Research on the Application Maturity of Enterprises’ Artificial Intelligence Technology Based on the Fuzzy Evaluation Method and Analytic Network Process

Abstract

The aim of this study was to study the impact of artificial intelligence (AI) on enterprises in terms of strategy, technology, business operations, and organizational management. This study used grounded theory analysis to identify the influencing factors of AI technology application maturity in Chinese enterprises. Taking Chinese film and television enterprises as an example, this study constructed an AI technology application maturity evaluation index system for enterprises based on the analytic network process (ANP) and evaluated the application maturity of AI technology in enterprises in terms of enterprise strategy, technology, business operations, and organizational management. To comprehensively evaluate and empirically analyze the application maturity of enterprise AI technology, this study calculated the index weight based on the ANP, and combined it with the fuzzy comprehensive evaluation method to construct a comprehensive evaluation model. The research results showed that intelligence strategy was the element that was believed to be most affected by the maturity of enterprise AI technology. For technology, intelligence technology and equipment were the elements that were believed to be affected the most. For business operations, smart shooting was the element that was believed to be affected the most. With respect to organizational management, corporate culture was the element that was believed to be most affected. The results showed that the proposed methods for evaluating the application maturity of enterprise AI technology are scientific and effective. The results of this study provide a reference for promoting the application of AI, implementing the intelligence transformation, and enhancing the core competitiveness of enterprises.

1. Introduction

At present, artificial intelligence (AI) is enabling the new intelligence revolution, and the evaluation of the application maturity of enterprise AI applications is becoming one of the forefront research topics in the intelligence era. A new scientific and technological revolution and industrial transformation is being led by AI. Since the Dartmouth Conference held in 1956 marked the birth of AI, and with the continuous breakthroughs and improvements in data, algorithms, and computing power, the intelligence degree of AI has increased rapidly [1]. In recent years, Artificial Intelligence-Generated Content (AIGC), represented by ChatGPT and Sora, is the result of high-quality data, a neural network model based on the deep learning model transformer architecture, and a large number of graphics processing unit (GPU) clusters. In the sixth technological revolution, the breadth and depth of the intelligence technology group led by AI are unmatched by previous technological revolutions. The fourth industrial revolution led by AI has driven the upgrading of digital intelligence in all walks of life, and has increasingly become the key driving force for promoting the sustainable development of enterprises. Under the profound influence of the internet and digital intelligence technology, the film and television industry are at the forefront of AI applications. Taking the development of AI empowering film and television enterprises as an example, AIGC’s core capabilities, such as its language expression ability, art video ability, and creative emission ability, can empower creators in four important business aspects—planning, development, production, and promotion—and will have a subversive impact on the film and television industry.

AI improves total factor productivity by optimizing factor allocation efficiency, reducing enterprise costs, and improving enterprise research and development and innovation capabilities, thus promoting economic growth. On the one hand, AI can enable efficient and accurate data sharing, information exchange, and technology sharing within and across the systems of various departments of the enterprise, as well as accelerate the factors to give full play to their comparative advantages, and optimize the efficiency of factor resource allocation. On the other hand, “AI + big data” assists enterprises in scientific production, governance, and decision-making to reduce risks, which can significantly reduce enterprise operating costs and improve efficiency and productivity. Currently, the process of enterprise intelligence transformation is gradually accelerating. In this context, exploring the application maturity of enterprise AI technology is highly important for seizing the opportunities brought by the AI revolution, improving the competitiveness of enterprises, promoting the growth of enterprise revenue, and implementing intelligence transformations. On this basis, through multiple rounds of expert interviews and questionnaire surveys, this study constructed a comprehensive evaluation model of enterprise AI technology application maturity based on the Fuzzy-ANP method, empirically analyzed enterprise AI technology application maturity from the aspects of enterprise strategy, technology, business operations, and organizational management, and explains the importance of AI in enabling enterprise development. This information can help enterprises judge their own AI technology application maturity stage and provide strategic suggestions for enterprises to improve their AI technology application maturity.

2. Literature Review

AI is an emerging technology science based on computer science, which simulates, extends, and develops human intelligence through the research and development of theories, methods, technologies, and application systems [2]. AI technology not only has the common characteristics of digital technology such as reprogramming and data homogeneity [3], but also has adaptive learning abilities and the operational decision-making black box behind it [4]. To date, the academic circle has conducted in-depth research on the application of AI technology, including the practice of AI, the development of and paradigm change in AI, and the governance and application of AI [5,6]. With the characteristics of intelligence and digitalization, AI can perform mechanized and repetitive work through software programming and complete a variety of high-precision tasks, which is widely used in various industries [7]. AI technology has brought changes to all dimensions of enterprise management and industrial development. The organizational management and marketing of enterprises have been profoundly affected by AI [8,9]. For example, some scholars found that AI has had an important impact on the audit industry [10]. Through the use of AI technology, enterprises can effectively enhance intelligent manufacturing, promote internal information communication, help reduce costs, and improve total factor productivity through a variety of mechanisms [11,12,13].

In terms of influence mechanism, it mainly includes reducing costs, improving production efficiency, enhancing management decision-making ability, and promoting product innovation and other mechanisms [14,15]. AI technology has greatly accelerated the speed of research and development, improved the efficiency of research and development [16], reduced the number of employees, improved the productivity of employees, and enhanced the competitiveness of enterprises [17]. Some scholars have studied the impact of AI on technological innovation in the Chinese manufacturing industry, and found that the application of AI technology accelerates knowledge creation and technology spillover, promotes enterprises to improve their learning ability and absorption ability, promotes innovation by increasing research and development and talent input [18], and realizes internal resource optimization and reorganization through the technology spillover effect [19]. AI technology can create innovation momentum for products, process structures, and business models and play an enabling role in the whole process of enterprise production and operation by promoting knowledge spillover, assisting precise decision-making, innovating digital thinking, innovating organizational structure, and other paths [20]. AI can improve the effectiveness of search and recombine broader knowledge, thereby improving the ability to produce knowledge, with the positive impact of accelerating economic growth. Some scholars regard the process of knowledge creation as a process of reorganizing existing knowledge, and found that the development of AI not only helps people to discover new knowledge, but it can also mine the existing knowledge and technology market data of enterprises, and optimize the grouping process for the existing knowledge [21]. Some scholars have pointed out through case analysis that AI helps optimize management decisions by promoting information feedback, thus promoting enterprise innovation [22]. The application of AI improves the market value of enterprises by promoting product innovation [23]. From the perspective of product innovation and innovation processes, scholars analyzed the positive role played by AI technology in the process of problem exploration and solution selection, and found that the application of AI can significantly promote the improvement of enterprise innovation performance [24,25].

Artificial intelligence-generated content (AIGC) technology has a great impact on the production and operation of enterprises. AIGC is a way to guide AI models to generate named content through instructions. The significance of the application of the AIGC large model is not only to improve work efficiency, but more importantly, but also the transformation of the production mode of enterprises and major innovations in products and services. As the underlying intelligent infrastructure of enterprises, the AIGC large model enables enterprises to conduct research and development, production, operation and management, marketing, and other businesses, and improves the production efficiency and decision-making ability of enterprises [26]. The development of AIGC technology represented by ChatGPT has changed the existing work flow. Various intelligent systems developed based on AIGC can greatly improve the production efficiency of enterprises, improve the service response speed and quality, provide better insights into industry data to assist decision-making [27,28], achieve cost reduction and efficiency increases, and enhance the competitiveness of enterprises. They have a great influence on the production practice of enterprises [29,30]. The technological evolution mode of “AIGC+” links intelligent production processes to intelligent product services, realizes the integration of multiple technologies, promotes the collaborative innovation of the supply chain, and enables the high-quality development of enterprises. AIGC products enhance the intelligent level of product design, production, marketing, and operations in the industrial field, which not only improves competitiveness, but also improves the market response speed [31]. AIGC technology will also have a profound impact at the organizational and individual levels, reshaping the knowledge-creation process within organizations [32].

The concept of the maturity model first appeared in the 1970s and was specifically used in the software engineering industry [33]. Due to the ease of use and effectiveness of the maturity model, the maturity model has been gradually applied to more fields, which has confirmed the versatility of the maturity model in multiple fields. The numerical intelligence maturity model in academic circles covers a wide range of areas. Scholars combine the maturity model with the application of digital technology to describe the relevant features of digital intelligent maturity models from the aspects of the model’s purpose, applicability, and application. Many studies have confirmed that improving the maturity of AI is conducive to the long-term development of enterprises, and the digital intelligence maturity model can help enterprises identify the current state and the process of evolution [34]. The higher the maturity of digital intelligence technology, the more far-sighted and innovative the enterprises can be to promote value co-creation and improve the efficiency and effect of digital intelligence transformation [35]. The purpose of using the digital intelligence maturity model is to evaluate the situation of the enterprise’s digital intelligence transformation, effectively describe the maturity level, find the path to improve the maturity, and then guide the enterprise to carry out the digital intelligence transformation. In addition, there are some studies that focus on specific areas, including the application of the model in the enterprise management dimension, video industry, education dimension, and so on. The application scope of the digital intelligence maturity model covers all aspects of enterprise business operations in the process of digital intelligence transformation [36,37]. Some scholars have empirically studied the factors influencing the application of AI video-generation tools in the video industry [38]. In terms of education, digital maturity models that integrate digital technologies into teaching and learning can improve educational experience and quality [39]. The factors influencing the long-term effective operation of artificial intelligence include technology, personnel, and data [40]. Value chain analysis is helpful for enterprises to fully grasp their competitive advantages, understand the development stage of digital intelligence, and then measure the degree of interconnection among enterprises forming the value chain [41]. The method used in the model is mainly based on the questionnaire set by the Likert scale for qualitative and quantitative measurements. The main qualitative methods used in the model development stage include the Delphi method and case study method, and the main quantitative measurement methods include the fuzzy analytic hierarchy process, hierarchical cluster analysis, Monte Carlo simulations, etc. [42].

On the basis of the above, scholars have studied the application maturity of enterprise AI technology from different perspectives. These studies have at least two shortcomings. First, previous research has not systematically carried out a quantitative analysis of the application maturity of AI technology at the enterprise level. At present, research on the application maturity of enterprise AI technology has focused mainly on explanatory qualitative analyses, and the amount of empirical research at the enterprise level is relatively insufficient. Second, at present, there are few studies directly related to this topic. There are a few studies on a single dimension, and the research methods are relatively undiversified. A set of feasible comprehensive methods that combine qualitative and quantitative methods has not been built to evaluate the application maturity of enterprise AI technology. On this basis, this study explored evaluation index systems for enterprise AI technology application maturity based on existing studies, evaluated enterprise AI technology application maturity from the aspects of enterprise strategy, technology, business operation, and organizational management, and used Chinese film and television enterprises as examples. The data were gathered through multiple rounds of expert interviews and questionnaires. Using the network analytic network process (ANP) and fuzzy comprehensive evaluation method, this study empirically analyzed the influencing factors and importance degree of enterprise AI technology application maturity, and calculated a comprehensive maturity evaluation score. These research results have theoretical value and important practical significance in the formulation of policies related to enterprise intelligence transformation. The results will guide enterprises to implement innovation and changes by choosing a path to establish value creation, and will provide references for promoting the application of AI in enterprises, encouraging intelligence transformation, and enhancing enterprises’ core competitiveness.

The rest of the article is divided into three sections: Section 3 describes the research method, the construction of the enterprise AI technology application maturity evaluation index system, which classifies enterprise AI technology application maturity, and the building of the ANP structure model. Section 4 discusses the enterprise AI technology application maturity evaluation results via the Fuzzy-ANP method through a reliability scale and validity analysis, evaluation index weight analysis, and fuzzy comprehensive evaluation analysis. The conclusions and some practical suggestions for promoting the application of AI in enterprises are outlined in Section 5.

3. Research Methods

In this study, the factors influencing enterprise AI technology application maturity were first identified via grounded theory, an evaluation index system for enterprise AI technology application maturity was established, the importance of the indexes was evaluated via the ANP, and a comprehensive evaluation model was established that uses the fuzzy comprehensive evaluation method.

3.1. Evaluation of the Application Maturity of Enterprise AI Technology

3.1.1. Construction of the Evaluation Index System

An evaluation index system to measure the application maturity of enterprise AI technology was constructed via grounded theory. The data, which were analyzed using grounded theory, came from expert interviews and enterprise AI-related data collected from various channels. Grounded theory is different not only from the single top-down logical deduction of quantitative empirical research but also from the inductive method usually used in qualitative research. In grounded theory, induction and deduction are used interchangeably. In the process of induction, researchers collect a large amount of data through participatory observation and in-depth interviews and then continuously analyze, compare, classify, conceptualize, and categorize the data, thus gradually forming a theoretical framework. After summarizing and absorbing the content of the data, this study deduced the possible connections between categories according to the existing data, and then deduced the preliminary concepts, which is the process of deduction. Previous studies rarely describe the structural dimension of the application maturity of AI technology, and the application content of AI technology in enterprises is diverse and complex. Therefore, this study adopted grounded theory to explore the composite dimensions of the application maturity of enterprise AI technology.

In accordance with the operating principles proposed by grounded theory, this study encoded the data systematically according to grounded theory operational procedures and encoded the data step-by-step using NVivo 11 software. Step-by-step coding can be divided into open coding, spindle coding, and selective coding. To check the validity of the results, a quarter of the sample was reserved for a theoretical saturation test before the study, which was again incorporated into the grounded theory analysis process to verify the extent of coverage of the categories constructed by the previous theoretical model. After the test, there were no new important concepts or categories that affect the original core category, and there were no new relationship structures between categories, indicating that the theoretical saturation test of the data was passed. Finally, this study extracted 67 initial concepts, 24 initial categories, 8 main categories, and 4 core categories after classifying and summarizing the original data. The specific meanings of the main categories are shown in

Table 1. Result of the selective coding of the application maturity of enterprise AI technology.

Core Category	Main Category	Specific Meanings of the Main Category
Intelligent strategy	Intelligence strategy	Understands the importance of intelligence from a strategic height, produces intelligent layouts, and implements an intelligence strategy
Intelligence technology	Intelligence technology level	Develops intelligence technologies and intelligent production systems, and uses intelligent technologies for platform construction
	Business process integration	Implements business process integration based on intelligence technology
Intelligent business operations	Film and television project planning and development	Will achieve intelligence in assisting script creation, script evaluation, and casting
	Film and television project production	Will achieve intelligence in shooting, production, post-production, and other production links
	Film and television project marketing	Will achieve intelligence in content distribution, IP management, and other aspects of intelligence
Intelligent organizational management	Organizational management mode	Will implement intelligence transformations of the organizational management mode in human resource management, financial management, and other aspects based on intelligent technology
	Corporate cultural conception	Will implement an enterprise culture construction which is suitable for intelligent construction

.

Through grounded theory analyses, and taking into account the basic principles of constructing indexes, 4 first-level indexes, 8 second-level indexes, and 24 third-level indexes were selected to construct an initial evaluation index system for the application maturity of the AI technology of film and television enterprises, which allows for the evaluation of the application maturity of enterprise AI technology at all levels. On the basis of the results of the expert interviews, the evaluation indexes were screened, and by deleting, merging, and adding indexes, an evaluation index system for the application maturity of the AI technology of film and television enterprises, which consisted of 4 first-level indexes, 7 second-level indexes, and 21 third-level indexes, was finally determined, as shown in

Table 2. AI technology application maturity evaluation index system for enterprises.

First-Level Index	Second-Level Index	Third-Level Index	Description of Index
Intelligence strategy (IS)	Intelligence strategy (ISA)	Intelligent strategic thinking	Enterprises can deeply understand the significance of intelligence from a strategic height
		Intelligent strategic planning	Enterprise management has a clear vision of digital intelligence empowerment and includes intelligence transformation into strategic planning
		Implementation of intelligence strategy	Enterprise management supports the necessary intelligent investment, supports the use of intelligent tools, and sets clear intelligence goals to ensure the implementation of intelligence strategy
Intelligence technology (IT)	Intelligence technology level (ITA)	Intelligence technology and equipment development	Enterprises focus on research and development of digital intelligence technology and equipment
		Introduction of intelligence technology and equipment	Enterprises focus on the introduction of digital intelligence technology and equipment
		Intelligence platform construction	Enterprises pay attention to the use of intelligence technology for platform construction
	Intelligent business process integration (ITB)	Intelligent business process integration	Enterprises pay attention to the use of intelligence technology for business process integration
Intelligent business operations (IB)	Film and television project planning and development (IBA)	Intelligent script creation	Enterprises use intelligence technology to assist script creation
		Intelligent script evaluation	Enterprises use intelligence technology to realize intelligent script evaluation
	Film and television project production (IBB)	Intelligent shooting	Enterprises use intelligence technology to realize intelligent shooting processes
		Intelligent production management	Enterprises use intelligence technology to realize intelligent production management processes
		Development and application of digital assets	Enterprises use intelligence technology to develop and apply assets
		Intelligent postproduction	Enterprises use intelligence technology to realize intelligent postproduction processes
		Intelligent review film	Enterprises use intelligence technology to realize intelligent film review processes
	Film and television project marketing (IBC)	Intelligent marketing	Enterprises use intelligence technology to carry out fine propaganda and realize the intelligent marketing tools and services
		Intelligent distribution	Enterprises often distribute film and television works through internet channels
		Intelligent IP management	Will achieve intelligence in content distribution, IP management, and other aspects of intelligence
Intelligent organizational management (IO)	Organizational management mode (IOA)	Intelligent cooperation alliance	Enterprises have established an intelligent cooperation alliance among enterprises
		Intelligent financial management	Enterprises use intelligence technology to realize intelligent financial management
		Intelligent human resource management	Enterprises use intelligence technology to realize intelligent human resource management
		Intelligent leadership	Business leaders have the ability to promote the development of intelligence
		Intelligent talent construction	Enterprises pay attention to the construction of intelligence talents, and often carry out intelligence skills training for employees
	Corporate cultural conception (IOB)	Enterprise culture conducive to the construction of intelligence	Enterprises have established and propagated the enterprise culture that is conducive to the construction of intelligence

. Taking film and television enterprises as examples, from the perspective of maturity composition, the application maturity of enterprise AI technology was assessed mainly based on the aspects of intelligence strategy, intelligence technology, intelligent business operations, and intelligent organizational management. The maturity of the intelligence strategies reflects the importance that enterprises attach to the application of AI technology, the intelligent layout, and the implementation of intelligence strategies from a global perspective. Intelligence technology maturity reflects the enterprise’s intelligent technology level and the integration levels of the business processes. Intelligent business operations included the application of intelligence technology in film and television project planning and development, project production, project marketing, and other links. Intelligent organizational management refers to the transformation of the organizational management mode and the cultural conception embodied by enterprises against the background of digital intelligence.

3.1.2. Application Maturity Level Classification of Enterprise AI Technology

This study classified the application maturity of enterprise AI technology, aiming to provide enterprises with quantitative evaluation tools to make targeted and progressive improvements, and to continuously improve the application level of enterprise AI technology. The capability maturity model (CMM), first proposed by Carnegie Mellon University, is a best practice tool used to guide the enterprise software quality improvement process. The starting point of the CMM is to look at the status of the enterprise, evaluate the process management level of the organization by identifying the maturity level, and continuously look for weak links, mainly to improve the existing software development process. The evaluation method based on the capability maturity model (CMM) has become a tool for science and technology evaluation and management, and many studies have introduced maturity into the evaluation of different industries. In essence, enterprise intelligence construction is also a dynamic process with specific goals, a continuous optimization of resources, combinations of capabilities, and it involves transformation and upgrading. On this basis, this study classified the levels of each index according to the evaluation index system mentioned above, combined with the existing maturity evaluation model and maturity classification criteria. The maturity evaluation level of this study mainly refers to the composition of the maturity assessment model, maturity level, capability elements, maturity requirements, maturity assessment content, and assessment process, and uses the maturity level determination method specified by China’s national standards: “Maturity model of intelligent manufacturing capability (GB/T 39116—2020)” and “Maturity assessment method of intelligent manufacturing capability (GB/T 39117—2020)” [43,44]. Moreover, based on the actual situation of the film and television enterprises, combined with expert opinions, the application maturity of enterprise AI technology was divided from low to high into five levels: the planning level, standard level, integration level, optimization level, and leading level. The maturity level divisions show a progressive state change in the degree of application of enterprise AI technology from a low level to a high level. Each level has its corresponding standards and requirements, and after meeting the corresponding standards, the enterprise can reach the next maturity level. Each level is necessary for the enterprise, and the lower level is a foundation for the higher level. A low maturity level represents a low level of AI technology application in the enterprise, and vice versa. The maturity levels and their explanations are shown in

Table 3. Application maturity levels of enterprise AI technology and their explanations.

Maturity Level	Explanation
Planning level	Enterprise management lacks intelligent thinking, insufficient understanding of the value of AI technology application, has not yet formed a consensus, has only formulated a preliminary intelligence strategic plan, and has just begun to promote intelligence construction.
Standard level	Enterprise management and a few technical personnel realize the importance of AI technology, the enterprise has developed intelligence strategic planning and continues to promote intelligence construction, but there is no system to implement intelligence.
Integration level	Enterprise management and most business personnel have entered a new level of intelligence cognition, the enterprise has developed intelligence strategic planning, begun to consciously prepare for the development of more comprehensive AI applications, and an intelligence strategy has been preliminarily implemented.
Optimization level	All employees of the enterprise can fully realize the important value of intelligence, and the enterprise has formulated a clear intelligence strategic plan, established a long-term mechanism for continuous optimization and innovation, and cooperated with multiple departments to smoothly promote the application of AI technology projects, and the intelligence strategy has been effectively implemented.
Leading level	All employees of the enterprise deeply understand the importance of intelligence, adhere to the strategic guidance, comprehensively promote the application of AI technology, and periodically organize a strategic review, summarize the phased results, evaluate the gaps, dynamically optimize and adjust the strategic plan, and comprehensively guarantee the implantation of an intelligence strategy.

.

The measurement of enterprise AI technology application maturity based on maturity levels relies mainly on questionnaires to obtain data. Therefore, the design of a questionnaire needs to be comprehensible, comprehensive, relevant, consistent, and applicable [45]. In terms of questionnaire design, based on the established evaluation index system and the combination of grounded theory analysis of the research results and the literature, the questionnaire was designed with four dimensions: intelligence strategy, intelligence technology, intelligent business operations, and intelligent organizational management. Enterprises at different maturity levels have the corresponding characteristics, and different indexes have different characteristics at different maturity levels. Based on these characteristics, the initial measurement scale of the application maturity of enterprise AI technology was designed, and the corresponding grade characteristics constituted the measurement scale used to measure the maturity of the third-level indexes. The model can be converted into a five-level Likert scale with 23 items; see previous studies for details.

3.2. Comprehensive Evaluation Model of the Application Maturity of Enterprise Artificial Intelligence Technology Based on the Fuzzy-ANP Method

Based on the evaluation index system for the application maturity of enterprise AI technology, this study used the ANP and the fuzzy comprehensive evaluation method to construct a comprehensive evaluation model for enterprise AI technology applications.

3.2.1. ANP Method

At present, many methods are used to determine the weights of evaluation indicators; the commonly used comprehensive evaluation methods include the expert scoring method, the analytic hierarchy process, data envelopment analysis, the artificial neural network evaluation method, and the grey comprehensive evaluation method. The expert scoring method and analytic hierarchy process are simple to perform; however, they are highly subjective and do not consider correlations between elements. The data envelopment analysis method is applied to evaluate the relative effectiveness of multi-input and multi-output methods. The artificial neural network evaluation calculation process is more complicated, and the grey comprehensive evaluation is mostly used in element correlation analysis.

Based on the above analysis, the ANP, as a multicriteria decision-making method, can integrate the knowledge and experience of experts through the comprehensive use of qualitative and quantitative indicators, and is especially suitable for dealing with complex engineering problems with multiple objectives and multiple levels as well as complex correlations among indicators, such as evaluation indicators of the artificial intelligence technology application maturity of enterprises.

The ANP is a multi-attribute decision model developed by the famous American operations research scientist Saaty. Its unique feature is that the ANP allows for the consideration of the interdependence and feedback between system elements [46]. With the development of the analytic hierarchy process (AHP), the ANP contains interdependencies within and between hierarchies, which objectively reflect the relative weights of various influencing factors. In complex environments, the ANP is more practical than the AHP is. Under the ANP method, the system elements are composed of two types of elements: control layer elements, including the overall objective and decision criteria, and network layer elements, which may be interrelated and influence each other, forming a network structure composed of individual element groups and influencing relationships. Thus, in the ANP, not only does the importance of the decision criteria determine the importance of the elements, but the importance of the elements also determines the importance of the individual criteria. Considering that the evaluation of the application maturity of enterprise AI technology is a complex system problem, it is more scientific and reasonable to evaluate the evaluation index model of the application maturity of enterprise AI technology using the ANP method in this study, which can consider the complex relationships among various indexes of the application maturity of enterprise AI technology.

The basic steps of the ANP method are as follows:

(1) Establish the ANP structure model. The decision problem of the system is analyzed, and an ANP structural model of the network structure is established.

(2) Construct the hypermatrix of the ANP structure. Suppose that the control layer has n criteria, denoted as B_s (s = 1, 2, …, n), and the network layer has m element groups, denoted as C_i (i = 1, 2, …, m), where each element group contains n_i elements, denoted as C_i1, C_i2, …, C_ini. Then, by a pairwise comparison of elements and between elements, the judgment matrix is constructed, the normalized eigenvector is calculated, and the supermatrix W is calculated.

(3) A weighted supermatrix of the ANP structure is constructed. Each column of the supermatrix is normalized, the normalized eigenvector (a_1k, a_2k,…, a_mk)^Τ is calculated, the weighted matrix A is summarized, and then the weighted supermatrix is obtained. The calculation formula is as follows:

$$\mathrm{A}=\left[\begin{array}{cccc}{a}_{11}& a_{12}& \cdots & a_{1m}\\ a_{21}& a_{22}& \cdots & a_{2m}\\ ⋮& ⋮& ⋮& ⋮\\ a_{m1}& a_{m2}& \cdots & a_{mm}\end{array}\right]$$

(1)

$$\overline{W_{ik}}=a_{ik}\left(W_{ik}\right)$$

(2)

where i = 1, 2, …, m and k = 1, 2, …, m. The values in the weighted supermatrix correspond to the global weights of each element, and the sum of the elements in each column of the matrix is 1.

(4) A limit-weighted supermatrix of the ANP structure is constructed. Owing to the complexity of the ANP calculation process, it is necessary to obtain stable element weights via the method of finding the limit. By using the limit existence theorem of the supermatrix, if the supermatrix W is an internally dependent hierarchical structure, and every column is a normalized eigenvector about the eigenvalue 1, then the limit $W^{\mathrm{\infty }}$ of the supermatrix exists.

$$W^{\mathrm{\infty }}=\underset{v⟶\mathrm{\infty }}{\lim }{W}^v$$

(3)

3.2.2. Fuzzy Comprehensive Evaluation Method

Fuzzy comprehensive evaluation is a process of considering and analyzing each index factor based on fuzzy mathematics theory, and transforming a qualitative evaluation into a quantitative evaluation. The theory was proposed by American cybernetics expert Professor Zadeh in 1965. To solve many fuzzy phenomena in which it is difficult to determine the boundaries and measurement criteria in the objective world, fuzzy mathematics theory uses grade assignments to determine the membership levels of factors that are difficult to quantify, such as qualitative indexes, and obtains the final evaluation results through evaluation sets.

The main steps of a fuzzy comprehensive evaluation are as follows:

(1) Establish the evaluation index set of the evaluation object. According to the description of the maturity evaluation index system, the evaluation index set for each level is established. The first-level index set of the maturity evaluation is U_Z = {IS, IT, IB, IO}, IS, IT, IB, and IO, which represent the intelligence strategy, intelligence technology, intelligent business operations, and intelligent organizational management, respectively. The second-level index sets of the maturity evaluation are as follows:

U_IS = {ISA}, U_IT = {ITA}, U_IB = {IBA, IBB, IBC}, U_IO = {IOA, IOB}

The same is true for the third-level index set.

(2) Construct a set of comments for each index. The comment set is a set of all possible evaluation levels, which is the overall distinction of the evaluated object, usually represented by V, where V = {V₁, V₂, V₃,…, V_m}. This study determined the evaluation criteria, divided the evaluation levels through expert consultation, and established a comment set composed of 5 evaluation levels, namely, m = 5. V = {V₁, V₂, V₃, V₄, V₅} = {planning level, standard level, integration level, optimization level, and leading level}.

(3) Determine the weight set of the evaluation indexes. The weight set of the evaluation index is the weight set corresponding to the index set at each level. In this study, the weights of all levels of indexes determined via the ANP method were used as the weight vector of a fuzzy comprehensive evaluation.

(4) Establish a fuzzy relation matrix. The fuzzy relation matrix is calculated according to the fuzzy mapping between the evaluation index set and the evaluation set.

(5) Determine a synthetic fuzzy comprehensive evaluation result vector. In accordance with the fuzzy comprehensive evaluation model, a single-factor fuzzy comprehensive evaluation and whole fuzzy comprehensive evaluation are carried out.

3.2.3. ANP Structure Model for the Maturity Evaluation of Enterprise AI Technology

When the ANP method is used to analyze the application maturity evaluation of enterprise AI technology, it is necessary to determine the evaluation target, elements, and element groups. According to the established evaluation index system, the maturity evaluation index of intelligence strategy, intelligence technology, intelligent business operations, and intelligent organizational management are were into 7 element groups: intelligence strategy, intelligence technology level, film and television project planning and development, film and television project production, film and television project marketing, the organizational management mode, and the corporate culture concept.

In the model, ISA = (ISA1, ISA2, ISA3), ITA = (ITA1, ITA2, ITA3), IBA = (IBA1, IBA2, IBA3), IBB = (IBB1, IBB2, IBB3, IBB4, IBB5), IBC = (IBC1, IBC2), IOA = (IOA1, IOA2, IOA3, IOA4, IOA5), and IOB = (IOB1).

Because the indexes are not independent of each other, there is a certain interdependence and feedback. To construct an ANP evaluation model, it is necessary to determine the influence relationships among the evaluation indexes of the application maturity of enterprise AI technology. After the influence relationships among the indexes were determined, the ANP network structure model of maturity evaluation was further constructed to objectively evaluate the application maturity of enterprise AI technology. The ANP structure model is shown in Figure 1. In the ANP network structure model, the first layer is the target layer, that is, the goal of building the model is “application maturity evaluation of enterprise AI technology”. The second layer is the decision-making criteria layer, which includes all first-level indexes, and the target layer and the decision criteria layer together constitute the control layer of the model. The network layer part is a set of network layer elements composed of 7 second-level indexes, and the third-level indexes are the elements that interact internally. The arched arrow on the element group indicates that there is a mutual influence between the elements within it. The arrow pointing from one element group to another element group indicates the influence relationship between the element groups. When the element groups are interrelated, a “bidirectional arrow” is used. If the influence relationship between element groups is one way, it is represented by a “single arrow”. Since the ANP method involves many calculations of supermatrices and limit supermatrices, it is difficult to solve manually. Therefore, this study used the Super Decision (SD) software (version 3.2) launched by Saaty, which is based on ANP theory, to construct and analyze the ANP structure model for evaluating the application maturity of enterprise AI technology [47].

3.3. Sample Selection and Data Sources

This paper discusses issues related to the application maturity evaluation of enterprise AI technology. Considering the availability of data, Chinese film and television enterprises were selected as research samples. In this study, the data for the evaluation index system for the application maturity of enterprise AI technology come from various channels. In this study, sample data were collected from multiple channels via data triangulation to ensure the reliability and validity of the research conclusions. To ensure that the application maturity of the enterprise AI technology index system constructed in this study is in line with the actual industry, several different types of film and television enterprises were investigated, interviewed, and analyzed. We invited 32 experts to participate in semi-structured in-depth interviews. Six of the experts were from relevant authorities in the industry, such as the department heads of the National Radio and Television Administration and the chairpersons of the China Television Production Committee. These experts are responsible for the industry’s supervision and management. They draw up guidelines so that enterprises can standardize the research and application of artificial intelligence technology. Another 26 AI experts came from the management, development, and business functions of large- and medium-sized Chinese film and television enterprises, such as the CEO of the listed company iQiyi and the vice president of Alibaba Pictures. The interviews included a combination of face-to-face interviews, field visits, and telephone interviews, and each interview lasted 30–60 min. The interview content focused on the dimensions of enterprise AI technology application maturity. An outline of the collected data is shown in

Table 4. Data collection.

Sample Type	Collection Channel	Sample Size
Interview report	Face-to-face interviews and field visits with 20 people	32 reports, approximately 40,000 words in total
	Telephone interviews with 12 people
Books, literature, expert comments, news reports, corporate official website materials	Network channel	96 documents, totaling approximately 60,000 words

.

To further improve the quality of the scale, before the formal questionnaire survey, data were collected in the form of small-scale questionnaires, and the initial scale was pretested. SPSS 26.0 and AMOS 26.0 software were used to analyze the reliability and validity of the scale, and appropriate modifications were made to the initial scale according to the evaluation results to provide appropriate data collection tools for empirical research and finally, a formal questionnaire was developed. Since completing this questionnaire requires a certain understanding and knowledge of the application of AI technology in enterprises, the test questionnaire was distributed to middle and senior managers of film and television enterprises, as well as film and television practitioners who are familiar with the application of AI technology in film and television enterprises, and who have worked for more than 10 years. A total of 76 questionnaires were sent out, 60 were recovered, and 58 were valid.

To comprehensively evaluate the application maturity of enterprise AI technology, questionnaires were used to collect the data. On the basis of previous relevant research results, this study used the ANP method to collect industry experts’ opinions through enterprise research, analyzed and determined the influence relationships between the indicators based on the evaluation index system, and built an ANP network structure model. A questionnaire was designed on the basis of the network structure model, and expert opinions were collected via a questionnaire survey. Sixteen industry experts, enterprise executives, and university professors were invited to score the importance of the indexes at all levels and compare the importance of the indexes in the evaluation index system; we recorded the original data and obtained a judgment matrix after data processing. Using SD software to construct the ANP supermatrix and weighted supermatrix, the quantitative relationships between the factors were obtained. Finally, the limit supermatrix was calculated, and the weights of each evaluation index were obtained. The index weight reflects the degree of importance of each index in the evaluation process, and is a measure of the relative importance of the evaluation index.

To prove that the maturity evaluation model constructed in this study has good applicability, several typical Chinese film and television enterprises were selected for the empirical research study. Using the given level and evaluation criteria, the questionnaire was designed according to the evaluation index system. Ten middle and senior managers and practitioners from each company formed an expert group, and the experts were invited to face-to-face interviews about their understanding of the company and their own work experience and actual feelings. The application maturity of enterprise AI technology was subsequently evaluated.

4. Results and Discussion

4.1. Reliability and Validity Tests of the Scale

In this study, SPSS 26.0 software was used to test the reliability of the questionnaire. The reliability test results of the variables in this study are shown in Table 4. The reliability test revealed that the Cronbach’s alpha values corresponding to the four dimensions of this study were all greater than 0.7 and were 0.926, 0.881, 0.938, and 0.947, respectively, indicating good internal consistency among all the dimensions of the questionnaire. After the deletion of ITB1, the Cronbach’s alpha value was 0.911, which was greater than the Cronbach’s alpha value of ITB1 (0.881), and the CITC value was low. After IBC3 was deleted, the Cronbach’s alpha value was 0.944, which was greater than the Cronbach’s alpha value for IBC3 (0.938). Therefore, the items were screened, ITB1 and IBC3 were deleted, and the remaining 21 items were ultimately eliminated. The two indexes of “intelligent business process integration” and “intelligent IP management” in the corresponding evaluation index system were also eliminated. After the first reliability test, adjustments were made and the second reliability test was carried out using the modified scale. The results of the second reliability test are shown in

Table 5. The reliability test results using the scale.

Variable Dimension	Item	CITC	Square Multiple Correlation	Cronbach’s Alpha Value of Deleted Items	Cronbach’s Alpha
Intelligence strategy (IS)	ISA1	0.893	0.802	0.862	0.926
	ISA2	0.849	0.752	0.895
	ISA3	0.812	0.673	0.922
Intelligence technology (IT)	ITA1	0.837	0.701	0.859	0.911
	ITA2	0.826	0.683	0.881
	ITA3	0.832	0.692	0.875
Intelligent business operations (IB)	IBA1	0.877	0.868	0.939	0.949
	IBA2	0.772	0.785	0.945
	IBB1	0.777	0.735	0.945
	IBB2	0.860	0.879	0.940
	IBB3	0.781	0.775	0.945
	IBB4	0.852	0.854	0.941
	IBB5	0.804	0.739	0.943
	IBC1	0.784	0.730	0.944
	IBC2	0.716	0.689	0.948
Intelligent organizational management (IO)	IOA1	0.841	0.748	0.939	0.947
	IOA2	0.841	0.743	0.937
	IOA3	0.807	0.698	0.941
	IOA4	0.783	0.631	0.943
	IOA5	0.901	0.877	0.929
	IOB1	0.880	0.859	0.932

.

In previous studies, validity was described from four main aspects, namely, content validity, structural validity, discriminative validity, and aggregate validity.

First, the questionnaire design of this study was based on the premise of combining the existing relevant research and referring to the relevant data in grounded theory, and the design was recognized by industry experts. The questionnaire was revised based on expert opinion, thus ensuring the content validity of the scale.

Measuring structural validity is usually performed using factor analysis. In this study, exploratory factor analysis and confirmatory factor analysis were used to test the structural validity of the scale. In this study, SPSS 26.0 software was used to perform the KMO test and Bartlett sphericity test on the sample data. The KMO value of the scale was 0.870, which is greater than 0.7, and the Bartlett sphericity test chi-square was significant (p < 0.001), indicating that the scale data met the factor analysis conditions. Principal component analysis was used to extract factors in the form of feature roots greater than 1. As shown in

Table 6. Total variance explained.

Component	Initial Eigenvalue			Sum of the Squared Loads Extracted			Load Sum of Squares Rotated
	Total	Variance%	Accumulate %	Total	Variance%	Accumulate %	Total	Variance%	Accumulate %
1	12.501	59.530	59.530	12.501	59.530	59.530	6.130	29.190	29.190
2	2.133	10.158	69.688	2.133	10.158	69.688	4.700	22.382	51.572
3	1.456	6.931	76.620	1.456	6.931	76.620	3.461	16.483	68.055
4	1.039	4.946	81.566	1.039	4.946	81.566	2.837	13.511	81.566
5	0.711	3.385	84.951
6	0.535	2.546	87.497
7	0.450	2.143	89.640
8	0.372	1.773	91.413
9	0.323	1.539	92.952
10	0.277	1.320	94.273
11	0.248	1.182	95.455
12	0.214	1.020	96.475
13	0.178	0.849	97.325
14	0.129	0.613	97.938
15	0.105	0.499	98.438
16	0.089	0.425	98.862
17	0.067	0.319	99.182
18	0.058	0.276	99.458
19	0.055	0.263	99.721
20	0.032	0.153	99.874
21	0.027	0.126	100.000

, there were four factors with eigenvalues greater than 1; the cumulative variance explanation rate of these four factors after rotation was 81.566%, which is greater than 60%, and the information expression met the statistical requirements.

Through the orthogonal rotation of the factor load matrix using the maximum variance method, combined with the factor load coefficient, the research scale in this section had a total of 21 items, which were divided into four factors, and the composition of the four factors was consistent with the theoretical construction of the research design. As shown in

Table 7. Load matrix after rotating the propagation effect index factor.

	Component
	1	2	3	4
ISA1			0.913
ISA2			0.878
ISA3			0.829
ITA1				0.661
ITA2				0.821
ITA3				0.744
IBA1	0.826
IBA2	0.807
IBB1	0.758
IBB2	0.794
IBB3	0.692
IBB4	0.773
IBB5	0.774
IBC1	0.636
IBC2	0.613
IOA1		0.735
IOA2		0.747
IOA3		0.753
IOA4		0.651
IOA5		0.718
IOB1		0.773

The value of the blank spaces is less than 0.5.

, the absolute value of the factor load corresponding to all the items in the scale was greater than 0.5. This shows that the scale has high structural validity.

Both aggregate validity and discriminative validity were studied via a confirmatory factor analysis. AMOS 26.0 software was used for the confirmatory factor analysis. The relevant results of the scale are summarized in

Table 8. Convergent validity test results.

Variable Dimension	CR	AVE
Intelligence strategy (IS)	0.930	0.817
Intelligence technology (IT)	0.914	0.779
Intelligent business operations (IB)	0.950	0.681
Intelligent organizational management (IO)	0.949	0.755

. After the aggregation validity test, the CR value of each factor was greater than 0.7, and the AVE value was greater than 0.5. Therefore, the scale had high aggregation validity. To evaluate whether the discriminative validity among various factors of enterprise AI technology application maturity was sufficient, this study used the AVE square root judgment method to test discriminative validity. The relevant results of the scale are shown in

Table 9. Discriminant validity test results.

Variable Dimension	Intelligence Strategy (IS)	Intelligence Technology (IT)	Intelligent Business Operations (IB)	Intelligent Organizational Management (IO)
Intelligence strategy (IS)	0.904
Intelligence technology (IT)	0.530	0.883
Intelligent business operations (IB)	0.453	0.747	0.825
Intelligent organizational management (IO)	0.575	0.821	0.761	0.869

The diagonal number is the AVE square root value.

. According to the discriminant validity test, the AVE square root value of the factor intelligence strategy was greater than the maximum value (0.575) of the absolute value of the correlation coefficient between the factor intelligence strategy and the other three factors, indicating that it has good discriminant validity. Similarly, the AVE square root values of the factors intelligence technology, intelligent business operations, and intelligent organizational management were greater than the correlation coefficients between them and the other three factors, indicating that the scale in this study has good discriminant validity.

After an expert consultation on the evaluation index system, and the reliability and validity tests of the scale, items ITB1 and IBC3 were deleted to increase the reliability and validity of the items and, ultimately, the remaining 21 items were retained in the enterprise AI technology application maturity scale to provide support for the subsequent research.

4.2. Weight Analysis of the Evaluation Indexes Using the ANP Method

To comprehensively evaluate the application maturity of enterprise AI technology, it is necessary to evaluate the impact of each index on maturity. By using the questionnaire method to collect data and constructing a pairwise-comparison judgment matrix, the quantitative relationships among the indexes in the evaluation index system can be clarified, the weight of each evaluation index can be calculated, and the foundation for a comprehensive evaluation based on the Fuzzy-ANP method can be laid.

The ANP model was constructed based on the structure of the nine-scale method, and two importance scale factors were compared to construct a questionnaire to obtain expert opinions, and a judgment matrix of the original data was constructed. The eigenvalue method was used to test the consistency of the constructed judgment matrix. According to the consistency test rule, if the judgment matrix has a determination coefficient CR < 0.1, it indicates that after the consistency test, the index weight is within the acceptable range. After testing, the consistency test results of all the judgment matrices constructed in this study were less than 0.1, and all passed the consistency test.

SD software was used to perform the calculations. The construction of a supermatrix is a process in which the relative importance of the elements in the network layer is obtained via pairwise comparisons. Based on the judgment matrix, a weighted matrix was constructed, the normalized eigenvectors of each judgment matrix were summarized to obtain the block of a supermatrix, and then the supermatrix was obtained. The value in the supermatrix represents the importance of the column elements to the row elements, and corresponds to the local weight of each element in the element group. In accordance with the ANP principle, column normalization of the supermatrix was carried out, and the weighted supermatrix was obtained by processing the supermatrix and the weighted matrix. To obtain stable element weights, the stability of the weighted supermatrix was processed, and the limit supermatrix was further calculated via the limit matrix calculation method. Finally, the limit supermatrix was obtained, and the global weight of the index system was obtained by solving it. The final index weight table shown in

Table 10. Weights and ranks of the indexes.

First-Level Indexes	Weight	Second-Level Indexes	Weight	Third-Level Indexes	Local Weight	Global Weight	Rank
IS	0.3568	ISA	0.3568	ISA1	0.1136	0.0405	10
				ISA2	0.3193	0.1139	2
				ISA3	0.5672	0.2024	1
IT	0.0805	ITA	0.0805	ITA1	0.3986	0.0321	12
				ITA2	0.3549	0.0286	14
				ITA3	0.2465	0.0198	18
IB	0.3617	IBA	0.0593	IBA1	0.4278	0.0254	15
				IBA2	0.5722	0.0339	11
		IBB	0.2348	IBB1	0.2984	0.0701	3
				IBB2	0.2122	0.0498	7
				IBB3	0.1899	0.0446	8
				IBB4	0.2239	0.0526	5
				IBB5	0.0756	0.0178	20
		IBC	0.0676	IBC1	0.7696	0.0520	6
				IBC2	0.2304	0.0156	21
IO	0.201	IOA	0.1361	IOA1	0.2146	0.0292	13
				IOA2	0.3260	0.0444	9
				IOA3	0.1604	0.0218	17
				IOA4	0.1333	0.0181	19
				IOA5	0.1657	0.0226	16
		IOB	0.0649	IOB1	1.0000	0.0649	4

reflects the experts’ judgment of the importance of the enterprise AI technology application maturity indexes, which can be used to guide subsequent maturity evaluation research and provide a decision-making reference for enterprises to promote the application of AI technology. The weights of the first-level indexes ranked from highest to lowest were “intelligent business operations”, “intelligence strategy”, “intelligent organizational management”, and “intelligence technology”. This result provides some enlightenment. Judging from the weights of the second- and third-level indexes, film and television enterprises should first attach great importance to intelligent strategies in the application process of AI technology. The most influential factor in technology was the research and development of intelligence technology and equipment. The most influential factor in business operations was smart shooting. The most influential factor in organizational management was corporate culture. From the perspective of first-level indexes, it is especially necessary to promote the intelligent business operations from multiple perspectives, such as film and television project planning and development, project production, and project marketing. In addition, we should pay attention to the intelligent organizational management models and constantly improve the application level of AI technology.

4.3. Comprehensive Evaluation of Enterprise AI Technology Application Maturity Based on the Fuzzy-ANP Method

The CMM was first published in the software industry by the Software Engineering Institute of Carnegie Mellon University. Owing to its dynamic evolution advantages, its scope of application is constantly expanding. In essence, the application of enterprise intelligence technology is also a dynamic process with specific goals to continuously optimize resources and combine capabilities to achieve transformation and upgrading. Therefore, this study introduced a fuzzy comprehensive evaluation method to build a comprehensive evaluation model based on the ANP method, evaluated the application maturity of enterprise AI technology, and divided the maturity level of enterprise AI technology application based on the fuzzy comprehensive evaluation method scores to ensure that the evaluation results have a quantifiable hierarchy.

4.3.1. Single-Factor Fuzzy Evaluation and Construction of the Fuzzy Relation Matrix

The weight set composition of the indexes at all levels was as follows:

The first-level index weight set of a maturity evaluation was W_Z = {0.3568, 0.0805, 0.3617, 0.2010}.

The second-level index weight sets of the maturity evaluation were as follows:

W_IS = {0.3568}, W_IT = {0.0805}, W_IB = {0.0593,0.2348,0.0676}, W_IO = {0.1361,0.0649}

The single-factor fuzzy comprehensive evaluation reflects the influence of a single factor on the maturity target. This study conducted a single factor evaluation on n_i elements of the third-level index set U_Ai = {u₁⁽ⁱ⁾, u₂⁽ⁱ⁾,…, u_n⁽ⁱ⁾} of enterprise AI technology application maturity from the bottom up, and a fuzzy map was obtained: f_Bi: U_Bi → F(V).

According to fuzzy map f, the fuzzy relation R_f can be derived, and the evaluation fuzzy set of all third-level indexes belonging to the two indexes can be calculated. The method used to determine the fuzzy matrix was to have industry experts to form an evaluation group and evaluate the grade through scoring, and a fuzzy relationship matrix of single-factor evaluations was obtained. For example, when evaluating the third-level indexes “intelligence strategic thinking” of Enterprise Q, 70% of the experts thought it was at the “leading level”, 10% of the experts thought it was at the “optimization level”, 10% of the experts thought it was at the “integration level”, 5% of the experts thought it was at the “specification level”, and 5% of the experts thought it was at the “planning level”. Then, the membership degree of the intelligence strategic thinking index for all levels of evaluation were obtained: ISA1→(0.7, 0.1, 0.1, 0.05, 0.05). The fuzzy relationship matrix of the single-factor evaluation can be obtained by analogy.

$$R_{xy}=\left[\begin{array}{cccc}{r}_{11}& r_{12}& r_{13}& r_{14}\\ r_{21}& r_{22}& r_{23}& r_{24}\\ ⋮& ⋮& ⋮& ⋮\\ r_{n1}& r_{n2}& r_{n3}& r_{n4}\end{array} \begin{array}{c}{r}_{15}\\ r_{25}\\ ⋮\\ r_{n5}\end{array}\right].$$

In this study, n is the number of the third-level indexes included in each secondary index, where n = 1, 2, 3, 5, x = 1, 2, 3, 4, and y = 1, 2, 3. r_ij represents the membership degree of factor i to the evaluation level j, reflecting the fuzzy relationship between factor set U and comment set V, according to which, multiple single-factor evaluation matrices can be obtained:

$$\begin{gathered} R_{11}=\left[\begin{array}{ccc}{r}_{11}^{\left(11\right)}& r_{11}^{\left(11\right)}& r_{13}^{\left(11\right)}\\ r_{21}^{\left(11\right)}& r_{22}^{\left(11\right)}& r_{23}^{\left(11\right)}\\ r_{31}^{\left(11\right)}& r_{32}^{\left(11\right)}& r_{33}^{\left(11\right)}\end{array}\hspace{0.17em}\hspace{0.17em}\begin{array}{c}{r}_{14}^{\left(11\right)}\\ r_{24}^{\left(11\right)}\\ r_{34}^{\left(11\right)}\end{array}\hspace{0.17em}\hspace{0.17em}\begin{array}{c}{r}_{15}^{\left(11\right)}\\ r_{25}^{\left(11\right)}\\ r_{35}^{\left(11\right)}\end{array}\right] \\ {R}_{21}=\left[\begin{array}{ccc}{r}_{11}^{\left(21\right)}& r_{11}^{\left(21\right)}& r_{13}^{\left(21\right)}\\ r_{21}^{\left(21\right)}& r_{22}^{\left(21\right)}& r_{23}^{\left(21\right)}\\ r_{31}^{\left(21\right)}& r_{32}^{\left(21\right)}& r_{33}^{\left(21\right)}\end{array}\hspace{0.17em}\hspace{0.17em}\begin{array}{c}{r}_{14}^{\left(21\right)}\\ r_{24}^{\left(21\right)}\\ r_{34}^{\left(21\right)}\end{array}\hspace{0.17em}\hspace{0.17em}\begin{array}{c}{r}_{15}^{\left(21\right)}\\ r_{25}^{\left(21\right)}\\ r_{35}^{\left(21\right)}\end{array}\right] \\ {R}_{41}=\left[\begin{array}{ccccc}{r}_{11}^{\left(41\right)}& r_{11}^{\left(41\right)}& r_{13}^{\left(41\right)}& r_{14}^{\left(41\right)}& r_{15}^{\left(41\right)}\\ r_{21}^{\left(41\right)}& r_{22}^{\left(41\right)}& r_{23}^{\left(41\right)}& r_{24}^{\left(41\right)}& r_{25}^{\left(41\right)}\\ r_{31}^{\left(41\right)}& r_{32}^{\left(41\right)}& r_{33}^{\left(41\right)}& r_{34}^{\left(41\right)}& r_{35}^{\left(41\right)}\\ r_{41}^{\left(41\right)}& r_{42}^{\left(41\right)}& r_{43}^{\left(41\right)}& r_{44}^{\left(41\right)}& r_{45}^{\left(41\right)}\\ r_{51}^{\left(41\right)}& r_{52}^{\left(41\right)}& r_{53}^{\left(41\right)}& r_{54}^{\left(41\right)}& r_{55}^{\left(41\right)}\end{array}\right] \\ {R}_{51}=\left[\begin{array}{ccc}{r}_{11}^{\left(51\right)}& r_{12}^{\left(51\right)}& r_{13}^{\left(51\right)}\end{array} r_{14}^{\left(51\right)} r_{15}^{\left(51\right)}\right] \end{gathered}$$

4.3.2. Comprehensive Fuzzy Evaluation Results

For the evaluation objects in this study, the final score was obtained by gradually calculating from the lowest level to the highest level. First, the lowest level of fuzzy comprehensive evaluation was carried out; that is, the weight matrix W_xy and fuzzy matrix R_xy were synthesized by the appropriate fuzzy operators, and the fuzzy comprehensive evaluation result vector B_xy of each third-level index was obtained. The rows in R_xy reflect the membership degree of each evaluation level from the perspective of each third-level index contained in the second-level index. Therefore, the fuzzy comprehensive evaluation result vector B_xy can be obtained by synthesizing different rows with the weight matrix W_xy:

$$\begin{array}{cc}{B}_{xy}=W_{xy}\cdot R_{xy}=W_{xy}\left[\begin{array}{cccc}{r}_{11}& r_{12}& r_{13}& r_{14}\\ r_{21}& r_{22}& r_{23}& r_{24}\\ ⋮& ⋮& ⋮& ⋮\\ r_{n1}& r_{n2}& r_{n3}& r_{n4}\end{array} \begin{array}{c}{r}_{15}\\ r_{25}\\ ⋮\\ r_{n5}\end{array}\right]& \\ B_{11}=W_{11}\cdot R_{11}=\left(W_{ISA1},,,W_{ISA2},,,W_{ISA3}\right)& \cdot \left[\begin{array}{ccc}{r}_{11}^{\left(11\right)}& r_{11}^{\left(11\right)}& r_{13}^{\left(11\right)}\\ r_{21}^{\left(11\right)}& r_{22}^{\left(11\right)}& r_{23}^{\left(11\right)}\\ r_{31}^{\left(11\right)}& r_{32}^{\left(11\right)}& r_{33}^{\left(11\right)}\end{array}\hspace{0.17em}\hspace{0.17em}\begin{array}{c}{r}_{14}^{\left(11\right)}\\ r_{24}^{\left(11\right)}\\ r_{34}^{\left(11\right)}\end{array}\hspace{0.17em}\hspace{0.17em}\begin{array}{c}{r}_{15}^{\left(11\right)}\\ r_{25}^{\left(11\right)}\\ r_{35}^{\left(11\right)}\end{array}\right]\end{array}$$

Similarly, B₁₂, B₁₃, …, B₄₁, B₄₂ can be obtained, and the fuzzy comprehensive evaluation result vector B₁, B₂, B₃, B₄ can be obtained via the same method for the fuzzy comprehensive evaluation of the second-level indexes. Finally, the fuzzy evaluation matrix R is formed by combining the fuzzy evaluation results of all the first-level indexes. There were four first-level indexes in this study, so the comprehensive fuzzy evaluation matrix is $R={\left[B_1, B_2, B_3, B_4\right]}^T$.

The comprehensive fuzzy evaluation results were as follows:

$$B=W_Z\cdot R=\left\{0.3568,0.0805,0.3617,0.2010\right\}\cdot {\left[B_1, B_2, B_3, B_4\right]}^T$$

Then, using the principle of maximum membership degree, the evaluation grade corresponding to the maximum membership degree was selected as the comprehensive evaluation result. The score S corresponding to the review set and the fuzzy comprehensive evaluation result vector B were combined to transform the evaluation result vector into the quantitative score of maturity: $M=B\cdot C^T$.

4.4. Sample Test Results of the Comprehensive Evaluation Model

To prove that the maturity evaluation model constructed in this study has good applicability, and to understand and evaluate the current development status of Chinese film and television enterprises in the application of AI technology to help these enterprises identify their existing problems and propose targeted suggestions, this study selected several typical Chinese film and television enterprises for empirical research. The application maturity model for enterprise AI technology was tested, and a typical Chinese film and television Enterprise Q was taken as an example to present the fuzzy evaluation process.

4.4.1. Fuzzy Evaluation

The score matrix shown in

Table 11. The score matrix for Enterprise Q.

Second-Level Index	Third-Level Index	Comment Set
		V1	V2	V3	V4	V5
ISA	ISA1	0.0	0.0	0.0	0.1	0.9
	ISA2	0.0	0.0	0.1	0.8	0.1
	ISA3	0.0	0.0	0.1	0.8	0.1
ITA	ITA1	0.0	0.0	0.0	0.1	0.9
	ITA2	0.0	0.0	0.0	0.1	0.9
	ITA3	0.0	0.0	0.0	0.1	0.9
IBA	IBA1	0.0	0.1	0.8	0.1	0.0
	IBA2	0.0	0.1	0.2	0.7	0.0
IBB	IBB1	0.0	0.0	0.1	0.1	0.8
	IBB2	0.0	0.0	0.1	0.1	0.8
	IBB3	0.0	0.0	0.2	0.8	0.0
	IBB4	0.0	0.0	0.2	0.8	0.0
	IBB5	0.0	0.0	0.1	0.1	0.8
IBC	IBC1	0.0	0.0	0.1	0.1	0.8
	IBC2	0.0	0.0	0.1	0.8	0.1
IOA	IOA1	0.0	0.0	0.1	0.8	0.1
	IOA2	0.0	0.0	0.9	0.1	0.0
	IOA3	0.0	0.0	0.9	0.1	0.0
	IOA4	0.0	0.0	0.8	0.2	0.0
	IOA5	0.0	0.0	0.1	0.1	0.8
IOB	IOB1	0.0	0.0	0.1	0.1	0.8

was obtained after the evaluation of the third-level indexes, and the scoring results were normalized. In the first row of the matrix, 90% of the respondents thought that “intelligent strategic thinking” was at the leading level, 10% thought that the index was at the optimization level, 0% thought that the index was at the integration level, 0% thought that the index was at the specification level, and 0% thought that the index was at the planning level.

Taking the “intelligence strategy (ISA)” as an example, its fuzzy comprehensive evaluation matrix was as follows:

$$R_{11}=\left[\begin{array}{ccc}0& 0& 0\\ 0& 0& 0.1\\ 0& 0& 0.1\end{array} \begin{array}{c}0.1\\ 0.8\\ 0.8\end{array} \begin{array}{c}0.9\\ 0.1\\ 0.1\end{array}\right]$$

The weight matrix was as follows:

$$W_{11}=\left[\begin{array}{ccc}0.1136& 0.3193& 0.5672\end{array}\right]$$

The fuzzy comprehensive evaluation matrix for intelligence strategy was obtained:

$$\begin{gathered} B_{11}=W_{11}\cdot R_{11}=\left[\begin{array}{ccc}0.1136& 0.3193& 0.5672\end{array}\right]\cdot \left[\begin{array}{ccc}0& 0& 0\\ 0& 0& 0.1\\ 0& 0& 0.1\end{array} \begin{array}{c}0.1\\ 0.8\\ 0.8\end{array} \begin{array}{c}0.9\\ 0.1\\ 0.1\end{array}\right] \\ =\left[0.000\begin{array}{l} 0.0000 0.0886 0.7205 0.1909\end{array}\right] \end{gathered}$$

According to the above steps, the fuzzy comprehensive evaluation matrix was calculated successively, and the fuzzy evaluation matrix of the first-level index was constructed:

$$\begin{gathered} R_1=\left[\begin{array}{ l}0.0000& 0.0000\end{array} \begin{array}{ccc}0.0886& 0.7205& 0.1909\end{array}\right] \\ {R}_2=\left[\begin{array}{ l}0.0& 0.0\end{array} \begin{array}{ccc}0.0& 0.1& 0.9\end{array}\right] \\ {R}_3=\left[\begin{array}{ ll}0.0000& 0.1000& 0.4567\\ 0.0000& 0.0000& 0.1414\\ 0.0000& 0.0000& 0.1000\end{array} \begin{array}{c}0.4433\\ 0.3897\\ 0.2613\end{array} \begin{array}{c}0.0000\\ 0.4689\\ 0.6387\end{array}\right] \\ {R}_4=\left[\begin{array}{ ll}0.0000& 0.1000& 0.5824\\ 0.0000& 0.0000& 0.1000\end{array} \begin{array}{c}0.2636\\ 0.1000\end{array} \begin{array}{c}0.1540\\ 0.8000\end{array}\right] \end{gathered}$$

Taking “intelligent business operations (IB)” as an example, the fuzzy evaluation matrix was calculated.

The weight matrix was as follows:

$$W_3=\left[\begin{array}{ccc}0.1639& 0.6492& 0.1869\end{array}\right]$$

The comprehensive fuzzy evaluation results were as follows:

$$\begin{array}{ccc}{B}_3=& W_3\cdot R_3=\left[\begin{array}{ccc}0.1639& 0.6492& 0.1869\end{array}\right]\cdot & \left[\begin{array}{ccc}0.0000& 0.1000& 0.4567\\ 0.0000& 0.0000& 0.1414\\ 0.0000& 0.0000& 0.1000\end{array} \begin{array}{c}0.4433\\ 0.3897\\ 0.2613\end{array} \begin{array}{c}0.0000\\ 0.4689\\ 0.6387\end{array}\right]\\ =& \left[0.0000 0.0164 0.1853 0.3745 0.4238\right]\end{array}$$

The fuzzy evaluation matrix of other first-level indexes was further calculated using the above calculation method, and the fuzzy evaluation vectors of the above dimensions were further integrated to construct the comprehensive evaluation fuzzy matrix of the target layer:

$$R=\left[\begin{array}{ ll}0.0000& 0.0000& 0.0886\\ 0.0000& 0.0000& 0.1000\\ 0.0000& 0.0164& 0.1853\\ 0.0000& 0.0000& 0.4267\end{array} \begin{array}{c}0.7205\\ 0.1000\\ 0.3745\\ 0.2107\end{array} \begin{array}{c}0.1909\\ 0.8000\\ 0.4238\\ 0.3626\end{array}\right]$$

According to the matrix, based on the principle of maximum membership degree, for the dimension of intelligent business operations, the membership degree belonging to the “optimization level” in the evaluation results was the largest, and its value was 0.7205. Therefore, the maturity under this dimension was at the “leading level”. Similarly, the evaluation result was that the intelligence technology dimension was at the “leading level”, the evaluation result for the intelligent business operations dimension was at the “leading level”, and the evaluation result for the intelligent organizational management dimension was at the “integration level”.

Finally, the fuzzy comprehensive evaluation results were calculated, and the weight set was as follows:

$$W=\left[0.3568 0.0805 0.3617 0.2010\right]$$

The fuzzy comprehensive evaluation results of the application maturity of the AI technology of Enterprise Q can be obtained through the fuzzy operation of the weighting evaluation matrix and the fuzzy evaluation matrix of each first-level index:

$$\begin{array}{cc}B=W·R& \\ =\left[\begin{array}{cccc}0.3568& 0.0805& 0.3617& 0.2010\end{array}\right]\cdot & \left[\begin{array}{ ll}0.0000& 0.0000& 0.0886\\ 0.0000& 0.0000& 0.1000\\ 0.0000& 0.0164& 0.1853\\ 0.0000& 0.0000& 0.4267\end{array} \begin{array}{c}0.7205\\ 0.1000\\ 0.3745\\ 0.2107\end{array} \begin{array}{c}0.2909\\ 0.8000\\ 0.4238\\ 0.3626\end{array}\right]\\ =\left[0.0000 0.0059 0.1925 0.4429 0.3944\right]& \end{array}$$

Based on the above data, the application maturity level of the AI technology of Enterprise Q is at the “optimization level”.

4.4.2. Comprehensive Evaluation Results and Analysis

The score value corresponding to the comment set was set as S = {s₁, s₂, s₃, s₄, s₅} = {1,2,3,4,5}; that is, values of 1, 2, 3, 4, and 5 were assigned to the planning level, standard level, integration level, optimization level, and leading level, respectively. A summary of the evaluation results is shown in

Table 12. The summary of the evaluation results for Enterprise Q.

Target Layer	Score	First-Level Index	Score	Second-Level Index	Score
Application maturity of enterprise AI technology	4.3328	Intelligence strategy (IS)	4.1026	Intelligence strategy (ISA)	4.1026
		Intelligence technology (IT)	4.7000	Intelligence technology level (ITA)	4.7000
		Intelligent business operations (IB)	4.2057	Film and television project planning and development (IBA)	3.3433
				Film and television project production (IBB)	4.3276
				Film and television project marketing (IBC)	4.5387
		Intelligent organizational management (IO)	3.9359	Organizational management mode (IOA)	3.5716
				Corporate cultural conception (IOB)	4.7000

.

The average value of the overall level of application maturity of AI technology predicted by the experts according to their own experience and understanding of the situation of Enterprise Q was 4.1 points. The score calculated by the above evaluation model was 4.3328 points, which is relatively close to the expert’s estimated score of the overall AI technology application maturity of Enterprise Q, indicating that the evaluation model constructed in this study is highly accurate.

5. Conclusions

With the rapid development of information technology, enterprises have both opportunities and challenges in the process of using AI empowerment. This study constructed a comprehensive evaluation method system for enterprise AI technology application maturity that combines qualitative and quantitative methods and performs an empirical analysis. The research results showed that the evaluation model is highly accurate.

First, based on a literature analysis and multiple rounds of expert interviews, this study explored the composite dimensions of enterprise AI technology application maturity in detail, and identified four dimensions: intelligence strategy, intelligence technology, intelligent business operations, and intelligent organizational management. A maturity evaluation index system for enterprise AI technology consisting of 4 first-level indexes, 8 second-level indexes and 23 third-level indexes was constructed. Combined with industry experts’ opinions, the evaluation scale was modified and improved, and an exploratory factor analysis and confirmatory factor analysis were carried out on the scale. The reliability and validity test results of the scale showed that the scale developed in this study had high reliability and validity, and could comprehensively reflect the component dimensions of enterprise AI technology application maturity.

Second, the ANP was used for an index weight analysis. A judgment matrix was established under the advice of experts, and consistency tests and supermatrix calculations were carried out. After weighted and stabilized processing, the index weight was determined, and the evaluation system was quantified. Based on the ANP method, a fuzzy comprehensive evaluation method was introduced to construct a comprehensive evaluation model, evaluate the application maturity of enterprise AI technology, and classify the maturity level of enterprise AI technology applications to ensure that the evaluation results had quantifiable levels. The maturity data of film and television enterprises for various indexes were collected via questionnaires, the maturity of each evaluation dimension was calculated, the membership degree was transformed according to the experts’ opinions, and a fuzzy evaluation matrix was generated. Then, the fuzzy comprehensive evaluation results of enterprise AI technology application maturity were empirically analyzed through a combination of fuzzy operations and weights, and the comprehensive score of maturity was calculated. The application maturity of enterprise AI technology was comprehensively evaluated.

The research results showed that intelligence strategy was considered to be the greatest factor affecting the maturity of AI technology in enterprises. The most influential factor in technology was the research and development of intelligence technology and equipment. The most influential factor in business operations was smart shooting. The most influential factor in organizational management was the corporate culture. Furthermore, this study selected a number of typical Chinese film and television enterprises for empirical research, and verified that the enterprise AI technology application maturity evaluation model constructed in this study has good applicability and high evaluation accuracy. This study examined the application maturity of AI technology in enterprises in terms of enterprise strategy, technology, business operations, and organizational management. However, for enterprises with different attributes, this influence shows heterogeneity. Due to time limitations, this study did not analyze the heterogeneity of the characteristics of different enterprises. Future research could conduct more in-depth analyses of this aspect.

With the advent of digital intelligence, the application of AI to enable the development of enterprises has become an irreversible trend, which is a significant fact that will promote the intelligence transformation of enterprises and enhance their core competitiveness. Therefore, this paper makes the following suggestions:

(1) Enterprises should actively deploy intelligent strategies and vigorously promote the application of AI technology. Enterprises should attach importance to the value of AI technology, seize the opportunities brought by the AI revolution, and improve the application level of enterprise AI technology from the aspects of strategy, technology, business operations, and organizational management. AI can improve productivity by relying on “AI + big data” to mine massive amounts of data, assisting enterprises in scientific production and governance, making better and more accurate decisions, achieving cost reductions and efficiency, and increasing the enterprise’s value.

(2) We should vigorously promote AI technology to enable enterprises to integrate, build, and reconfigure internal and external resources. On the one hand, with the empowerment of AI technology, enterprises can better cope with the rapidly changing external environment, thus improving their innovation abilities and allowing them to maintain sustainable competitive advantages in a competitive environment. On the other hand, driven by an intelligent strategy, the enhancement of digital intelligence capabilities brought by the development of AI-enabled enterprises will help them integrate the new generation of information technology resources with other existing resources, and will help enterprises produce more advanced and intelligent products by strengthening their creative divergence abilities, decision support abilities, resource collaboration abilities, and value creation abilities. All enterprise products and business models should be innovated to effectively improve performance.