Smart Finance Transformation in ChatGPT Perspective: Logic, Framework, and Application Scenarios

In the era of artificial intelligence, enterprise financial accounting is undergoing digital transformation. Intelligent technology reshapes the financial process and improves operational efficiency, but it also brings challenges in technology, data, talent and organisation, etc. The response includes optimising the technical architecture, strengthening data management, cultivating professional talent and promoting cultural construction, and enterprises need to systematically promote these initiatives to realise intelligent upgrading of the financial function, improve operational efficiency, enhance the decision-making support capability, and better cope with changes in the market. Through comprehensive transformation, enterprises can take advantage of the competition [1-3].

Enterprise intelligent financial system refers to the use of advanced information technology and data analysis methods, combining financial management with intelligent technology to build an efficient and intelligent financial management system. In the era of digital intelligence, it is of great significance for group enterprises to build an intelligent financial system [4]. Specifically, it mainly includes the following four aspects, to improve the efficiency of financial management, to improve the level of risk management, to support the enterprise to make strategic decisions scientifically, and to promote the optimisation of the overall resource allocation of the enterprise [5].

To sum up, the intelligent financial system has significant advantages in improving the quality of decision-making, optimising resource allocation, and enhancing the ability of risk prevention and control. In the era of digital intelligence, the construction of intelligent financial system has become a key strategy for group enterprises to achieve sustainable development [6]. In the future, with the continuous progress of digital technology and application scenarios are increasingly rich, intelligent finance will bring more application value for group enterprises. Therefore, enterprises need to continue to learn and improve the application level of intelligent finance, so as to continuously improve the market competitiveness of enterprises and ensure that enterprises achieve long-term sound development [7].

In recent years, the rapid development of a new generation of information technology, such as big intelligence, cloud and object area, digital, intelligent accounting, financial reporting, management accounting, internal control and other aspects of the depth of the application, to promote the accounting industry into a high-quality development stage.Musleh Al-Sartawi, et al. outlined the positive role played by artificial energy technology in support of the enterprise, the investor decision-making and then to ensure the sustainable development of financial They also provide an in-depth analysis of the potential problems facing the practice of AI technology in the field of financial accounting [8]. Gotthardt, M et al. illustrate the potential value of automation and AI technology in empowering accounting and auditing, and establish a theoretical framework for analysing this automated intelligent technology in accounting and auditing practice [9]. Zhang, Y et al. systematically review the potential of Artificial Intelligence and Blockchain technology in the field of accounting development, pointing out the increased demand for accounting jobs in IT professions, which also reveals the future expectations and preferences of the accounting field in terms of the professional expertise of accounting students [10]. Parimi, S. S. R discusses how the introduction of machine learning technology into the SAP system can improve the compliance of financial reports and optimise them, especially in the financial results. The application of fraud detection contributes positively to the promotion of financial reporting compliance and quality optimisation in SPA systems [11]. Hajek, P. et al. developed a financial fraud detection model based on intelligent feature selection and classification machine learning algorithms and noted that among them the Bayesian Belief Networks (BBN)-based detection model demonstrated the optimal performance [12]. Artificial intelligence technology empowered financial accounting research, mainly with the financial report quality optimisation, financial work framework and financial fraud detection, and the depth of the research is insufficient, the need for further in-depth exploration of the depth of the practice of artificial intelligence technology in the field of financial accounting.

Based on the current domestic economic environment, domestic companies need to redefine their governance objectives, determine the focus of financial reform, promote the optimisation of the governance model, and ensure the improvement of the quality of financial management.Reid, L. C et al. examined the changes in audit reporting in the UK and the changes in the international audit reporting regulations, and pointed out that the new rules on audit reporting have effectively improved the quality of financial reporting, and that the cost of maintaining the audit report remains unchanged [13]. Dillard, J et al, in order to address how accounting accountability systems can contribute to democracy, systematically examined the relevant literature reports, pointing out that along with the maturation of accounting systems, the political and social implications of critical accounting research need to be further advanced and developed [14]. Grossi, G et al. reviewed research papers and journals related to accounting accountability systems and were informed that there are variations in accounting accountability systems in different public organisations due to differences in value orientation and behavioural logic in different public organisations [15]. Morales-Díaz, J et al. analysed how the reformed accounting model affects entity finances, and based on the results of the study it was analysed that the retail, hospitality and transport sectors were the most affected [16]. Furqan, A. C et al. analysed data related to 491 municipalities using cross-sectional regression methodology and showed that the quality of financial reporting and the quality of public services are highly correlated in the context of accrual accounting system, where audit recommendation tracking contributes positively to the quality of public services [17]. Cutting-edge research reveals that financial reforms involve systems, and rules, and analyses the effects of these reforms, contributing positively to the adaptation of stakeholders to reformed financial systems and financial rules.

In the face of the rise and development of financial intelligence, this paper proposes the logic of change of intelligent finance based on artificial intelligence technology, respectively on the two aspects of financial data processing and financial risk control. Extract the features of the clustering algorithm through feature engineering, capture the pattern and structure of the data, and use the clustering modeling algorithm to divide the financial data into different clusters, organize the intrinsic structure in the data, and combine with the data visualization technology to present the complex clustering results in an intuitive way, so as to realize the intelligent processing of financial big data. The fuzzy control function of constrained variable model and variable model is established, the learning degree of intelligent control algorithm is calculated by using statistical feature analysis, and the artificial intelligence model for the control of risk factors of enterprise financial informatization is established, and adaptive optimization search is carried out through the learning algorithm of artificial intelligence, and the optimization design of control algorithm is carried out by combining the least squares planning model and fitting algorithm. In this paper, the effectiveness and performance of intelligent financial big data processing technology and intelligent enterprise financial risk control technology are respectively experimented, and the financial data of Company A is used as the research sample to carry out intelligent financial change simulation experiments. At the end, the ChatGPT perspective is used as an entry point to present the application of this paper’s intelligent financial change technology and the safeguard measures for conducting financial construction.

2

Smart Financial Transformation in the ChatGPT Perspective

2.1

Overview of ChatGPT

ChatGPT is a natural language processing model driven by artificial intelligence technology, whose core lies in processing linguistic and textual data, which can learn patterns and statistical laws through the pre-training process, generate logical answers, and interact with the context of the chat, forming a chat scenario that is almost indistinguishable from that of a real human being for communication.ChatGPT is characterized by three capabilities, including natural language processing, reasoning and adaptive learning ChatGPT is characterized by three capabilities: natural language processing, reasoning and adaptive learning, and multimodal processing.

ChatGPT is a natural language processing technology that is based on artificial neural networks trained and learned through big data, so that computers can understand, process, and generate natural language like humans.The logical steps are mainly divided into three steps: supervised fine-tuning model, training return model with PPO model, and fine-tuning SFT model.

2.2

Smart Finance Framework Logic

The essence of smart finance is a new management model, which is a reconstruction of the traditional financial management model. The basic framework of intelligent finance contains two parts, broad and narrow, at the broad level, intelligent finance is an intelligent ecosystem, intelligent finance involves the main body in addition to the application of enterprises and institutions, but also includes government management, industry organizations, supply chain and other related main. At the narrow level, the smart finance framework mainly involves the main body of the application of smart finance.

Intelligent finance is a man-machine symbiosis, a collaborative evolution of accounting management activities. Its underlying technical architecture mainly includes four parts: intelligent perception, network, data, and intelligent engine.The main function of the intelligent perception system is to collect data about business activities and process it in a structured way. The main technologies used include barcodes, radio frequency identification, OCR text recognition, and others.Network system function is mainly to maintain the transmission and sharing of data. The realization of this function is mainly through mobile Internet and Internet of Things technologies.The data system’s primary purpose is to support the intelligent financial application layer, which includes meta-database data and data generated by the enterprise’s financial and tax business processing.The intelligent engine system improves the operational efficiency of the intelligent financial system through multimodal means.

The left side is the source of information input for the intelligent financial system, including the data of business activities of the enterprise and external environment subjects such as upstream and downstream enterprises, government authorities, investors and public big data resources in the market, and the input of information is completed through human-machine collaboration. The underlying intelligent engine provides real-time and dynamic correlation data in a targeted manner that meets the needs of different business managers in information output.

The middle information processing contains three financial platforms such as intelligent accounting, intelligent management, and intelligent strategy. Intelligent accounting financial platform is the primary stage of the development of intelligent finance, mainly relying on RPA, OCR and other technologies to intelligently complete the financial accounting work, for the large number of repetitive and standardized financial work that exists in the enterprise, ChatGPT can use its natural language processing technology and adaptive learning mechanism to effectively deal with it, and the intelligent management financial platform is the intermediate stage of the development of intelligent finance, which On the basis of intelligent accounting, it gradually shifts to the integration of industry and finance, big data financial analysis, intelligent financial analysis visualization and other management activities. Intelligent strategic financial platform is the mature stage of the development of intelligent finance, which focuses on the play of intelligent decision-making functions on the basis of intelligent accounting and intelligent management, including revenue forecasting and decision-making, cost forecasting and decision-making, investment and financing forecasting and decision-making, financial risk forecasting and decision-making, and will be computer-based. Aspects will be completed through computer-based human-computer integration.

2.3

Intelligent Finance Application Scenarios and Countermeasures

With the rise of ChatGPT, digital intellectualization technology is increasingly being penetrated into various industrial fields, and the financial management of modern enterprises has gradually entered the era of intelligent change. This section will address the two application scenarios of financial data processing and financial risk control, and propose corresponding financial management countermeasures on the basis of artificial intelligence technology respectively, for the sake of going forward, conveniently following up the functional docking with complex artificial intelligence models such as ChatGPT, and proposing safeguards for intelligent financial change based on ChatGPT’s perspective, to enhance the efficiency and accuracy of financial processing, and to further in the financial change to Promote the development of artificial intelligence applications.

1)

Financial data processing.

In the financial data processing scenario of modern enterprises, the general framework of the financial big data intelligent analysis system is constructed, and clustering algorithms, feature extraction and other technologies are utilized to intelligently analyze and process corporate financial big data.

2)

Financial risk control.

Propose the financial risk factor control algorithm based on artificial intelligence, apply it to the financial risk control scenario of modern enterprises, and realize the quantitative assessment and automatic prediction of the risk factors of enterprise financial informatization.

3

Smart financial change technologies

In the previous chapter, this paper introduces the current emerging artificial intelligence ChatGPT and puts forward the framework logic of intelligent finance, and proposes corresponding intelligent finance strategies and methods from two application scenarios of financial data processing and financial risk control. At the end of this paper, it will also provide safeguards for its smart finance construction effect from the ChatGPT perspective and promote functional docking with ChatGPT.

This chapter will focus on how to propose intelligent financial big data processing technology and intelligent enterprise financial risk control technology based on artificial intelligence technology in the enterprise financial management scenarios of financial data processing and financial risk control, respectively.

3.1

Intelligent financial big data processing technology

3.1.1

Feature engineering extraction

Feature engineering is a very important aspect of data analytics and machine learning, which mainly involves selecting, constructing or transforming features from raw data to improve the effectiveness and performance of subsequent clustering algorithms [18]. The main goal of feature engineering is to reduce the dimensionality of data, improve data separability, and eliminate redundant features so as to better capture the essential characteristics of financial data. The specific steps are as follows.

1)

Selection of extracted features

In the process of feature selection, the variance threshold is used to select the most informative and relevant features, which are retained and irrelevant or redundant features are removed. The calculation process of specific variance threshold is shown in formula (1): (1) $V (x) = \frac{1}{n} \sum_{i = 1}^{n} {(x_{i} - u)}^{2}$

Where V(x) is the variance of feature x, n is the number of samples, x_i is the eigenvalue of each sample, and u is the mean value of feature x. The larger the variance value is, the more dispersed the distribution of the data is, and the smaller the variance value is, the more centralized the distribution of the data is. When the variance value is lower than a predetermined threshold, the feature is deleted.

2)

Construct transformed features

Create new features to capture patterns in the data, feature engineering combines different features or calculates statistical metrics to create new features. Use interaction features to create interaction terms between 2 or more features to capture the correlation between those features. Specific variance thresholds are calculated as shown in equation (2): (2) $X (i) = x_{1} \cdot x_{2}$

Where X(i) is the new feature interaction information, x₁ and x₂ are the two original features or variables respectively, which can be any feature in the dataset. Simple features are multiplied together during the computation of interaction features to create a new feature which captures the relationship between the original features.

3)

PCA feature dimensionality reduction

Dimensionality reduction is one of the key steps in feature engineering used to reduce the dimensionality of the data while retaining as much information as possible. High dimensional data may increase computational complexity, reduce model performance and suffer from dimensionality catastrophe problems. The main adaptation of Principal Component Analysis (PCA) linear dimensionality reduction technique is to project the original high dimensional features to a new low dimensional subspace by linear transformation [19]. The main goal of PCA is to maximize the covariance of the projected data while retaining as much information as possible. The steps of PCA are as follows.

Calculate the covariance matrix C of the original features as shown in equation (3): (3) $C = \frac{1}{m} x^{t} \cdot x$

where m is the number of samples, x is the dataset for each feature, and x′ is the transpose of x. The eigenvectors corresponding to the first k eigenvalues are selected, where k is the dimension that is desired to be preserved, and the data are projected into the subspace consisting of the selected eigenvectors, and the original data X are projected into the subspace consisting of the selected k eigenvectors to obtain the new feature matrix Y. The computation is shown in Eq. (4): (4) $Y = X \cdot V_{K}$

Where Y is the dimensionality reduced dataset, X is the original data, and V_K is the matrix consisting of the selected feature vectors. In conclusion, feature engineering needs to be based on data preprocessing to further optimize the data.

3.1.2

Analysis of clustering modeling algorithms

Cluster modeling algorithms are used to group financial data into groups (clusters) with similar characteristics, with the goal of identifying patterns, trends, and clusters in the data for further intelligent analysis and decision support. The clustering modeling algorithm selects appropriate clustering algorithms (including K-means, hierarchical clustering algorithms) and sets relevant parameters in order to group data into meaningful clusters. The clustering modeling algorithm specifically consists of the following two computational components.

1)

K-mean clustering [20].

The specific subset of the cluster modeling algorithm is an iterative clustering algorithm that divides the data points into K clusters, where K is the number of pre-specified clusters and the goal is to minimize the sum of the squares of the distances between the data points within each cluster and the centroid (center of mass) of the cluster in which they are located. The specific distance calculation is shown in equation (5): (5) $d (x, y) = {\sqrt{\sum_{i = 1}^{n} (x_{i} - y_{i})}}^{2}$

where d(x,y) is the distance between two points and x_iy_i is the eigenvalue of the two points. K Mean clustering has obvious assumptions about the data distribution, i.e., each cluster is convex and has equal variance, and it is sensitive not only to outliers but also to the choice of the initial center of mass.

2)

Hierarchical clustering

Hierarchical clustering is a bottom-up or top-down clustering method in clustering modeling algorithms, which creates a tree structure (dendrogram or dendrogram) for the hierarchical structure of the data, and the tree can be pruned to obtain different numbers of clusters as needed.

3.1.3

Data visualization

In clustering algorithm-based intelligent analytics processing techniques for financial big data, data visualization is the presentation of data in the form of graphs or images to help users better understand the data, spot patterns and trends, convey information, and support decision-making.

Clustering results visualization is the first step in data visualization and shows the overall resultant composition of the data by labeling or coloring data points by clusters. Feature Importance Visualization distinguishes the formation and characteristics of the data clusters after the clustering results are visualized, and the user can see how the data points are distributed into different clusters. The feature importance score of the tree is mainly calculated using the feature split gon, and the calculation is shown in Equation (6): (6) $S c = \sum_{i = 1}^{m} I (f) \cdot \frac{G_{l} \cdot G_{r}}{G_{t}}$

where Sc is the splitting contribution of feature f in tree t, n_t is the number of nodes in tree t, I(f) is an indicator function that is 1 if node i is split using feature f and 0 otherwise. G_l is the impurity of the left subtree, G_r is the impurity of the right subtree, and G_t is the total impurity of tree t. Intra-cluster data distribution visualization further deepens the understanding of the data structure within each cluster, by plotting histograms or density plots of clusters, the user can identify the shape and characteristics of the data distribution within the cluster, and also measure the degree of dispersion of the data points with respect to the mean based on the calculation of standard deviation of the data distribution within the clusters, which is calculated as shown in Eq. (7): (7) $S D = \sqrt{\frac{1}{n - 1}} \sum_{n - 1}^{n} {(x_{i} - x_{\min})}^{2}$

Where SD is the standard deviation, n is the number of data points in the cluster, x_min is the data mean, and x_i is the value of each data point.

Decision support visualization is the ultimate goal of the entire process, integrating clustering results, feature importance, and other business information to help users better understand the information behind the data during the decision-making process, as well as communicating the clustering results to the decision maker, allowing the user to further explore the data and make practical decisions. The completeness and coherence of the data visualization process are ensured by the linkage between these steps, which allows users to fully comprehend the clustering structure of financial data and use it for practical business decisions.

3.1.4

Experiments and analysis of results

In order to verify the effectiveness of the intelligent financial big data processing technology based on clustering algorithm proposed in this paper, a comparison experiment with the traditional intelligent machine learning algorithm SVM is carried out in this section. In the experiment, the parameter settings of each model are kept consistent, and the loss function change trend and classification accuracy are plotted and obtained specifically as shown in Figure 1. Comparison of the loss function change graph can be found, when epochs = 30, this paper’s financial big data processing method’s train_loss value and val_loss value were 0.026, 0.365, respectively, while the SVM model’s train_loss value and val_loss value were 0.061, 0.197. Obviously, this paper’s design of financial big data processing method drops to a stable value of 0.061, 0.197. Data processing method designed in this paper decreases to a stable value a little slower but relative to the SVM model finally trained to get a lower value of the loss function, that is to say, indicating that the classification effect is the best. By comparing the classification accuracy graphs, it can also be found that when epochs=30, the values of train_acc and val_acc of this paper’s financial big data processing method reach 0.99 and 0.915, respectively, while the values of train_acc and val_acc of the SVM model are 0.981 and 0.911. The classification accuracy of this paper’s financial big data processing method is higher than that of the comparative SVM model. Accuracy is higher than the comparison of SVM, a traditional intelligent machine learning method.

3.2

Financial Risk Control Techniques for Intelligent Enterprises

3.2.1

Constrained variable model

In order to realize risk factor control, it is necessary to first construct a constrained variable model of risk factor control of enterprise financial informationization, and adopt the constrained feature analysis method to make quantitative decisions on risk factor control. Assuming that there are n samples in the quantitative set of the level of risk factor control of enterprise financial informationization, in which the sample characteristic distribution of risk factor assessment of enterprise financial informationization is x_i,i=1,2,⋯,n, which indicates the number of items of the associated data for risk factor control of enterprise financial informationization, in the model of information management, the set of fuzzy quantitative regression distributions of risk factor control of enterprise financial informationization can be expressed as follows: (8) $M_{D a p} = M_{D B} + Q \sum_{i = 1}^{n} \sum_{j = 1}^{l_{i} - 1} m$

When the number of subgroups Q=m, the fuzzy association rule scheduling method is used for the decision-making of risk factors of enterprise financial informatization, and the target decision vector is described as x_i=(x₁,x₂,⋯,x_i). The Lyapunov general function for the control of risk factors of enterprise financial informatization under the decision-making of the association rule is satisfied [21]: (9) $0 \leq f^{r} (y (- σ) V f (y (- σ))]$

Based on the above analysis using statistical feature analysis method to construct a constraint variable model for the control of risk factors of corporate financial information technology, assuming that the constraint variable model is: (10) $m i n_{0 < α_{i} < c} W = \frac{1}{2} \sum_{i, j = 1}^{l} y_{i} y_{j} α_{i} α_{j} K (x_{i}, x_{j}) - \sum_{i = 1}^{l} α_{i} + b M_{D u p}$

In the formula, (x_i,x_j) represents the enterprise financial information technology risk factor control level characteristics of the sample, b for the enterprise financial information technology risk factor control level characteristics of the classification of attributes, the model as a decision-making objective function, need to be added to the fuzzy control function, in order to improve the control effect of the method in this paper.

3.2.2

Fuzzy control functions for variable models

Establishing the big data analysis model of risk factor control of enterprise financial informationization, adopting the association rule mining method to carry out the statistical characterization of risk factor control of enterprise financial informationization, and obtaining the formula of balanced metric factor D_m for quantitative assessment of risk factor control of enterprise financial informationization as follows: (11) $D_{m} = \frac{d_{e} + d_{b} + \sum_{i = 1}^{n - 1} | d_{i} - \frac{\sum_{i - 1}^{n - 1} d_{i}}{n - 1} |}{d_{e} + d_{b} + (n - 1) \frac{\sum_{i = 1}^{n - 1} d_{i}}{n - 1}}$

Where, d. Is the distribution distance of the feature points of the level of control of risk factors of enterprise financial informationization in the space of searching for excellence, and d_b is the average adaptation degree of the control of risk factors of enterprise financial informationization. Using the descriptive statistical analysis method, the fuzzy association rule feature quantity description of enterprise financial informatization risk factor control is obtained as: (12) $T = \frac{1}{t p}, i, j \in {0, 1, \dots, v + 1}$

Analyzing the explanatory variables of informatization management on the control of risk factors of corporate finance informatization, the average confidence level of informatization risk factor assessment is obtained: (13) $k = I n t (\frac{n \bar{Q}}{1 - \bar{Q}}) + 1$

Where, $\bar{Q}$ is the endogenous control variable of corporate financial management, under the Probitv multiple regression analysis model, the least squares fitting method is used to obtain the factor analysis results of the control of risk factors of corporate financial informatization are described as follows: (14) $p (e_{k} | v_{k}) \sim t_{({\tilde{v}}_{k} + d_{s})} ({\tilde{u}}_{e | s, k}, {\sum^{˜}}_{e | s k})$

The statistical mean of the control of risk factors of corporate financial informatization is: (15) $\tilde{y} (t) = \frac{k T}{D_{m}}$

The fuzzy control function that describes the risk factors of corporate financial informatization using factor analysis and deep learning methods under continuous bounded conditions is [22]: (16) $V_{0} (k) = {\begin{matrix} γ [1] γ^{T} (1) & k = 1 \\ \frac{[ρ V_{0} (k - 1) + γ k) γ^{T} (k)]}{1 + ρ} & k > 1 \end{matrix}$

In the above equation, γ(k) and $\hat{x} (k | k - 1)$ are the detection statistics of risk factor control of enterprise financial informatization, respectively. Combining the fuzzy control function for risk factor control can improve the confidence level of risk factor control of enterprise financial informatization.

3.2.3

Intelligent Control Algorithm for Financial Informatization Risk Factors

1)

Learning degree selection for intelligent control

On the basis of the above construction of the constraint variable model and the control feature quantity analysis of risk factor modeling by combining the correlation between the corporate finance equity market and the corporate finance market, the optimization design of the control algorithm is carried out. The finite set for constructing the quantitative assessment of the control level of risk factors of corporate finance informatization is: (17) $f_{\lg - M} (z) = (f_{\lg} (z) h_{x} * f_{\lg} (z) h_{y} * f_{\lg} (z))$

In the above equation, f_1g(z) represents a set of regression analysis values of financial information system. The statistics are obtained by analyzing the risk factor control model of financial informatization under informatization management conditions under relevance constraints, cash flow level, financial redundancy conditions, and combining financial policy and fiscal relevance factors: (18) $C_{x x} (j τ) = \sum_{r = 1}^{t} \sum_{q = 1}^{k_{2}} {‖ W_{i}^{T} x_{i r} - W_{i}^{T} x_{i r q} ‖}^{2} B_{i r q}$

Where x_i denotes the principal component feature component of enterprise financial informatization risk factor control, x_inq is the fuzzy kernel of enterprise financial informatization risk, B_inq is the fuzzy state feature quantity of enterprise financial informatization risk factor control, and W_i is the full sample regression coefficient. Establish the big data analysis model of enterprise financial informatization risk factor control, and use the association rule mining method to carry out the statistical feature analysis of enterprise financial informatization risk factor control, and get the optimization function noted as: (19) $f (x) = \frac{1}{2} + C \sum_{i = 1}^{n} (ξ_{i} + ξ_{i}^{*}), ξ_{i}, ξ_{i}^{*} \geq 0, i = 1, 2, \dots, n; C > 0$

Where, ξ_i and $ξ_{i}^{*}$ are the detection statistics, respectively, combined with the segmented sample regression analysis method to realize the control of risk factors of enterprise financial informatization, and get the expression of learning degree calculation as: (20) $W_{x} = β K_{p o l y} + (1 - β) K_{r b f}, β \in (0, 1)$

Where K_poly denotes the association rule term. In summary, we construct the statistical feature analysis model of risk factor control of enterprise financial information technology, and carry out the optimization design of risk factor control algorithm.

2)

Optimization of control algorithm

The artificial intelligence control method is adopted to carry out the risk factor control of enterprise financial informatization, and the artificial intelligence learning process of risk factor control is obtained as: (21) $X_{i} = (α_{c}^{[1]}, α_{c}^{[2]}, α_{c}^{[3]}, \dots)$

The learning algorithm of artificial intelligence is used to carry out adaptive optimization search for the control of risk factors of enterprise financial informationization, and combined with the least squares planning model, the expression of constraint rules for the control of risk factors of enterprise financial informationization is obtained as follows: (22) $\min (f) = \sum_{i = 1}^{n} \sum_{j = 1}^{n} C_{i j} X_{i j}$ (23) ${\begin{array}{l} \sum_{j = 1}^{m} X_{i j} = a_{i}, i = 1, 2 \dots m \\ \sum_{i = 1}^{m} X_{i j} = b_{i}, j = 1, 2 \dots n \\ X_{i j} \geq 0, i = 1, 2 \dots m, j = 1, 2 \dots n \end{array}$

Adaptive training of the control level of risk factors of enterprise financial informationization is carried out by using the extreme learning method, and the adaptive learning model for the control of risk factors of enterprise financial informationization under artificial intelligence learning is as follows: (24) $G_{i} = \sum_{j} α_{j} y_{i} y_{j} K (x_{i}, x_{j}) + y_{i} b - 1$

Through the above steps to realize the adaptive control of the risk factors of enterprise financial information technology, the expression of the optimized control output is: (25) $Q' = k \frac{G_{i} X_{i}}{M_{D u p}}$

3.2.4

Experiments and analysis of results

In this section, the SPEA2 algorithm will be selected as the experimental comparison object to verify the performance of enterprise financial information technology risk control in this paper. Through the F-test method, the risk control convergence curve of the intelligent enterprise financial risk control technology proposed in this paper and the SPEA2 algorithm is obtained as shown in Figure 2. Analyzing the information in the figure, it can be seen that with the increase in the number of iterations, the convergence distance of this paper’s enterprise financial risk control technology is always lower than that of the SPEA2 algorithm, and the difference in convergence distance between the two can reach a maximum of 0.22. This paper’s enterprise financial risk control technology can effectively control risk factors in enterprise financial information systems, and the convergence is good.

4

Company A smart financial change simulation experiment

4.1

Overview of Company A

Company A is a major agent of kitchen and sanitary household in Hangzhou, China, founded in 2007, and now has branches in Shanghai, Jinan, Shanxi and Shandong, with more than 1,000 employees, and its business scope has been steadily expanding from sanitary and bath category to high-end cabinet and electric appliances, smart home and live water plumbing system and many other categories.

Since January 2023, Company A has been exploring the construction of smart finance. This paper will take the financial data of Company A as the research sample in this chapter, assume that Company A fully applies the intelligent financial big data processing technology and financial risk control technology proposed in this paper in January 2024, carry out the simulation experiment of intelligent financial change, and presuppose the relevant financial data changes after it carries out the practice of intelligent financial change.

4.2

Simulation and analysis of financial data

4.2.1

Efficiency of financial operations

According to the financial operation data of Company A in 2023, compare it with the financial operation simulation data in 2024, use the relevant financial data to calculate the ratio relationship between the total assets over the administrative management personnel and the total assets over the financial personnel, and analyze the impact of Company A’s introduction of the intelligent financial big data processing technology and financial risk control technology proposed in this paper in the intelligent financial change simulation experiment by means of the magnitude change. The impact of corporate financial operation efficiency, the specific data situation is shown in Table 1. The higher the ratio of total assets to total administrative personnel and total assets to total financial personnel, the more assets each manager and financial personnel is responsible for on average, reflecting the department and personnel with higher efficiency, and vice versa for the same reason. As can be seen from the table, the ratio of “total assets/administrative staff” to “total assets/finance staff” is 6.99 and 10.30 in 2023, and increases to 11.12 and 16.54 in 2024, respectively. The efficiency of the managers in the finance and administration departments has significantly improved. In terms of the growth rate of each growth rate, the growth rate of each growth rate in 2023 shows positive value, the growth rate of administrative management personnel, the growth rate of management expenses, the growth rate of financial personnel is 28.49%, 10.06%, 6.88%, respectively, showing different degrees of growth, while the growth rate of total assets is only 3.18%. In the smart financial change simulation experiment in 2024, Company A’s in the number of management personnel and the cost of a significant downward trend in the cost of administrative personnel growth rate, management fee growth rate, financial personnel growth rate of -20.98%, -21.02%, -21.61%, respectively, while the growth rate of total assets is as high as 25.82%. Obviously, in the intelligent financial change simulation experiment, the total asset growth rate of company A grows rapidly, and the growth rate of administrative costs and management personnel decreases significantly, and under the support of the intelligent financial change technology proposed in this paper, the financial operational efficiency of company A has a remarkable effect.

Table 1.

Financial operation efficiency

Index	2023	2024
Administrator	348	275
Management fee (ten thousand yuan)	28.17	22.25
Financial officer	236	185
Total assets (ten thousand yuan)	2431.49	3059.36
Total assets/administrators	6.99	11.12
Total assets/financial personnel	10.30	16.54
Rate of growth of administrator	28.49%	-20.98%
Management rate growth rate	10.06%	-21.02%
Rate of growth of financial personnel	6.88%	-21.61%
Total assets growth rate	3.18%	25.82%

4.2.2

Change in financial performance risk

In the comprehensive evaluation analysis of the impact of adopting the smart financial change technology proposed in this paper on financial performance, the financial risk level of the enterprise is divided into six levels for better judgment of the enterprise risk level, as shown in Table 2.

Table 2.

Risk condition

Integrated rating	Risk condition
W≥90	Low risk
75≤W<90	Lower risk
60≤W<75	Medium risk
45≤W<60	Higher risk
W<45	High risk

The simulation experiment data of financial performance risk change of Company A in 2024 is shown in Table 1, and the risk evaluation value of financial performance of Company A in 2023 is also listed in the table as a comparison. It can be seen that the risk assessment value of all financial performance indicators of Company A in 2024 is higher than that in 2023, of which the difference in the risk assessment value of operating income growth rate is 22.6, the largest difference. The second is the current asset turnover ratio and inventory turnover ratio, and the difference in risk assessment value reaches 13.6 and 12.9. Moreover, in the simulation experiment of applying the intelligent financial change technology proposed in this paper, the risk assessment value of the three financial performance indicators of Company A, namely, the gearing ratio, the net sales interest rate, and the growth rate of operating income, grows to 93.9, 93.5, and 93.4, which is greater than 90, and is at the low-risk status The risk rating of the remaining financial performance indicators increased to 93.9, 93.5 and 93.4 for the gearing ratio, net sales margin and operating income growth rate. The remaining financial performance indicators are all within the range of [75,90], which is a low risk level. In contrast, in 2023, Company A’s risk evaluation value of current asset turnover, accounts receivable turnover, inventory turnover, operating income growth rate and other risk performance indicators is lower than 75, which is at a medium-risk level. In terms of the overall comprehensive evaluation, the simulation result of the comprehensive evaluation value of financial performance risk of Company A in 2024 is 87.4375 higher than that of 2023 by 8.875.

4.2.3

Changes in overall enterprise risk

In the previous section, this study explored and analyzed the change in financial performance risk of Company A in 2024 in the smart financial change simulation experiment using 2023 as a comparison. In this section, the overall corporate risk level and each risk factor associated with implementing smart finance in Company A in the smart finance change simulation experiment are assessed as a whole. Each risk factor is specifically shown in Table 3.

Table 3.

Financial risk factor

Overall risk	Primary risk	Risk factor	Title
Overall financial risk	Strategic risk	The accuracy of intelligent financial positioning and understanding	A1
		The rationality of the design of intelligent financial construction road diameter	A2
		The implementation of the system and clothing provided by third-party suppliers is appropriate	A3
	Process risk	Rationality of business flow design	B1
		Effectiveness of process supervision	B2
		Business flow process standardization	B3
		Process execution and specification	B4
	Organizational risk	The rationality of the organization system	C1
		The rationality of organizational architecture	C2
		Run rule acceptance	C3
		The effectiveness of organizational information communication	C4
	Technical risk	Technical effectiveness	D1
	Technical risk	Information system security	D2
	Management risk	The degree of position of operation management	E1
	Management risk	The degree of position of organization operation management	E2
	Personnel risk	Dependence on accounting robots	F1
		Employee sentiment	F2
		Employee expertise	F3

The assessment was conducted using a quantitative evaluation methodology that assigns a value of [0,10] to “low risk”, [10,20] to “medium-low risk”, [20, 30] to “medium risk”, [30,40] to “medium-high risk”, and [40,50] to “high risk” is assigned [40,50]. In the smart financial change simulation experiment. The risk score and risk level of each risk factor is shown in Table 4. Overall, in the smart financial change simulation experiment, the overall risk level of Company A after the adoption of this paper’s smart financial change technology is in the “medium-low risk” level, and the overall risk assessment value is 16.05. Among the six types of level 1 risks, the risk level of strategic risk, technology risk, management risk, personnel risk is in the “medium-low risk” level. Among the six categories of Tier 1 risks, the risk ratings of strategic risk, technology risk, management risk and personnel risk are at the “medium-low risk” level, with scores of 19.07, 19.69, 15.23 and 14.18, while the risk ratings of process risk (9.54) and organizational risk (8.2) are at the “low risk” level. Because the higher the risk level, the more likely to have a greater impact, so in the simulation experiment, the risk of the implementation of smart finance in Company A mainly comes from the strategic risk, technology risk, management risk, personnel risk, of which the specific risk factors need to pay special attention to the “information system security” and “effectiveness of technical means” factors, which are the most important factors. Among the specific risk factors, special attention should be paid to the factors of “information system security” and “effectiveness of technical means”, which are the factors with the highest risk score among all the factors.

Table 4.

Risk score

Overall risk	Overall risk score	Overall risk level	Primary risk	Risk score	Risk level	Risk factor	Risk score	Risk sort
Overall financial risk	16.05	Medium low risk	Strategic risk	19.07	Medium low risk	A1	7.49	6
						A2	1.58	15
						A3	10	5
			Process risk	9.54	Low risk	B1	2.88	11
						B2	0.35	18
						B3	4.57	8
						B4	1.74	13
			Organizational risk	8.2	Low risk	C1	1.4	16
						C2	1.62	14
						C3	4.02	10
						C4	1.16	17
			Technical risk	19.69	Medium low risk	D1	9.5	2
			Technical risk	19.69	Medium low risk	D2	10.19	1
			Management risk	15.23	Medium low risk	E1	8.9	3
			Management risk	15.23	Medium low risk	E2	6.33	7
			Personnel risk	14.18	Medium low risk	F1	2.15	12
						F2	4.2	9
						F3	7.83	4

5

Safeguards for building smart finances in the context of ChatGPT

In the previous chapter, this study used the financial data of Company A as a research sample to conduct a simulation experiment of intelligent financial change. In the simulation experiment analysis, it can be clearly seen that the application of the intelligent financial change proposed in this paper in the financial operation efficiency and financial performance risk, the overall risk of good simulation results. In reality, if enterprises want to formalize smart financial change and smart financial construction, one of the important issues is how to guarantee the progress and effectiveness of the construction.

In the beginning of this study, the association between ChatGPT and smart financial change was proposed, and the wide application of ChatGPT in the financial field will play a positive role in the timeliness of financial data collection, standardization of financial advice, automation of financial decision-making, and intelligent enterprise management. This chapter will continue to use the ChatGPT perspective as an entry point to propose measures for intelligent financial construction safeguards.

1)

Promote the development of accounting profession

With the organic integration of ChatGPT technology and intelligent finance, the automation and intelligence of financial work will be greatly improved, standardized and repetitive financial accounting and financial analysis positions will be replaced by intelligent technology, and accountants need to change from accounting to management. Enterprises need to expand their accounting organizational structure and strengthen the professional skills training of financial personnel. Colleges and universities need to carry out the transformation of intelligent financial personnel training, integrate big data and intelligent courses into the curriculum system, and cultivate digital intelligent financial and tax talents adapted to the era of digital economy.

2)

Enhance risk response capability

Introducing ChatGPT technology and related intelligent technology into corporate finance and carrying out intelligent financial change practices can indeed change many shortcomings and improve overall efficiency, but this is after all a bold attempt and change, which will give rise to many uncertainties affecting the achievement of the goal, which requires colleges and universities to enhance their risk response capabilities. Enterprises in addition to organizing relevant learning and training, but also with the financial staff should be frequent and close communication, understand the work of the worries, and actively guide employees to quickly integrate into the new working environment, in some cases can take certain incentives for employees. In the optimization of the implementation process can not be rushed, you can first choose the core module for optimization, successful operation of the gradual progress.

3)

Provide technical support

Whether it is the intelligent financial change technology proposed in this paper, or ChatGPT technology, how to integrate these technologies with each other and better applied in practice, which requires a certain degree of technical guarantee before implementation. With the changes in business scenarios and technology applications, it is also necessary to continuously improve the application of technology to meet the needs of cost-smart financial optimization, and to establish a corresponding technical guarantee system to promote the integration and development of technology.

6

Conclusion

This paper focuses on the two aspects of financial data processing and financial risk control, and proposes intelligent financial big data processing technology and intelligent enterprise financial risk control technology respectively to realize the intelligentization of enterprise finance.

Focusing on the three aspects of financial operation efficiency, financial performance risk change and overall financial risk change, this study takes the financial data of Company A as the research sample and carries out the intelligent financial change simulation experiment. In the indicators of financial operational efficiency, the ratio of “total assets/administrative management personnel” and “total assets/financial personnel” of Company A is 11.12 and 16.54, respectively, showing a significant decline in the number of management personnel and the cost of expenses. In terms of the number of management personnel and cost of expenses, there is a significant downward trend, while the growth rate of total assets is as high as 25.82%, which improves the efficiency of financial operation obviously. In the financial performance risk change, A company in 2024 the financial performance risk indicators score are higher than in 2023, of which the asset-liability ratio, net sales interest rate, operating income growth rate of the three financial performance indicators of the risk assessment value is greater than 90, at the level of low-risk situation, the rest of the indicators are in the [75,90] range, at the level of lower risk. In the simulation experiment, the overall risk evaluation value of the enterprise after the smart financial change of Formula A is 16.05, and all risk evaluation indicators are at the level of “medium-low risk” or “low risk”.

Funding:

1) Shaanxi Province’s “14th Five-Year Plan” Education Science Plan 2023 Project: “Research on the New Business Talent Training Model of Application-oriented Undergraduate Colleges in Shaanxi Province under the Background of Digital Economy” (Project No.: SGH23Q0359)

2) Shaanxi Provincial Social Science Foundation Project: “Research on the Integration of Industry and Education and Collaborative Education Ecosystem in Colleges and Universities in Shaanxi Province under the “Four Chains” Collaborative Mechanism (Project No.: 2024P048)

3) Shaanxi Province’s “14th Five-Year Plan” Education Science Plan 2024 Project: “Research on the Practical Path of Financial Accounting Talent Training and Empowering New Quality Productivity from the Perspective of “Five Chains” Integration (Project No.: SGH24Y2586)

4) Shaanxi Province’s “14th Five-Year Plan” Education Science Plan 2024 Project: Research on the Training Model of Interdisciplinary Integrated and Compound Financial Accounting Talents in the Context of Artificial Intelligence (Project No.: SGH24Y2572).

Sprache:: Englisch

Zeitrahmen der Veröffentlichung:: 1 Hefte pro Jahr
Fachgebiete der Zeitschrift:: Biologie, Biologie, andere, Mathematik, Angewandte Mathematik, Mathematik, Allgemeines, Physik, Physik, andere

Zeitschrift RSS Feed

Smart Finance Transformation in ChatGPT Perspective: Logic, Framework, and Application Scenarios

Xiang Wu

Online veröffentlicht: 17. März 2025

Eingereicht: 13. Okt. 2024

Akzeptiert: 10. Feb. 2025

DOI: https://doi.org/10.2478/amns-2025-0270

SchlüsselwörterChatGPT, Clustering Algorithm, Fuzzy Control Function, Financial Risk Control Algorithm, Intelligent Finance

© 2025 Xiang Wu, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Schlüsselwörter
ChatGPT, Clustering Algorithm, Fuzzy Control Function, Financial Risk Control Algorithm, Intelligent Finance