Relationships Between AI Tools, Social Media, and Performance via Ensemble Bayesian Network: A Survey Among Chinese Lawyers

Abstract

Amidst the rapid digital transformation reshaping the legal profession globally, this study examines the interplay between AI tools, social media usage, and lawyer job performance in China. While prior research has extensively explored factors influencing the job performance of lawyers, due to the relatively small number of lawyers in China and the legal and ethical limitations in their use of social media and AI tools, systematic investigations into the roles of AI and social media in this context remain limited. We use an ensemble Bayesian network model to examine causal mechanisms, analyzing 313 questionnaires on their use of AI and social media. This study constructs a robust causal network to analyze the impacts of nine key variables, including excessive social use of social media at work, AI-supported employee training and development, AI-driven workload reduction for employees, and strain, among others. The findings reveal that AI-driven workload reduction, AI-supported leadership, and strain directly influence lawyer job performance. Notably, excessive cognitive use of social media at work (ECU) exerts the most significant impact, while strain and work–technology conflict serve as critical mediators in the relationship between ECU and performance. The ensemble Bayesian network framework not only enhances the methodological rigor of this research but also facilitates a comprehensive understanding of the complex interdependencies among the considered factors. Based on the results, practical recommendations are proposed for the optimization of the job performance of lawyers. This study contributes to the growing body of literature on lawyer job performance through the introduction of an advanced analytical approach, as well as offering actionable insights for law firms and informing legal technology legislation and policy development navigating the digital era.

1. Introduction

In the era of rapid digital transformation, internet technologies have become deeply integrated into various industries, including the legal profession [1,2]. According to ISAAC [3], internet technologies have transformed organizational learning and implementation processes, posing new challenges for law firms and lawyers in terms of adopting and applying these technologies. Research on lawyer performance is of significant importance, as it not only helps law firms to accurately assess work outcomes, optimize resource allocation and management decisions, and enhance their overall competitiveness, but also provides lawyers with clear career development directions, promotes professional growth, and ultimately drives high-quality development in the legal services industry. Studies have shown that the use of information technology in organizations significantly enhances the productivity and work performance of employees [4,5]. As a knowledge-intensive profession, those practicing lawyers can effectively improve their performance by leveraging information technology for knowledge acquisition [6]. However, the low tolerance for errors in legal work means that improper use of technology in applying knowledge may lead to mistakes, affecting judicial cases or rendering legal advice invalid [7].

Existing research on the use of information technology by lawyers has primarily focused on two areas: AI tools and social media. Based on the Competency and Quality Theory, employees’ professional and communication skills directly influence their job performance [8,9]. In this regard, social media can enhance their communication skills, while AI tools can help improve their professional capabilities [10,11,12,13,14,15,16,17,18]. With the rapid growth of computational power and data volume, AI technology has advanced significantly [11] and, in recent years, many studies have applied AI technologies—such as large language models (LLMs)—to the legal field, addressing challenges that traditional methods struggle to solve [12,13,14]. AI has made notable progress in tasks such as legal prediction [14], interpretation of legal provisions, dispute mediation, legal information extraction [15], and legal reasoning [16]. With the increasing number of judicial cases due to population growth, the application of LLMs and other AI technologies in the legal field is expected to alleviate the “more cases, fewer personnel” dilemma, thereby improving the efficiency of the judicial system [17,18]. Work-oriented AI tools have become essential for enhancing the professional capabilities of lawyers, and an increasing number of law firms are adopting AI technologies to handle their daily tasks and conduct online operations to improve performance. AI skills encompass various technologies and applications, including legal research platforms (e.g., Westlaw and LexisNexis), contract review software (e.g., Kira Systems), predictive analytics tools (e.g., Premonition and Lex Machina), and document automation software (e.g., HotDocs and ContractExpress) [19,20,21].

Simultaneously, social media platforms (e.g., X, Facebook, and instant messaging applications such as WeChat and QQ in China) have facilitated the construction of professional networks and knowledge sharing [22,23,24,25]. The use of social media enhances lawyers’ communication skills, helps to build personal branding, and expands business channels, thereby improving business performance [26,27,28]. Therefore, in the digital age, how lawyers effectively adopt and utilize new technologies to enhance their work performance and how law firms achieve digital transformation to help lawyers better leverage AI tools and social media have become pressing practical issues. Combining AI tools and social media in research not only deepens the application of competency theory in the legal profession but also allows for a more comprehensive exploration of the mechanisms through which AI and social media influence the performance of lawyers.

However, no study has yet clearly revealed whether the use of AI tools and social media in law firms can enhance lawyer performance or the mechanisms through which related factors operate. Numerous studies have shown that various factors associated with AI tools and social media significantly impact lawyer performance. In terms of AI tools, Hamid [29] used structural equation modeling (SEM) to demonstrate that employees inclined to use digital technologies exhibit stronger adaptability, leading to improved performance; Adeyemi [7] employed quantitative research methods to prove that knowledge transfer using AI technologies significantly enhances law firm performance; and Alonderienė [30] found that self-directed learning using information technology, characterized by determination, initiative, confidence, and reflection, significantly impacts the individual performance of lawyers. Regarding social media, Cao [31] noted that the learning- and social-oriented use of social media effectively enhances performance, serving as an important supplementary resource for improving team and employee performance.

Although numerous studies have explored factors influencing lawyer performance, most are limited to analyzing single factors, lacking a macro-systematic investigation into how AI tools and social media jointly affect lawyer performance. Traditional SEM requires pre-specified models and hypothesis testing, making it inadequate for addressing the complex, inter-related indicators in AI and social media environments that collectively influence lawyer performance. To fill this research gap, this study employs a data-driven averaged Bayesian network model [32,33], which allowed for the construction of a causal network which captures the mechanisms through which AI tools and social media influence lawyer performance. Compared to earlier studies that analyzed only partial factors, this study’s main advantage lies in its ability to analyze joint probability distributions and derive causal models through a data-driven approach. Unlike traditional causal analysis methods such as SEM [34,35,36], this method does not require pre-constructed complex models, relies less on domain-specific expertise, and generates models with higher relevance and interpretability.

This study collected 322 questionnaires targeted at Chinese lawyers about their social media and AI use, of which 313 were deemed valid and used for analysis. All findings are based on statistical data obtained from these questionnaires. Compared to previous studies using interpretive structural modeling methods, this research offers the following innovations:

(a)

Breaking away from traditional single-factor research, this study is the first to explore the combined impact of AI and social media on lawyer performance, addressing the research gap relating to the integrated mechanisms of these technologies.

(b)

The use of a data-driven averaged Bayesian network model, which requires minimal expert knowledge, effectively dissects complex variable relationships, and establishes a causal framework for lawyer performance mechanisms, overcoming the limitations of traditional SEM when considering high-dimensional complex problems.

(c)

The findings reveal that AI-driven workload reduction, AI-supported leadership, and strain directly influence lawyer performance. Notably, excessive cognitive use of social media at work (ECU) exerts the most significant impact, while strain and work–technology conflict (WTC) serve as critical mediators in the relationship between ECU and performance. Practical recommendations are provided for lawyers to scientifically use AI tools and rationally utilize social media.

This remainder of this paper is divided into the following sections: In Section 2, a literature review focused on the impacts of AI tools and social media on lawyer job performance is given, and the causal analysis methodology is also overviewed. In Section 3, we introduce the ensemble Bayesian network and provide the candidate variables we used to study the mechanisms driving lawyer job performance. The lawyer job performance mechanism is then detailed, including the data collection and model building processes, in Section 4. Section 5 discusses the obtained lawyer job performance mechanism. Section 6 concludes by summarizing this study’s contributions and suggesting directions for future research, while Section 7 acknowledges this study’s limitations and provides recommendations for further exploration.

2. Literature Review

2.1. Progress in Research on the Impacts of AI Tools on Lawyer Job Performance

In recent years, the multi-dimensional impacts of AI technology on lawyer job performance have attracted extensive attention. In previous research, some researchers have proposed indicators reflecting the impact of AI technology on the job performance of employees. Rozman et al. [37] have put forward a multi-dimensional talent management model incorporating AI technology into the human resource management process, which specifically includes four key impact indicators: AI-supported employees’ training and development (AET), AI-driven workload reduction for employees (AWR), AI-supported organizational culture (AOC), and AI-supported leadership (AL). These indicators provide a comprehensive framework for understanding how AI technology can improve lawyer job performance.

First, in terms of AET, the maturity of AI technology has significantly enhanced lawyers’ professional capabilities. With the progress of legal reasoning models and data mining technologies, the accuracy and efficiency of AI systems in handling complex legal issues have been remarkably improved [37]; for example, AI technology can provide accurate legal consultations and solutions for lawyers by means of deep learning and analysis of legal provisions [20]. Moreover, the adaptability of lawyers and their learning attitude towards AI technology can directly affect their job performance. Davis [21] found that AI can improve lawyer job performance in areas such as litigation review, legal research, and predictive analysis. Additionally, lawyers who are proactive in learning and applying AI technology can master new tools more quickly and integrate them into their work processes, thus improving their work efficiency and quality. However, lawyers who are resistant to technology may not be able to fully utilize the advantages of AI technology, thereby affecting their job performance [38].

Second, in terms of AWR, AI technology has significantly improved lawyer job performance through the provision of automated processes. Wijayati et al. [39] have used AMOS22 SEM and found that AI has a significant impact on employees’ job performance and work contributions. For example, AI technology can quickly retrieve and organize a large amount of legal literature, providing accurate information support for lawyers and reducing the time and energy they spend on cumbersome information retrieval [20]. In addition, AI performs excellently in tasks such as legal document processing, case analysis, and contract drafting. It can automatically generate documents and extract key information, thus greatly enhancing work efficiency [40]. Surden [41] has pointed out that, although AI cannot completely replace lawyers’ high-order cognitive abilities, when considering tasks such as legal retrieval and statistical analysis, AI technology can significantly reduce lawyers’ work burdens and contribute to better job performance.

Third, in terms of AOC, AI technology not only provides valuable resources for employees but also promotes team collaboration and information sharing. The research of Zhang and Venkatesh [42] showed that effective AI technology can help employees to make timely use of organizational resources, thereby improving their job performance. For example, in terms of team collaboration, AI technology enhances communication and cooperation among lawyers through sharing information and resources, thus improving the efficiency and quality of case handling, which has beneficial effects on lawyer job performance. In addition, the application of AI technology has also promoted the transformation of organizational culture, encouraging employees to actively explore new technologies and achieve innovation in their work [37].

Finally, in terms of AL, the degree of attention and support of law firms for AI technology directly affects the effect of application of AI technology by lawyers, thus having a significant impact on lawyer job performance. Wang [43] and Asad [44] have pointed out that leaders play a crucial role in any successful enterprise, as they provide a vision that inspires and guides other employees to achieve goals. If the leadership of a law firm can actively promote the innovation of intelligent AI applications, it can significantly improve lawyer job performance; for example, through AI-based empowerment, the leadership can provide technical training, equipment investment, and resource support for young lawyers, helping them to better master and apply AI technology. This not only accelerates the professional growth of lawyers but also directly improves the service quality and client satisfaction of the law firm, which is reflected in better lawyer job performance. In addition, Narayanamurthy and Tortorella [45] have further emphasized that using information technology for knowledge transfer and application is an effective mechanism for improving lawyer job performance. Through implementing Industry 4.0 technologies (e.g., big data, the Internet of Things, and machine learning), law firms can optimize resource allocation and improve lawyers’ work efficiency and performance. The in-depth application of AI technology, such as automated document processing, intelligent legal retrieval, and predictive analysis, not only reduces the repetitive work to be carried out by lawyers but also improves the accuracy and efficiency of case handling, thus significantly enhancing lawyer job performance. In summary, AI has a profound impact on lawyer job performance through the four-dimensional impact indicators of employee training, workload reduction, organizational culture, and leadership.

2.2. Progress in Research on the Impacts of Social Media on Lawyer Job Performance

In the digital age, social media has been deeply integrated into lawyers’ professional lives, and its impacts on lawyer job performance have drawn significant attention from both academia and the industry. The prevalence of social media has presented new opportunities for lawyers’ professional development, yet it also brings about challenges. Existing research has predominantly focused on the positive changes that social media brings to lawyers, such as expanding information access, enhancing social interaction, and facilitating professional image building. Regarding the impact indicators of social media on lawyer job performance, Cao [46] has pointed out that excessive use of social media may exert a negative influence on lawyer job performance and categorized it into three key dimensions: excessive social use of social media at work (ESU), excessive hedonic use of social media at work (EHU), and ECU. Additionally, WTC and strain, as mediating variables, can further unveil the intricate impact mechanisms of social media on lawyer job performance.

First, we consider the impact of ESU on lawyer job performance. ESU refers to lawyers spending an excessive amount of time on non-work-related social activities on social media at work, such as frequently browsing feeds and engaging in irrelevant discussions. This behavior can cause lawyers to become distracted, thus reducing their work efficiency. Research has indicated that ESU significantly boosts the frequency of distractions and interruptions during lawyers’ work, making it difficult for them to concentrate on core tasks. For instance, although frequent interactions with clients or peers on social media can assist lawyers in building a relationship network, excessive socializing may lead to a loss of time management control, thereby affecting the efficiency and quality of case handling, ultimately influencing lawyer job performance [47]. Moreover, ESU may also result in information overload, compelling lawyers to consume a great deal of energy in screening and discerning information, further diminishing their job performance [48]. In research on the impact of social media on lawyers’ information acquisition, while lawyers can obtain information via social media, they might also become trapped in a quagmire of invalid information due to excessive socializing, leading to a decline in work efficiency and job performance [22,23]. Additionally, although the multi-functionality of social media platforms can enhance the efficiency of team communication, excessive socializing may divert lawyers’ attention from non-work-related conversations and activities, thus impacting their overall business performance and lawyer job performance [24].

Second, we consider the impacts of EHU on lawyer job performance. EHU refers to lawyers’ use of social media for entertainment or pastime purposes at work, rather than for work-related tasks. This behavior can distract lawyers during working hours and even cause them to veer away from work tasks. Research has shown that EHU significantly decreases lawyers’ work focus and task-completion efficiency [46]; for example, if a lawyer browses entertainment content on social media during their work breaks, it may lead to distraction and difficulty in promptly returning to the working state. Moreover, EHU may also trigger psychological strain, particularly when lawyers are confronted with urgent tasks. The conflict between entertainment consumption behavior and work responsibilities will exacerbate emotional exhaustion, thus negatively impacting lawyer job performance [47]. In research on the shaping of lawyers’ professional image in the social media era, while lawyers showcase their professional image through social media, they may also damage their professional reputation due to hedonic behavior; for example, excessive focus on entertainment content may lead to inappropriate words and deeds of lawyers on social media, thereby affecting clients’ trust in their professional capabilities and lawyer job performance.

Third, we consider the impacts of ECU on lawyer job performance. ECU refers to lawyers paying excessive attention to work-related information on social media at work, such as legal developments and industry trends. Although this behavior seemingly aids in enhancing their professional capabilities, over-attention may lead to information overload and psychological strain. Research has revealed that ECU causes lawyers to become interrupted frequently when handling daily tasks, making it challenging to maintain an efficient work rhythm [46]. For example, if a lawyer constantly refreshes legal information on social media or over-publishes work-related content, it may lead to distraction and affect their efficiency in terms of case analysis and decision making, thus influencing their job performance. Additionally, ECU may also trigger anxiety, especially when the volume of information is overwhelming and difficult to sift through. Lawyers may feel resource-depleted, thus affecting their job performance [47]. In research on the role of social media in lawyers’ social networks, while lawyers acquire industry trends through social media, they may also fall into the predicament of information overload due to excessive cognitive use, resulting in a decline in work efficiency and lawyer job performance [27]. Furthermore, the vast amount of information on social media platforms may make it difficult for lawyers to identify effective information, thereby influencing the quality of their decisions and job performance [28].

Fourth, we consider the mediating effect of WTC, which pertains to the mismatch or conflict between the use of social media and a lawyers’ current job responsibilities. This conflict significantly elevates lawyers’ strain levels, thereby having a negative impact on their job performance. Research has indicated that WTC is a crucial mediating variable between excessive social media use and the decline in lawyer job performance [47]; for example, when lawyers spend too much time on non-work-related activities on social media at work, they may feel pressed for time and struggle to complete assigned tasks, thus generating strain. This strain not only reduces work efficiency but may also lead to emotional exhaustion, further weakening their job performance [46]. While lawyers can discover cooperation opportunities through social media, they may also encounter difficulties in balancing the use of social media and work tasks due to WTC [26]; for example, if a lawyer frequently communicates with potential partners on social media, it may lead to improper time allocation, thus affecting their case handling efficiency and job performance [49].

In addition, we consider the mediating effect of strain: another vital mediating variable through which excessive social media use negatively impacts lawyer job performance. Research has shown that excessive social media use can significantly heighten lawyers’ psychological strain, especially in the face of information overload and loss of time management control [46]. For example, if a lawyer frequently receives and processes information on social media, it may lead to emotional fatigue and resource depletion, thus affecting their job performance. Moreover, strain may also trigger a series of negative emotions, such as anxiety and burnout, further reducing lawyers’ work efficiency and quality, consequently harming their job performance [47]. In research on the reputation management of lawyers in the social media era, the remarks and behaviors of lawyers on social media may spread widely. Once inappropriate remarks surface, they may severely damage the reputation and professional image of a lawyer, thus increasing psychological strain and negatively affecting their job performance [31]. Additionally, false propaganda and malicious competition on social media may also embroil lawyers in public opinion storms, divert their attention from handling cases, trigger trust crises among peers, and impede cooperation, all of which have negative impacts on lawyer job performance [31].

In conclusion, considering the impacts of social media on lawyer job performance, it can be seen as a double-edged sword. On one hand, it offers lawyers opportunities for information acquisition, social interaction, and professional image building. On the other hand, excessive use of social media (including ESU, EHU, and ECU) may, through the mediating effects of WTC and strain, have detrimental impacts on lawyer job performance.

2.3. Progress in Research on Causal Analysis Methodologies

In this study, we adopt a causal analysis method using a Bayesian network to investigate the effects of AI and social media on lawyer job performance. With the soaring prominence of AI and social media, data-driven methodologies have become an integral part of research in the social sciences. As an example, Kim has employed Bayesian network analysis to delve into the profound influence of R&D collaboration on Korea’s innovation performance [50]. Causal analysis, widely used in the natural [32,33] and social sciences [34], involves methods such as covariance matrices, regression coefficients, SEM [35,36], and multivariate linear regression [51]. Unlike fixed-structure SEMs, Bayesian networks allow for the direct study of the relationships between data. Traditionally, SEM was applied to explore the correlations between variables; however, this approach does not fully reveal their relationships. In this study, we utilized a Bayesian network to describe the correlations between AI, social media, and lawyer job performance. The network’s nodes represent specific factors, while the arcs between them indicate causal relationships. By learning from the data, a set of indicators and a causal network were identified. This study provides a succinct overview of existing research, which can help policymakers to better understand how to enhance lawyer job performance. We followed a research process that consisted of defining the question, selecting relevant studies, and analyzing data. Therefore, we select indicators based on relevant studies, thus extending the research context to the environment of law firms. This study enhances our theoretical understanding of the impacts of AI and social media on lawyer job performance and provides policymakers with insights into the use of AI and social media. In addition, we also provide empirical evidence regarding how lawyers can control when and how they use social media, in order to avoid the negative consequences of abuse.

3. Method

3.1. Data Collection

This research survey comprised two parts, which are provided in

Table A1. Questionnaire designed for this study.

Factors		Items
Personal information	Gender	Male\Female
	Age	My age
	Educational	1. Associate degree; 2. Bachelor’s degree; 3. Master’s degree; 4. Doctor’s degree
	Experience	Personal experience in the field.
	Frequency	Average daily time spent employing social media in my law firm.
		Frequency of daily social media use during working hours.
		I have come to use online platforms more and more often in my work to communicate with my clients, customers and/or team members.
Excessive use of social media	ESU	I always dedicate a significant time to employing social media as a means for establishing new professional relationships during working hours.
		During my employment, I devoted an unusually extensive time towards utilizing social media platforms to cultivate strong interpersonal bonds with my co-workers.
	EHU	During my downtime at the law firm, I frequently indulge in social media usage, which consumes a notably substantial portion of my time.
		For entertainment purposes, I find myself dedicating a significant time to employing social media platforms within my law firm.
	ECU	Within my law firm, I invest an unusually significant time in utilizing social media platforms to collaborate with my co-workers, generating valuable content together.
		Within my law firm, I allocate an unusually considerable time towards utilizing social media as a means to develop work-associated content.
AI-supported impact on personal development	AET	AI technology minimizes the duration required for in-firm training programs.
		AI technology has reduced the decline in attention span that attorney previously experienced during the course of traditional in-firm training programs.
	AWR	AI has enabled the prevalence of open communication in law firms, enabling timely on-the-spot resolution of attorney issues.
		The AI technology applied by our law firm can be used to communicate with users/clients, reducing the workload of lawyers.
AI-supported impact on the organization	AOC	AI makes the team culture of law firms very responsive and easy to change.
		AI allows lawyers to familiarize themselves with all the services they can provide to their clients in a law firm.
	AL	AI allows us to collaborate with data scientists, other lawyers and clients and can identify opportunities that may present themselves to our firm.
WTC		In my law firm, social media application has taken me further away from my work than I would like.
		In my law firm, social media use takes away from the time I feel I should be spending at work.
Strain		The tasks required by my law firm to be handled on social media leave me feeling overwhelmed as they drain my energy.
		In my law firm, working on social media all day is stressful for me.
Performance		I consistently fulfill all the duties and responsibilities outlined in my role.
		I consistently comply with all the formal performance criteria mandated in my company.
		I often fail to perform basic duties.
		AI technology improves the effectiveness of lawyers’ decisions and actions.
		AI technology provides accurate data and information to attorneys and clients.
		The use of AI enables us to enhance our firm’s performance faster than our main competitors.

Note: This study employed a Likert scale based on five points to evaluate responses, covering “completely disagree” to “completely agree”, with each question evaluating the corresponding five dimensions.

. The initial portion of the questionnaire collects essential personal details from the participants, including their gender and age. It also inquiries about the average time respondents spend using social media and AI platforms in their daily work, and the associated results are presented in

Table 1. Demographic data and the frequency of social media/AI usage among respondents.

Items	Category	Frequency	Percentage (%)
Gender	Male	159	50.8
	Female	154	49.2
Age Groups	20–30	148	47.3
	31–40	98	31.3
	41–50	28	8.9
	51–60	37	11.8
	60 years and above	2	0.6
Education	Associate degree	15	4.8
	Bachelor’s degree	194	62
	Master’s degree	96	30.7
	Doctoral degree	8	2.6
Experience	<1 year	31	9.9
	<5 years	139	44.4
	≥5 years	143	45.7
Average Daily Time Spent on Social Media at Work	Seldom	13	4.2
	<1 h	19	6.1
	1–2 h	56	17.9
	2–3 h	57	18.2
	≥3 h	168	53.7
Frequency of Daily Social Media Usage at Work	Seldom	20	6.4
	1–3 times daily	31	9.9
	4–6 times daily	42	13.4
	7–9 times daily	22	7
	≥10 times daily	198	63.3
I have increasingly used online platforms for communication with my colleagues, clients, and/or team members during work	Completely disagree	14	4.5
	Disagree	13	4.2
	Inconclusive	42	13.4
	Agree	144	46
	Totally agree	100	31.9
Total		313	100

. The second part of the questionnaire is the main section, focusing on lawyer job performance and the impacts of using AI and social media. The questions in the questionnaire were derived from existing scales [37,46,52]. This survey was conducted through a combination of online and offline methods. The target participants were Chinese lawyers. The National Unified Legal Professional Qualification Examination in China is highly demanding, which leads to a relatively small number of lawyers. Additionally, when lawyers engage with AI and social media, they must comply with the specific requirements of nearly 17 laws and relevant statutes. Consequently, the process of gathering questionnaires from this group is fraught with difficulties. Due to the special nature of the lawyer population, we primarily invited lawyers working in law firms to participate and spread the survey through WeChat groups to include more participants from various age groups and educational backgrounds. The online method mainly involved posting the questionnaire on the QuestionStar website and inviting participants to fill it out. Additionally, we also conducted offline distribution of survey questionnaires. Of the total 322 questionnaires received for this survey, 313 were deemed valid and suitable for analysis. This implies a valid response rate of 97%, which meets the requirements for statistical analysis. The criteria used to determine the validity of the questionnaires were as follows: (1) whether respondents participated in the survey and answered all the questions and (2) whether the majority of the answers were consistent or showed obvious patterns, indicating invalid responses. Ultimately, the 313 valid questionnaire responses were included for analysis. The statistical characteristics of the sample structure are as follows:

The statistical findings reveal a balanced representation of both male (50.8%) and female (49.2%) lawyers in the survey. A significant portion of the respondents fell into the age range of 20–30 (47.3%), with the majority holding a bachelor’s degree or higher (85.3%). Moreover, the majority of the lawyers had more than five years of professional experience (45.7%). In terms of job experience, most lawyers reported spending more than three hours each day utilizing social media for work-related activities on average (53.7%). Additionally, a significant 63.3% of lawyers reported engaging with social media over ten times a day, with a substantial segment acknowledging the increasing ubiquity of AI and online platforms within the professional sphere. An analysis of the questionnaire’s reliability and validity was conducted, and the questionnaire’s internal reliability was measured by adopting Cronbach’s alpha. Cronbach’s alpha coefficients for each of the latent indicators in this study exceeded the basic criterion of 0.7, with most of them surpassing 0.8, indicating the well-established latent indicators of the respective question items. Therefore, the questionnaire demonstrated high reliability.

The test results shown in

Table 2. KMO and Bartlett’s test results for the questionnaire.

Kaiser–Meyer–Olkin Measure of Sampling Adequacy		0.909
Bartlett’s Test of Sphericity	Approx. Chi-Square	3993.115
	df	253
	Sig.	0.000

indicate that the KMO (Kaiser–Meyer–Olkin) test value for the survey data was 0.909, which exceeds the recommended minimum value of 0.70. This suggests that the questionnaire is applicable for factor analysis. Moreover, Bartlett’s test of sphericity yielded a relatively large approximate chi-square value of 3993.115 with a significance probability of 0.000 (p < 0.01). Therefore, the null hypothesis of Bartlett’s test of sphericity could be rejected, indicating that this scale is applicable to and effective for performing factor analysis and possesses a strong validity structure.

3.2. Ensemble Bayesian Network

In this study, we utilized an averaged ensemble Bayesian network to describe the correlations among AI, social media, and lawyer job performance. In a Bayesian network, the nodes of the network represent specific factors, while the arcs between them indicate causal relationships. In some practical problems, only a limited sample is available. Usually, we directly estimate a single model from the data and then draw our conclusions. However, this approach may overlook the fact that the estimated model is not “fixed”, and can lead to the interference of noisy samples on the overall structure [53].

We employed model averaging to enhance the reliability of structure learning [54,55]. According to the central limit theorem, the population distribution can be estimated by estimating sub-samples. By resampling the data using the bootstrap method, 100 independent and identically distributed sub-datasets were obtained. We then used each sub-dataset to construct a Bayesian causal network separately and superimposed these sub-networks to calculate the average strength of the arcs and retain only the causal structures with a high occurrence frequency. The Bayesian network constructed in this study mainly focuses on analysis of the correlations between the impacts of AI tools and social media on lawyer job performance. During the analysis, each node represents a specific factor, and the arcs connecting the nodes indicate causal relationships. Through Bayesian network analysis, we aim to reveal the causal network among the selected variables, providing strong support for a deeper understanding of the impacts of AI and social media on lawyer job performance.

3.3. Main Variables in the Bayesian Network

Based on the literature review, we considered several key variables that may potentially influence lawyer performance. This study adopts the classification method proposed by Cao [46], categorizing the excessive use of social media at work into three impact indicators: ESU, EHU, and ECU. Additionally, conflicts and strain arising from work and technology are also important factors to consider. In terms of AI technology, as mentioned earlier, this study refers to Rozman’s [37] classification of AI. When examining the impacts of AI on lawyer work performance, AI is divided into four impact indicators: AET, AWR, AOC, and AL. All nine indicators mentioned above are supported by relevant studies.

3.3.1. ESU

ESU in the workplace refers to the excessive investment of time and energy in establishing and maintaining social relationships within the work environment. Socializing on social media is crucial for lawyers to establish and maintain work connections and personal relationships; however, ESU can lead to exhaustion, strain, and lower work lawyer job performance. Therefore, if lawyers excessively engage in socializing on social media, it may create a lack of balance between utilizing social media as a tool to expand their client base and network and the potential negative impacts on their performance [47].

3.3.2. EHU

Hedonic use of social media refers to individuals using these platforms primarily for entertainment and pleasure [52]. Individuals who engage in hedonic use frequently allocate a substantial amount of time to social media usage while at work, as it enables them to maintain a high level of enjoyment [46]. This excessive usage can lead to distractions and difficulties in focusing on work tasks. Consequently, lawyers may experience strain as they struggle to strike a balance between indulging in the hedonic use of social media and fulfilling their work responsibilities.

3.3.3. ECU

This indicator primarily emphasizes the importance of using social media for work-related purposes, such as promoting work content or seeking job-related information. When the level of cognitive use of social media for work exceeds the optimal threshold, it can lead to increased strain for employees [46]. Lawyers commonly engage in frequent utilization of social media platforms to actively pursue work-related information and expand their client base; however, excessive posting of work-related content on social media may also potentially cause stress for lawyers.

3.3.4. AET

The Spartak [56] study demonstrated the positive outcomes of utilizing AI tools for employee education. The findings revealed a significant 30% increase in productivity, an 18% improvement in employee engagement, and a notable 65% reduction in learning time. Additionally, AI technology plays a valuable role in identifying the training needs of and advancement opportunities for employees [57]. In general, the introduction of efficient training and development programs supported by AI tools has been proven to yield favorable results in terms of both company performance and employee engagement [39].

3.3.5. AWR

AI can save nearly one-third of the time spent on simple and repetitive tasks, resulting in improved job performance and increased employee engagement [39]. AI can also assist employees with time-consuming repetitive tasks, thereby reducing their workload and enhancing productivity [43]. Moreover, AI can significantly reduce the work stress and overall workload for employees [58].

3.3.6. AOC

A company that successfully cultivates a positive AI culture and nurtures an inclusive and motivating environment can effectively manage, transform, and engage all its teams of employees through AI [59].

3.3.7. AL

The development and application of AI have greatly transformed the role of leaders [39]. Therefore, lawyers need to enhance their leadership through AI to stay competitive and provide guidance for their teams. Thus, we hypothesize that AL may have an impact on lawyer job performance in law firms.

3.3.8. WTC

Technostress originates from the stimuli that individuals encounter and can trigger diverse psychological responses [48]. The WTC caused by excessive use of social media hinders individuals from fulfilling their job responsibilities, resulting in poor job performance [47]. Lawyers have limited time and energy, and the irrational use of social media is likely to conflict with their job responsibilities, leading to increased strain and affecting their job performance. Thus, it can be inferred that there is a negative correlation between work–technology conflict and lawyer job performance.

3.3.9. Strain

When employees face overwhelming work demands due to excessive use of social media, they will feel resource-depleted and fall into psychological stress [60]. Past research has shown that psychological strain has an adverse effect on job performance. Therefore, it is hypothesized that excessive use of social media causes work demands to exceed lawyers’ coping abilities, triggers psychological stress, and affects job performance.

3.3.10. Lawyer Job Performance (Performance)

Adeyemi et al. have proposed that performance evaluations can adopt both financial and non-financial indicators [7]. Given the sensitivity of the financial data of law firms, this study evaluates lawyer job performance from the perspective of non-financial criteria of financial performance through the perceptions of the participants. When constructing the measurement criteria for lawyer job performance, we adopted the method proposed by Wijayati et al. [39]. This method includes various job performance indicators, such as continuously fulfilling the responsibilities listed in the job description, continuously meeting all formal performance requirements, and assessing the frequency of failure to fulfill basic responsibilities.

4. Experiments and Results

4.1. Model Building

In the first step, we calculated the correlations among all the variables used, including 13 variables such as gender, age, usage frequency, years of practice, ESU, EHU, WTC, AOC, AET, AWR, AL, and strain. Before constructing the Bayesian network, we first needed to eliminate some variables with low correlations from the indicators in the questionnaire. This process was based on the traditional assumption that if there is a causal relationship between two variables, there should be a high correlation between them, where this correlation can be either linear or non-linear. Correlation analysis is a valuable data analysis method. Through analyzing the connections between different features, a potential relationship can be found. The correlation coefficient is a numerical measure of some correlations, indicating the statistical relationships between different variables. These variables can be two components of a random multi-variable with a known distribution or two columns of an observation matrix. There are several types of correlation coefficients, such as the Pearson correlation coefficient [61], the Spearman correlation coefficient [62,63], the Kendall correlation coefficient [64], and the maximal information coefficient [65,66]. Each method has its own definition and scope of application, and the coefficients usually range from -1 to 1. The more significant the absolute value of the coefficient, the greater the correlation. According to our previous work [67], we adopted the mixed correlation coefficient (MCC) to quantify the correlation strength among the indicators, and the results are shown in Figure 1.

After conducting the MCC analysis, we identified and eliminated three indicators—namely, “gender”, “experience”, and “frequency”—which had low correlations with other indicators. Subsequently, we established a Bayesian causal inference model for ten indicators (including “ESU”, “EHU”, “ECU”, “AET”, “AWR”, “AOC”, “AL”, “WTC”, “Strain”, and “performance”) and used the data-driven method to analyze the causal relationships among these indicators.

We adopted the average Bayesian method, aiming to mitigate the impacts of singular values and randomness in the questionnaire data on the results of causal inference. Specifically, we first collected the questionnaire data and screened out 313 valid responses. Then, 100 independent and identically distributed sub-questionnaire datasets were generated using the sampling-with-replacement technique. For each sub-dataset, the Bayesian network analysis method was adopted, combined with the Max–Min Hill-Climbing algorithm [68], with the goal of minimizing the Bayesian Information Criterion to construct a network structure that describes the causal relationships of lawyer performance. Eventually, 100 sub-Bayesian networks were generated.

In the subsequent integration step, we superimposed the causal graphs of these sub-networks and calculated the average strength and direction of each arc. In the case where the frequencies from ESU to EHU and from EHU to ESU were both 50%, we considered these two indicators to be mutually causal and, thus, retained the two-way arc. In addition, the directed arcs with an occurrence frequency of less than 45% were eliminated, and the direction with higher frequency was selected for each arc. In this way, we obtained the average Bayesian causal graph network.

The final constructed causal graph is shown in Figure 2, revealing the impact mechanism of AI technology and social media on lawyer performance. In this study, the independent variables included ECU, ESU, EHU, WTC, AOC, AET, AWR, AL, and strain, while the dependent variable was lawyer performance. The average Bayesian causal graph network removed the causal relationships that were regarded as noise or false positives in the samples, due to their low occurrence frequencies. For example, the causal association between EHU and AL was only found in 7 out of 100 sub-networks and, thus, was eliminated. In addition, the advantage of the data-driven method lies in its ability to automatically discover mediating effects. For example, ESU can indirectly affect performance through WTC. After considering this mediating effect, as the direct causal relationship between ESU and performance is not significant, the corresponding arc was replaced by ESU → WTC → performance. This adjustment not only simplifies the structure of the causal model but also enhances its ability to explain the real world, avoiding the difficulty of interpretation caused by overly complex causal relationships.

4.2. Key Findings of Ensemble Bayesian Network

Through the Bayesian causal inference network capturing the impact mechanisms relating to lawyer job performance, law firm managers can obtain some ideas for improving lawyer job performance. The first finding of this study is that strain and WTC have important mediating effects between ECU and lawyer job performance. In the graph, the intensity of the influence among various influencing factors can be observed. The intensity of the direct effect of ECU on performance was 0.82; however, when ECU was mediated through strain or WTC, the impact intensity on performance rose to 1. This indicates that, compared with the direct influence, the indirect paths through strain and WTC have a more significant effect on performance. In reality, when lawyers over-cognitively use social media at work, such as over-publishing work-related information and over-focusing on legal information and industry trends, it will first directly distract their attention from core work, such as case handling, and then have a direct negative impact on performance, with an intensity of 0.82. More crucially, this excessive cognitive use will trigger strain. When facing the published work information, lawyers need to spend a lot of energy to prepare in order to maintain their professional image as lawyers, and they may also face negative online public opinions, which increases their strain. The intensity of the relationship between strain and performance was 1, thus having a more intense negative impact on performance. At the same time, over-cognitive use of social media may also lead to WTC; for example, excessive use of social media takes up too much time, and publishing too much work-related content on social media conflicts with lawyers’ current job responsibilities. This conflict further exacerbates strain, having a negative impact on performance with an intensity of 1. Lawyers’ over-cognitive use of social media means that they frequently disperse their attention between various social information during working hours, which is highly likely to trigger psychological anxiety and unease.

The second finding of this study is that AWR, AL, and WTC can directly affect lawyer job performance, for the following reasons.

First, AWR can directly affect lawyer job performance. Traditional lawyer business processes are often filled with a large number of repetitive and time-consuming basic tasks; for example, when handling complex commercial litigation cases, lawyers need to manually retrieve relevant provisions and cases from mountains of regulatory documents and past precedents, which not only can take days or even weeks but also is very likely to lead to the omission of key information due to human negligence. After introducing intelligent legal software and using AI tools, with their powerful natural language processing capabilities and the advantage of massive data storage, they can accurately locate the required materials and quickly sort out the context of the case, greatly shortening the preparation time. In this way, lawyers can redirect their saved time and energy to high-value-added tasks such as analysis of the core controversial points of the case and the formulation of strategies, directly promoting an exponential increase in their case-handling efficiency and, thus, significantly and positively affecting their job performance. In addition, AI optimizes the work process, reduces repetitive work, and thus reduces the workload and strain.

Second, regarding AL, this does not simply mean that AI replaces humans as the leaders of law firms but, rather, that AI technology empowers the management decision-making processes of law firms. In terms of team-building, AI tools can analyze multi-dimensional data such as the types and difficulties of past cases, as well as lawyers’ personal professional expertise and case-winning rates, in order to accurately match the most suitable team member combinations for law firms, ensuring that each case can be undertaken by the most advantageous team. In terms of business allocation, AI can scientifically and reasonably allocate tasks based on real-time case flows, lawyers’ current workloads, and the fit of professional fields, avoiding uneven workloads or professional mismatches among lawyers. When a law firm undertakes large-scale and complex projects, such as cross-border mergers and the acquisition of legal services, AI can comprehensively consider various factors to plan a detailed task roadmap and guide team members to collaborate in an orderly manner. Through optimizing all aspects of the operation of the law firm, it indirectly creates a good working environment and an efficient cooperation atmosphere for lawyers, strongly boosting improvements in lawyers’ personal performance.

Finally, strain is a direct influencing factor affecting lawyer job performance. The lawyer profession itself is characterized by high intensity, high complexity, and high uncertainty, and the fast-paced nature of modern society and the popularization of information technology and digital tools have further exacerbated the strain levels of lawyers. The impact of this strain on lawyer job performance is mainly reflected in the following aspects: First, there is an imbalance in time management and energy allocation. High-intensity strain may cause lawyers to be unable to concentrate; for example, they may be slow to react or have unclear thinking before a court session due to excessive strain. Second, there is information overload and distraction. Strain exacerbates the burden of information processing, leading to frequent errors when writing legal opinions or analyzing cases, thus affecting the progress and quality of the work performed. Finally, the ability to apply technology declines. In a high-pressure environment, lawyers may find it difficult to focus on mastering new legal technology tools; for example, in complex litigation, they may not be able to proficiently use electronic evidence analysis software, which may lead to the omission or misjudgment of key evidence, directly affecting the outcome of the case.

The third finding of this study is that ECU has the most significant impact on lawyer job performance in the Bayesian network. As presented in

Table 3. Measures of centrality for indicators in the obtained network.

Indicators	In-Degree	Out-Degree	Total
Excessive cognitive use of social media at work (ECU)	0	6	6
Excessive social use of social media at work (ESU)	2	3	5
Excessive hedonic use of social media at work (EHU)	2	2	4
Work–technology conflict (WTC)	3	2	5
Strain	4	1	5
AI-supported employee training and development (AET)	1	2	3
AI-driven workload reduction for employees (AWR)	1	1	2
AI-supported leadership (AL)	1	1	2
AI-supported organizational culture (AOC)	1	2	3
Lawyer job performance	7	0	7

, the centrality measures for the indicators in the network clearly illustrate the significant influence of ECU. ECU has an in-degree of 0 and an out-degree of 6, making it the independent variable with the most significant impact on lawyer job performance. Excessive cognitive use of social media refers to cognitive-level behaviors of lawyers in the workplace, such as being overly immersed in the in-depth interpretation of social media information and over-analyzing the impacts of social feedback on their own professional image. This may cause lawyers to spend a great deal of energy on the psychological construction of virtual social interactions, deviating from the core handling of cases; for example, when lawyers are at a crucial stage of preparing for a court trial, instead of focusing on the case, they may repeatedly think about the social impact of their brand on social media or overly worry that not getting enough likes for their published professional insights will damage their reputation. As a result, they fall into anxiety and self-doubt, thus diverting their energy away from delving into case facts and sorting out legal logic, ultimately having a serious negative impact on their case-handling efficiency, quality, and overall performance. Therefore, it is recommended that lawyers strive to control the frequency of cooperation with colleagues on social media platforms and limit the posting of work-related content on their personal social media accounts to relieve work strain and improve work performance. Constantly checking, participating in discussions, or posting content during working hours will divert attention from key tasks, reduce work efficiency, and may lead to disputes and risks. Negative feedback on social media requires lawyers to exert more effort in managing their professional image and reputation, which exerts pressure on their work performance. Social media is filled with a complex array of information, including the dynamics of peer competition and public sentiment. When lawyers continuously pay attention to this information, they will involuntarily engage in self-comparison, worrying that their business capabilities and customer resources are inferior to those of others, thus generating huge psychological stress. As this strain accumulates, when they deal with actual cases, it becomes difficult for them to concentrate on deeply analyzing legal provisions and constructing a rigorous litigation strategy. This greatly reduces their work efficiency and can ultimately seriously affect their performance.

5. Discussion

In the contemporary legal landscape, AI tools and social media have emerged as transformative forces, permeating every aspect of legal practice. According to our study, the first finding of this study is that strain and WTC have important mediating effects between ECU and lawyer job performance. The second finding of this study is that AWR, AL, and WTC can directly affect lawyer job performance. The third finding of this study is that ECU has the most significant impact on lawyer job performance in the Bayesian network. On one hand, although wise use of social media can expand case resources, build a lawyer’s personal brand image, and reduce the communication cost with clients, the over-cognitive use of social media may damage professional performance, thus reducing job performance [69,70,71]. Excessive use of social media can cause lawyers to become distracted and reduce their work efficiency [72,73]. This result is consistent with the research conclusions of Cao [46] and Maier [47].

Our finding that AWR directly affects job performance is consistent with the conclusion of Savelka [13] and others, stating that AI is of great help in case handling and information retrieval. Through automating routine activities such as document organization and data archiving, AI enables lawyers to focus on core tasks, thereby improving their work efficiency, job satisfaction, and sense of achievement. In addition, the use of AI can also enable lawyers to improve their work efficiency, exert leadership skills, and cultivate a positive organizational culture, consistent with the conclusions of previous studies [22,74,75].

This study is dedicated to furnishing lawyers with actionable insights to harness these two elements in manner that enhances their lawyer job performance. Extensive research has underscored the pivotal role of optimizing AI and social media usage in improving lawyer job performance. Drawing on three key findings, the following recommendations are put forward across three distinct levels, including individual lawyers, law firms and lawyers’ associations, and the legislative and policy-making domains.

5.1. Suggestions for Lawyers to Enhance Their Job Performance

In the legal sphere, AI tools—especially LLMs—have remarkable potential. They are capable of automatically generating specific legal documents, a technological advancement that holds the promise of streamlining legal processes, reducing the cost of legal services, and bolstering access to justice. A research endeavor has delved into whether legal practitioners—including lawyers—perceive legal documents differently based on their presumed origin: human-written or AI-generated. Participants evaluated the documents primarily in terms of accuracy and language quality. The results demonstrated that they had a marked preference for human-written documents over those thought to be AI-generated. Simultaneously, the majority of participants expressed anticipation for the future automatic generation of documents [76]. Consequently, legal practitioners must be aware of this cognitive divergence. When using AI-generated documents, they should integrate their professional acumen with the output, conducting meticulous reviews to guarantee their accuracy and legality. Paramount among these is the lawyer’s capacity to critically assess AI-generated content. The American Bar Association has already flagged the importance of lawyers reviewing the content generated by generative AI. To boost lawyer job performance, legal professionals should embrace AI technologies wholeheartedly, enhance their technical proficiency, and actively partake in AI-related training courses and seminars in the legal domain. They should delve deep into the technical underpinnings and operational methodologies of AI tools; for instance, through specialized online courses or in-person workshops, they can master the art of using AI tools to initiate the creation of legal documents and construct frameworks for contract and legal instrument drafting. Additionally, staying attuned to the latest developments in legal technology and continuously updating their knowledge reservoir are essential for leveraging these technologies to enhance work efficiency. Lawyers must also fortify their content-review consciousness, and AI-generated legal documents should never be used without due diligence. Instead, they should be subjected to comprehensive scrutiny, examining every aspect from the precision of legal citation, to the cogency of logical reasoning, and to the propriety of language expression. For example, when reviewing contract terms, lawyers must ensure they align with the clients’ intentions and legal requirements, thereby averting potential legal pitfalls. Establishing a symbiotic human–machine collaboration model can significantly enhance lawyer job performance. Lawyers can proactively explore AI tools tailored to their practice areas, experimenting with general-purpose AI software such as ChatGPT4.0 and Deepseek-V3 for tasks including legal provision retrieval and preliminary case-analysis screening. This can expedite information acquisition and inspire novel case-handling strategies. Regularly reflecting on the utility of AI tools, through comparing metrics such as the time and accuracy of handling similar cases before and after AI adoption, can help lawyers to fine-tune their usage, achieving the dual goals of reducing workload and enhancing efficiency, while alleviating work-related stress and technological frictions.

The integration of social media into the legal profession has ushered in both opportunities and risks. In the context of China’s legal career development, social media-induced professional ethical risks mainly encompass violations of confidentiality obligations, non-compliance with lawyer advertising regulations, and inappropriate interactions with judicial personnel. Article 38 of the Lawyers Law of the People’s Republic of China [77] and Article 9 of the Rules of Professional Conduct for Lawyers [78] serve as the legal and ethical bedrock for lawyers’ confidentiality obligations; however, these provisions harbor certain ambiguities, such as the scope of “the lawyer’s side”, the temporal scope of “information known within a specific period”, the definition of “relevant information”, and the confidentiality recipients. These uncertainties become more pronounced when social media infiltrates legal practice, exposing lawyers to the risk of breaching professional ethics. Some lawyers—due to a lack of clear demarcation between personal opinion expression and confidentiality obligations on social media—may unknowingly violate ethical norms, remaining oblivious even when subject to disciplinary actions by the lawyers’ association [79]. Furthermore, the Rules of Professional Conduct for Lawyers and the Rules for the Disciplinary Sanctions of Members of Lawyers’ Associations offer scant guidance on potential ethical issues arising from social media use by lawyers and law firms. This can lead to anxiety and stress among lawyers when using social media, for fear of unknowingly breaching confidentiality ethics. When using social media, lawyers should safeguard the privacy of case parties, scrupulously adhering to the confidentiality provisions in the Rules of Professional Conduct for Lawyers. They should select communication methods that prioritize the protection of clients’ personal information. For example, lawyers should gain a thorough understanding of the functions and information-sharing mechanisms of social media platforms and utilize privacy settings to curtail information leakage. When relying on AI to generate legal documents, strict compliance with legal standards is imperative. For instance, after using AI-assisted tools for contract drafting, lawyers must painstakingly review the terms to ensure compliance with relevant laws, such as the Civil Code. Lawyers should also demarcate clear boundaries between their personal and professional lives. Developing a personalized social media usage plan can help them to distinguish between work-related and personal use, preventing excessive social media engagement from causing anxiety and distraction. Limiting the dissemination of work-related content, minimizing work-hour collaborations with colleagues on social media, and restricting access to entertainment-oriented social media sections can enhance focus. Setting limits on the frequency of checking and replying to work-related group chats can curtail unproductive discussions. Cultivating healthy social media usage habits, such as following professional legal accounts to augment professional knowledge and refraining from posting unvetted work-related updates, can mitigate professional risks and information leakage. Simultaneously, fostering a scientific understanding of stress, recognizing its objectivity and the positive role of moderate stress while remaining vigilant against excessive stress, is crucial. Lawyers can manage stress through prioritizing cases, seeking team collaboration, or accessing the psychological counseling resources provided by their law firms, thus maintaining optimal job performance.

5.2. Suggestions for Law Firms and Lawyers’ Associations to Optimize Lawyer Job Performance

Research has revealed that AI-driven leadership, along with the mitigation of workload and WTCs, can directly impact lawyer job performance. Lawyers’ associations—in light of their mandate and the current dearth of in-depth research on the awareness and application of AI by legal practitioners—should assume a leadership role in this regard. They should organize comprehensive AI application training programs for all lawyers, inviting luminaries from the legal technology domain, including experts, scholars, and seasoned lawyers, to conduct lectures. These training courses should cover a gamut of topics, from fundamental AI concepts, legal application scenarios, and practical operation skills, to risk-mitigation strategies. This initiative can galvanize the legal community to enhance their AI-related knowledge and skills, propelling the industry forward. Lawyers’ associations should also formulate industry-specific norms to guide the healthy development of AI use in the legal profession. Anticipating the potential issues that lawyers may encounter when using AI, these norms should clearly define requirements for generating legal documents with AI, such as source attribution, review responsibilities, and information-security safeguards. Concomitantly, a set of disciplinary measures for non-compliance should be established to safeguard the industry’s integrity and reputation. To foster knowledge-sharing and innovation, lawyers’ associations should create diverse communication platforms; these can include themed forums, case-sharing sessions, and online communities, where lawyers can exchange experiences, discuss challenges, and proffer solutions related to the use of AI. From a law firm management perspective, internal work processes should be optimized through the integration of AI-assisted elements; for example, intelligent comparison software can be deployed for contract review, and intelligent classification tools can be utilized for case archival. Establishing dedicated technical support positions within law firms can help to bridge the gap between the latest AI technologies and the practical needs of lawyers, providing personalized tool-configuration advice and usage consultations. In line with China’s push for the digital transformation of the legal industry, law firms can participate in government-sponsored training initiatives, such as local government-organized training camps that feature the latest AI-based legal technology applications, which can enhance the digital capabilities of the local legal sector.

Law firms must also prioritize ethics and risk-awareness training. Given China’s stringent data protection laws, such as the Data Security Law [80] and the Personal Information Protection Law [81], ethics training should center on regulatory compliance. Legal experts should be invited to conduct training sessions, which elucidate how to handle client data securely when using AI systems. These sessions should cover the significance of algorithm transparency and analyze real-world AI-related legal risk cases, such as data leakage incidents. In accordance with the Cybersecurity Law [82], employees should be made aware of their legal obligations regarding data protection, including security incident reporting requirements. Law firms should also formulate robust data management and security policies. Client data should be categorized according to their sensitivity level, with the strictest security measures reserved for highly sensitive information, such as personal financial data and business secrets. Complying with the Regulations on the Security Management of Network Data [83], law firms should establish a clear data governance framework, defining the data lifecycle to ensure that data use aligns with client consent and legal requirements. Adhering to international standards, such as the General Data Protection Regulation of the European Union [84] and staying abreast of domestic regulatory changes, law firms can either establish in-house regulatory compliance teams or appoint dedicated compliance officers to ensure the legality of operations [85]. These concerted efforts can help to familiarize lawyers with AI, reduce work-related stress and technological frictions, and ultimately enhance lawyer job performance.

The International Bar Association posits that, when formulating guidelines for social media use in the legal profession, the overarching goal should be to facilitate and encourage the legal and effective utilization of social media in the legal realm, while adhering to six fundamental principles [78]: maintaining independence, fostering integrity, clarifying responsibilities, emphasizing confidentiality, preserving trust, and refining policies. These principles serve as guiding beacons for lawyers and law firms in their social media endeavors. The ‘All China Lawyers Association’ can draw inspiration from the New York State Bar Association. Combined with the actual situation of the legal profession in China, it can modify the Rules of Professional Conduct for Lawyers [78] or specifically formulate a Code of Conduct for Lawyers’ Use of Social Media. Lawyers’ associations should formulate norms for social media use in law firms, clarify the external publicity tone and communication taboos with clients, and guide lawyers to correctly use social media to expand their business and maintain their image. Creating an intelligent and friendly team culture, organizing internal training and sharing sessions, and encouraging lawyers to share the usage skills of AI tools and successful cases regarding the use of social media to legally and reasonably build brand images and obtain case sources can form an atmosphere of mutual help and learning. This can help to eliminate the fear and resistance of some lawyers to new technologies, further contributing to improvements in lawyer job performance.

5.3. Legislative and Policy Recommendations to Enhance Lawyer Job Performance

Legislators and policymakers should actively promote legislation which prompts government departments to pay attention to issues such as the protection of rights and interests and data privacy in the context of the digital transformation of the legal profession, ensuring that the legitimate rights and interests of lawyers are not infringed when using AI tools and social media platforms. For example, the application of the Data Security Law [80] in the scenarios of data storage and transmission by lawyers can be refined. Building an AI innovation platform for the legal industry which brings together the forces of law firms, technology enterprises, and scientific research institutions may allow for the joint development of intelligent solutions for the entire legal business process. This may include creating an integrated intelligent service ecosystem from client management and case handling to knowledge precipitation, making up for the current deficiencies of scattered and unprofessional tools. Policymakers should formulate supportive policies to promote the in-depth application of such technology. Setting up special scientific research funds to support the joint research of universities, scientific research institutions, and legal technology enterprises can help to overcome the key technical problems of AI application in the legal field. China should revise its laws to clarify the rules relating to the application of technology application. With the extensive application of AI in the legal field, the existing laws and regulations may be difficult to adapt to the new technological environment. Legislators should pay attention to the development trends of AI technologies, promptly revise and improve relevant laws and regulations, and clarify key issues such as the legal status, legal effect, and liability attribution of AI-generated content. For example, formulating legal provisions to specify the circumstances in which AI-generated legal documents have legal effect—as well as the responsibilities of developers, users, and relevant institutions when errors or disputes occur—can provide a clear legal basis for lawyers to use AI technologies and ensure the orderly development of the legal industry in the new technological environment. China should also carry out cross-domain collaborations to ensure the compliant development of technology. As the application of AI technologies involves multiple fields such as law, technology, and ethics, policymakers and legislators should strengthen the cooperation and coordination among different fields, for example, through organizing research teams composed of legal experts, technical experts, ethicists, and so on to conduct in-depth research on the complex issues of AI application in the legal field, allowing for the formulation of comprehensive solutions ensuring that the application of AI technologies in the legal industry complies with the laws of technological development, as well as legal and ethical norms. For example, when formulating AI-related policies and laws, fully considering the technical feasibility, legal normativity, and ethical rationality can avoid the abuse of technology or legal loopholes, thus ensuring the healthy and sustainable development of AI technologies in the legal field, consequently promoting improvements in lawyer job performance.

At present, there is no special code of conduct for lawyers’ use of AI tools or social media in China. The Commercial and Federal Litigation Department of the New York State Bar Association formulated the Code of Ethics for Social Media on 18 March 2014, and revised it on 9 June 2015 [86]. This code provides lawyers with new perspectives on the application of traditional professional ethical rules in social media, including seven parts focused on lawyers’ skills, lawyer advertising, providing legal advice through social media, the review and application of social media evidence, contacting clients, investigating jurors and reporting jurors’ improper behavior, and communicating with judicial officers through social media [79]. Chinese Lawyers’ Associations should draw on the experience of the New York State Bar Association, combined with the actual situation of the legal profession in China, and jointly put forward suggestions to the legislative authorities with industry experts and universities. In turn, the legislative authorities should modify the Rules of Professional Conduct for Lawyers or specifically formulate the Code of Conduct for Lawyers’ Use of Social Media and the Guidelines for the Application of AI in the Legal Profession. When modifying or formulating the rules, aiming to “promote and encourage the legal and effective use of AI technologies and social media in the legal profession” and clarifying the professional ethics and operation boundaries of lawyers when using new technologies and new media are essential. Legislators should promote the deep integration of social media and the legal industry. Negotiating with mainstream social media platforms to launch exclusive service sections for lawyers, providing functions such as client legal consultation matching and popular science-style publicity of professional content, as well as using the algorithm supervision function of the platform to assist lawyers in controlling the intensity of social media use during working hours can help to achieve a win–win situation between the industry and the platform. Setting up a technical support fund for the legal industry to encourage innovative law firms and teams to carry out pilot projects relating to the application of AI and social media, through mechanisms such as financial support and result sharing, can accelerate the popularization and promotion of new information technologies in the industry. This can improve the ability of lawyers to cope with digital transformation, ultimately enhancing lawyer job performance.

6. Conclusions

Based on in-depth research, we conducted careful observations and investigations based on the concept of lawyer job performance. Through constructing a Bayesian network to analyze 313 questionnaires about Chinese lawyers’ use of AI and social media tools, we successfully drew a causal graph, which clearly revealed how variables related to AI and social media affect the work performance of lawyers. This study, based on the averaged ensemble Bayesian network method, reveals the impact mechanism of AI tools and social media on lawyer job performance, providing important insights for the digital transformation of the legal industry. The research findings show that strain and WTC have significant mediating effects between ECU and lawyer job performance. The direct effects of AI, in terms of AWR and AL, as well as the direct impact of strain on performance, further highlight the complexity of the interactions between technological and human factors. Strain was observed as a key factor directly affecting lawyer job performance, which is consistent with the research of Cao [46], among others. In addition, as a key node in the Bayesian network, ECU was found to have the most significant impact on lawyer job performance, indicating that the excessive cognitive use of social media (i.e., irrational use) may become an important obstacle to performance improvement. Based on these findings, we suggest that individual lawyers should optimize the use of AI tools and scientifically manage their social media behaviors, particularly through reducing the posting of work-related information on social media. Law firms need to establish an AI-friendly culture and optimize their work processes accordingly. At the industry level, it is necessary to formulate regulatory guidelines, build innovation platforms, and promote technical support policies. Through multi-level collaborative efforts, the legal industry is expected to achieve comprehensive performance improvement and sustainable development driven by both AI and social media. This study not only provides specific methods through which lawyers can improve their work performance but also offers valuable references for future law students. We hope that lawyers can improve their legal literacy by learning how to effectively utilize AI tools and demonstrate professionalism on social media, thereby promoting the legal profession. It is particularly noteworthy that this improvement in ability is expected to help them provide higher-quality legal services.

7. Limitations

In a society governed by the rule of law, the legal profession exists in various countries. However, this study only selected Chinese lawyers as the research subjects, and the impacts of AI and social media on their performance may vary for lawyers in different countries and regions, characterized by different judicial environments and work requirements. Therefore, it would be interesting to conduct further research utilizing comparative methods between countries with different judicial environments and legal cultures, analyzing the commonalities and specific characteristics of influencing factors to obtain more intriguing results.

The future work associated with our model not only holds relevance for investigating the impacts of AI and social media on lawyer performance but also serves as a reference point for examining the influence of other variables. Its applicability extends beyond the legal profession and can be adapted to explore various factors and their effects on performance in different domains. Our future plan is to select additional indicators to more comprehensively analyze the factors influencing lawyer job performance. We will also further refine the problem setting and network structure, in order to enhance the model’s effectiveness. Additionally, we hope to collect feedback from lawyers across different countries and backgrounds to ensure the reliability and explanatory power of the proposed models. Through incorporating these steps, we aim to obtain more robust and informative models, thus promoting a better understanding of lawyer job performance. As our survey only targeted Chinese lawyers, the results may lack generalizability, and worldwide research is encouraged to verify our propositions.