hoc test for a point-biserial correlation, with α = 0.05 and an effect size |ρ| = 0.14, which corresponds to a small effect size in Cohen’s classification [71]. Because 0.82 exceeds the conventional 0.80 cut-off, the sample’s statistical power was shown to be adequate.

3.2. Sample Characteristics and Descriptive Analysis

In terms of demographics,

Table 1. Description of the respondent sample (n = 331).

	Frequency	Percentage
Gender
Man	88	26.6%
Woman	214	64.7%
I prefer not to disclose	29	8.8%
Age
19–29 years old	89	26.9%
30–40 years old	95	28.7%
41–50 years old	85	25.7%
51–60 years old	55	16.6%
Over 60 years old	7	2.1%
Education level
Did not complete primary/secondary education	0	0%
High school diploma/Vocational training	9	2.7%
Three-year college degree/Some Bachelor’s degree	98	29.6%
Bachelor’s degree	200	60.4%
Master’s degree	19	5.7%
Ph.D./Doctorate	5	1.5%
Classification Statutory Staff of
Health	239	72.2%
Non-health (Management and Services)	92	27.8%
Duration of employment contract
Permanent/Long-term temporary	258	77.9%
Short-term temporary	73	22.1%
Ownership type
Government-owned	130	39.3%
Privately owned	201	60.7%
Hospitals and health centers
Shandong Hospital of TCM	27	8.2%
Taixing No. 2 People’s Hospital	81	24.5%
Yangzhou maternal and childcare service centre	11	3.3%
Yangzhou Hospital of TCM	18	5.4%
Friendliness Hospital. Yangzhou	194	58.6%

presents the distribution characteristics of the sample. Among the respondents, 88 are male, representing 26.6%; 214 are female, representing 64.7%. Additionally, 29 individuals preferred not to disclose their gender. We respected each respondent’s choice regarding the disclosure of their personal information. The age distribution was relatively balanced for those under fifty years old. There were 55 individuals in the age range of 51 to 60 years, representing 16.6%, and 7 individuals over 60 years old. In China, the legal retirement age is 65. We observed that the survey results included both retired individuals and those rehired after retirement (in traditional Chinese medicine, older individuals are often rehired due to their greater experience).

Regarding educational level, the majority of respondents were concentrated in the bachelor’s or equivalent degree category, with a total of 200 individuals, representing 60.4%. The number of individuals with higher educational levels was smaller; however, this figure is expected to continue to grow in the future.

Finally, we surveyed respondents from a total of five hospitals: (1) Shandong Hospital of Traditional Chinese Medicine (TCM) (www.sdzydfy.com) (public, 3163 employees, 2451 beds, located in Jinan city with a population of 9 million); (2) Taixing No. 2 People’s Hospital (public, 765 employees, 700 beds, located in Taixing city with a population of nearly 1 million); (3) Yangzhou Maternal and Child Health Care Service Centre (www.yzfybj.com) (private, 831 employees, 450 beds, located in Yangzhou city with a population of nearly 4.5 million); (4) Yangzhou Hospital of TCM (www.yzszyy.com) (public, 628 employees, 700 beds, located in Yangzhou city); and (5) Friendliness Hospital Yangzhou (www.yzyhyy.com) (private, 565 employees, 496 beds, also located in Yangzhou city). Of these, 130 respondents were from public hospitals, representing 39.3%, and 201 were from private hospitals, representing 60.7%.

3.3. Selection and Measurement of Variables

The instrument consisted of a questionnaire comprising 51 items across nine dimensions (Appendix B,

Table A1. Questionnaire of latent constructs.

Constructs	Items	Sources
AI Capability		[5]
Tangible
Data	D1. We have access to very large, unstructured, or fast-moving data for analysis.
	D2. We integrate data from multiple internal sources into a data warehouse or mart for easy access.
	D3. We integrate external data with internal to facilitate high-value analysis of our business environment.
	D4. We have the capacity to share our data across business units and organizational boundaries.
	D5. We are able to prepare and cleanse AI data efficiently and assess data for errors.
	D6. We are able to obtain data at the right level of granularity to produce meaningful insights.
Technology	T1. We have explored or adopted cloud-based services for processing data and performing AI and machine learning.
	T2. We have the necessary processing power to support AI applications (e.g., CPUs, GPUs).
	T3. We have invested in networking infrastructure (e.g., enterprise networks) that supports efficiency and scale of applications (scalability, high bandwidth, and low-latency).
	T4. We have explored or adopted parallel computing approaches for AI data processing.
	T5. We have invested in advanced cloud services to allow complex AI abilities on simple API calls (e.g., Microsoft Cognitive Services, Google Cloud Vision).
	T6. We have invested in scalable data storage infrastructures.
	T7. We have explored AI infrastructure to ensure that data is secured from to end to end with state-of-the-art technology.
Basic Resources	BR1. The AI initiatives are adequately funded.
	BR2. The AI project has enough team members to get the work done.
	BR3. The AI project is given enough time for completion.
Human Skills
Technical Skills	TS1. The organization has access to internal and external talent with the right technical skills to support AI work.
	TS2. Our data scientists are very capable of using AI technologies (e.g., machine learning, natural language processing, deep learning).
	TS3. Our data scientists have the right skills to accomplish their jobs successfully.
	TS4. Our data scientists are effective in data analysis, processing, and security.
	TS5. Our data scientists are provided with the required training to deal with AI applications.
	TS6. We hire data scientists that have the AI skills we are looking for.
	TS7. Our data scientists have suitable work experience to fulfill their jobs.
Business Skills	BS1. Our managers are able to understand business problems and to direct AI initiatives to solve them.
	BS2. Our managers are able to work with data scientists, other employees and customers to determine opportunities that AI might bring to our organization.
	BS3. Our managers have a good sense of where to apply AI.
	BS4. The executive manager of our AI function has strong leadership skills.
	BS5. Our managers are able to anticipate future business needs of functional managers, suppliers and customers and proactively design AI solutions to support these needs.
	BS6. Our managers are capable of coordinating AI-related activities in ways that support the organization, suppliers and customers.
	BS7. We have strong leadership to support AI initiatives and managers demonstrate ownership of and commitment to AI projects.
Intangible
Inter-departmental Coordination	Please indicate to what extent do departments within your organization engage in the following activities:
	IC1. Collaboration.
	IC2. Collective goals.
	IC3. Teamwork.
	IC4. Same vision.
	IC5. Mutual understanding.
	IC6. Shared information.
	IC7. Shared resources.
Organizational Change Capacity	OCC1. We are able to anticipate and plan for the organizational resistance to change.
	OCC2. We consider politics of the business reengineering efforts.
	OCC3. We recognize the need for managing change.
	OCC4. We are capable of communicating the reasons for change to the members of our organization.
	OCC5. We are able to make the necessary changes in human resource policies for process re-engineering.
	OCC6. Senior management commits to new values.
Risk Proclivity	RP1. In our organization we have a strong proclivity for high risk projects (with chances of very high returns).
	RP2. In our organization we take bold and wide-ranging acts to achieve firm objectives.
	RP3. We typically adopt a bold aggressive posture in order to maximize the probability of exploiting potential opportunities.
HR Functions	HRF1. IA for technology awareness is more cost-effective than other technologies.	[60,72]
	HRF2. IA technology helps HR managers to select the right candidates.
	HRF3. IA technology helps HR managers to conduct online training and development sessions for new and existing employees.
	HRF4. IA technology provides user-friendly mediums to monitor employees’ performance.
	HRF5. Tracking employees’ activity through artificial intelligence technology is more efficient and time-saving.

) and is scored on a five-point Likert scale ranging from “strongly disagree” to “strongly agree” (1 to 5), where higher scores indicate a higher level of agreement. Additionally, information on other sociodemographic variables of interest, such as age, gender, job position, profitability of the hospital, size, and number of beds, among other details, was collected.

The scale developed by Mikalef and Gupta [5] was used to measure AI capability. AI capability is distinct from other digital capabilities, such as IT capability, as its constituent resources are specific. This study conceptualizes AI capability as a third-order mode A composite, comprising the following specific AI dimensions: tangible resources, human resources, and intangible resources. These dimensions, in turn, are modeled as second-order mode A composites, encompassing eight first-order mode A composites. These are grouped as follows: (a) tangible resources: data (six items), technology (seven items), and basic resources (three items); (b) human resources: technical skills (seven items) and business skills (seven items); (c) intangible resources: inter-departmental coordination (seven items), organizational change capacity (six items), and risk proclivity (three items).

To measure human resource functions (the dependent variable), the five items of Shahzad et al. [60] from the original scale of Pillai and Sivathanu [72] were employed. These were considered a mode A composite.

These scales, originally in English, were translated into Chinese using Brislin’s [73] back-translation procedure to ensure semantic equivalence. A bilingual professional first translated the items into Chinese; another professional then re-translated the items back into English.

4. Results

The results and analyses are presented here. The form chosen to construct the higher-order model used latent variable scores, employing the “Two-Step Approach” [74,75]. For the interpretation and analysis of the proposed model in PLS-SEM, two distinct stages were developed: (1) measurement model analysis; and (2) structural model analysis. This sequence ensured that the proposed measurement scales were valid and reliable.

4.1. Measurement Model Analysis

The individual item reliability is considered adequate when the item loading is greater than 0.707 (the acceptance value recommended by Carmines and Zeller [76]). In this study, the indicators and reflective dimensions met this requirement, being well above 0.7 (

Table 2. Item loadings, construct reliability and convergent validity.

Construct	Measures	Loading	Cronbach’s Alpha	Composite Reliability	AVE
Data	D1	0.922	0.964	0.971	0.847
	D2	0.916
	D3	0.922
	D4	0.913
	D5	0.933
	D6	0.917
Technology	T1	0.915	0.974	0.978	0.863
	T2	0.940
	T3	0.926
	T4	0.934
	T5	0.909
	T6	0.952
	T7	0.928
Basic Resources	BR1	0.949	0.952	0.969	0.913
	BR2	0.964
	BR3	0.952
Technical Skills	TS1	0.918	0.975	0.979	0.869
	TS2	0.936
	TS3	0.925
	TS4	0.938
	TS5	0.938
	TS6	0.921
	TS7	0.949
Business Skills	BS1	0.927	0.975	0.979	0.870
	BS2	0.921
	BS3	0.936
	BS4	0.938
	BS5	0.943
	BS6	0.932
	BS7	0.932
Inter-departmental Coordination	IC1	0.929	0.974	0.978	0.865
	IC2	0.922
	IC3	0.936
	IC4	0.929
	IC5	0.918
	IC6	0.934
	IC7	0.943
Organizational Change Capacity	OCC1	0.909	0.966	0.973	0.856
	OCC2	0.918
	OCC3	0.940
	OCC4	0.925
	OCC5	0.944
	OCC6	0.913
Risk Proclivity	RP1	0.936	0.943	0.964	0.898
	RP2	0.951
	RP3	0.957
Tangible Resources	Data	0.926	0.940	0.962	0.894
	Technology	0.967
	Basic Resources	0.943
Human Resources	Technical Skills	0.985	0.969	0.985	0.970
	Business Skills	0.985
Intangible Resources	Inter-departmental Coordination	0.961	0.961	0.974	0.927
	Organizational Change Capacity	0.972
	Risk Proclivity	0.956
AI Capability	Tangible Resources	0.980	0.977	0.985	0.956
	Human Resources	0.978
	Intangible Resources	0.974
HR Functions	HRF1	0.903	0.950	0.961	0.833
	HRF2	0.902
	HRF3	0.937
	HRF4	0.907
	HRF5	0.914

).

The construct reliability assessment employed Cronbach’s alpha (α) and composite reliability (ρ). Both indices serve the same purpose: to measure the internal consistency of a construct, although composite reliability is more appropriate for use in PLS as it is superior to Cronbach’s alpha [77]. For both indices, 0.7 is the basic benchmark [77,78,79]. In this research, all constructs and reflective dimensions were reliable, even exceeding values of 0.8 (strict reliability). The average variance extracted (AVE) measures convergent validity; that is, whether the set of indicators represents the same latent variable. All constructs and reflective dimensions achieved convergent validity, surpassing the recommended value of 0.5 [75] (Table 2).

Finally, the results of the discriminant validity assessment (the degree to which a construct differs from others) show that for each construct, the square root of the AVE was higher than the correlations between the constructs [77] and that the loadings of the constructs are higher in their respective constructs than in the rest of them (cross-loadings) [80]. This indicated the discriminant validity of the measures used (

Table 3. Correlation matrix and discriminant validity.

Measures	1. AI Capability	2. HR Functions
1. AI Capability	0.978
2. HR Functions	0.867	0.913

). In addition, the Heterotrait–Monotrait (HTMT) value is 0.89, below the threshold of 0.90 and significantly lower than 1, which supports the discriminant validity of the variables [66].

4.2. Structural Model Evaluation

With the convergent and discriminant validity of the measurement model and its reliability ensured, the relationships between the different variables were tested. Significance testing was carried out by performing the bootstrapping procedure (10,000 subsamples) to assess the significance of the path coefficients. Thus, the t-statistic (pseudo-parametric test) and the estimated 95% percentile confidence interval for the path coefficient were obtained—if it did not contain zero, it was said that the estimated path coefficient was significantly different from zero. In this study, hypothesis 1 (H1) states that the effect of AI capability on HR functions is positive and significant (β = 0.87; the t-value exceeds the minimum level indicated by the one-tailed Student’s t-distribution with n–1 degrees of freedom, where n is the number of subsamples, at a confidence level of 99.9% and, therefore, a probability of being wrong in rejecting the null hypothesis: p < 0.001). Finally, the standardized root mean square residual (SRMR) was calculated for a composite factor model to determine the exact fit of the composite factor model. The model (total effect) achieved a composite factor model SRMR of 0.03, which indicates an appropriate fit, assuming the usual cut-off of 0.10 or a more conservative 0.08 [81]. Therefore, the hypothesis was supported (

Table 4. Causal relationships: total effect.

Hypothesis	Path Coefficient (β)	t-Value (Bootstrap)	Percentile 95% Confidence Intervals		Support
			Lower	Upper
Model (SRMR = 0.03)
H1 = AI Capability → HR Functions	0.867 ***	35.671	0.815	0.909	Yes

*** p < 0.001, (based on t (9999), one-tailed test).

, Figure 1).

5. Discussion

The findings of this study have important implications for academics, executives, human resource professionals, and policy makers in China’s health sector, who can leverage artificial intelligence technologies to optimize and enhance the organizational performance of human resources. Their adoption should be carefully planned and managed to fully exploit the benefits of this transformative technology.

5.1. Theoretical Implications

The findings of this study present several significant theoretical implications, which can be related to existing studies in the field of artificial intelligence and human resource management.

The research demonstrates that AI capability has a positive and significant impact on HR functions in China’s health sector. This finding aligns with previous studies that have explored the adoption of emerging technologies in organizational management. For example, Mikalef and Gupta [5] developed a comprehensive framework for AI capabilities based on the resource-based view theory, arguing that these capabilities enable firms to coordinate and effectively utilize their AI-specific resources.

The ability of AI to enhance HR functions suggests that organizations can increase their operational efficiency through the automation of repetitive tasks and the optimization of processes. This is consistent with studies in the literature identifying AI as a key tool for improving organizational performance through augmented intelligence and data-driven decision-making. The study also highlights the importance of both tangible resources (such as advanced technology and high-quality data) and intangible resources (such as organizational change capability and interdepartmental coordination) for the successful implementation of AI. This multidimensional approach to AI capabilities coincides with the work of various researchers who have explored how intangible resources can drive innovation and organizational performance.

The findings also underscore the challenges organizations face in terms of developing specific AI technical and business skills. This is reflected in the literature, where it is highlighted that the lack of AI capabilities is a major barrier to its effective adoption. The training and support of business leaders are essential to overcome these challenges.

Finally, this study provides a theoretical framework that can be used in future research to further explore the dynamics between AI and HR in different contexts and sectors. Understanding how AI capabilities can be cultivated and effectively leveraged remains a critical area for both theoretical and practical development.

In summary, the findings of this study not only corroborate existing theories on the integration of AI into HR management but also expand our knowledge of how these technologies can transform organizational practices in the health sector.

5.2. Practical Implications

The application of artificial intelligence technology in China’s healthcare sector has received strong support from government policies. The development of “Internet + Health” is accelerating the integration of the Internet, blockchain, IoT, artificial intelligence, cloud computing, and big data into the healthcare field, strengthening the exchange and protection of health data.

Furthermore, the planning of medical institution configurations also emphasizes the supportive role of informatization. In the Guidelines for the Planning of Medical Institution Configurations (2021–2025) [82], it is explicitly proposed to promote the deep integration of new technologies, such as artificial intelligence, with medicine, driving the construction of smart hospitals and the standardization of hospital information. This indicates that China’s healthcare sector is committed to building a more intelligent and standardized medical system.

The results of this study will help medical institutions in China to understand the importance of artificial intelligence in optimizing human resource functions. Taking favorable measures concerning each of the studied variables—data, technology, basic resources, technical and business skills, inter-departmental coordination, organizational change capacity, and risk proclivity—will positively impact HR functions operating in the Chinese healthcare sector. With the aid of intelligent tools and systems, HR leaders in the healthcare sector can carry out selection, training, and management more efficiently, thereby retaining more outstanding talent. At the same time, artificial intelligence can analyze data, monitor employee behavior, and predict staff attrition trends. With the continuous advancement of technology and ongoing policy support, artificial intelligence is expected to play an increasingly important role in the future of the healthcare sector.

5.3. Limitations and Future Research Directions

First, the number of samples selected for this study is limited, with a total of 331 questionnaires collected. The small sample size may lead to results that do not fully represent the entire medical community, introducing potential sampling bias. Second, the sample selection scope includes only five hospitals in three cities. In cities with different levels of development, the use of artificial intelligence in human resource management within the healthcare system may vary.

This study conducted a sampling survey solely within China’s medical industry. In different countries, given varying levels of economic development, healthcare standards, and policies, different scenarios may emerge, making the conclusions potentially inapplicable to other regions or industries.

We encourage future research to improve data diversity by collecting information from various aspects, such as human resource expenditures, industry efficiency improvements, and changes in human resource investment, to quantitatively analyze the relationship between artificial intelligence and the enhancement of human resource management in the medical industry. Analyzing whether artificial intelligence can improve risk resilience and optimization capability in the HR domain through its multiple roles in human resource management could be another call for research.

6. Conclusions

In China, the healthcare sector is actively responding to the call for artificial intelligence technology. The government has implemented numerous policies aimed at improving the quality and efficiency of medical services through technological innovation while optimizing human resource management in the healthcare sector. In this era of rapid AI development, the medical industry is undergoing a profound transformation. This research aims to demonstrate that AI capabilities have had a positive impact on human resource functions within China’s healthcare sector.

The results indicate that the healthcare sector in China already recognizes the importance of AI technology in medical human resource management. Through the formulation and implementation of various policies, healthcare institutions are encouraged to reform staffing and personnel systems; establish a dynamic personnel expansion mechanism; promote coordinated workforce allocation within public medical institutions in healthcare consortia; and achieve unified recruitment and management of personnel. These efforts will contribute to improving the efficiency of human resource allocation and optimizing the talent structure in the healthcare sector.

In summary, China’s healthcare policies regarding AI’s impact on human resource functions reflect comprehensive support for the digital transformation of the medical industry. Through technological innovation, the aim is to enhance the quality and efficiency of medical services, optimize healthcare resource allocation, and improve the accessibility and equity of healthcare services. With the continuous implementation of these policies, AI is expected to play an increasingly vital role in the healthcare sector, driving revolutionary changes in human resource management and providing strong technological support for the sustainable development of the medical industry.