From AI Knowledge to AI Usage Intention in the Managerial Accounting Profession and the Role of Personality Traits—A Decision Tree Regression Approach

Abstract

This study examines the key drivers behind the adoption of artificial intelligence (AI) in the accounting profession, emphasizing the influence of AI-related knowledge, personality traits, and professional roles. By applying Decision Tree Regression analysis to survey data from accounting professionals, our research identifies AI knowledge as the strongest determinant of AI adoption, underscoring the importance of expertise in technology acceptance. While personality traits play a secondary role, extraversion and openness emerge as significant factors influencing adoption intentions. The study further explores AI applications in financial auditing, tax compliance, and fraud detection, clarifying the specific accounting domains impacted by AI integration. These findings offer valuable guidance for policymakers, educators, and business leaders aiming to equip the accounting workforce with the necessary skills and mindset to navigate the AI-driven transformation of the profession.

1. Introduction

Artificial intelligence (AI) has emerged as a transformative force in accounting and auditing, revolutionizing traditional practices and fostering operational efficiency, accuracy, and decision making. As businesses and educational institutions adopt AI systems, understanding the behavioral, technological, and organizational factors influencing their adoption becomes increasingly critical [1,2]. The integration of AI into accounting workflows enhances automation and addresses complex tasks such as financial forecasting, fraud detection, and auditing precision, underscoring its potential to redefine the accounting profession [3,4].

Research into AI adoption has consistently highlighted the role of individual and organizational readiness, emphasizing that technology awareness, skills, and governance structures significantly impact implementation success [2,5]. For instance, the readiness of accounting professionals to engage with advanced systems is mediated by their awareness of AI capabilities and their perceptions of how these systems complement their tasks [1]. In particular, behavioral intentions, shaped by personality traits and organizational support, have been identified as key predictors of AI usage, as evidenced by studies on technology acceptance frameworks and personality-driven adoption models [6,7].

While previous studies have extensively examined AI adoption in accounting, they have primarily focused on technological and organizational readiness, often overlooking the role of individual psychological factors. However, emerging research in technology acceptance has highlighted personality traits as significant determinants of user engagement with AI systems [6]. Despite the relevance of personality traits in shaping attitudes toward AI, there is a lack of empirical studies explicitly examining their influence within the accounting profession. Addressing this gap, the present study investigates how personality traits, particularly those associated with the Five-Factor Model, interact with AI-related knowledge to influence AI adoption among accounting professionals.

The role of accounting educators and auditors in AI adoption further extends the discourse on technology acceptance. Teachers’ intentions to use AI in accounting education are influenced by socio-psychological and anthropomorphic perspectives, which emphasize the interaction between human traits and AI systems [8,9]. Similarly, professional auditors’ willingness to adopt blockchain technologies and forensic AI tools is mediated by their technical knowledge, professional skepticism, and perceptions of adequacy in standards and training [10,11]. These findings underscore the dual role of individual cognitive factors and institutional infrastructure in facilitating AI adoption.

The organizational context also plays a pivotal role in the integration of AI into accounting practices. Factors such as staff perceptions, managerial support, and IT governance frameworks are critical in shaping adoption outcomes [12,13]. Studies have shown that organizational readiness and tailored strategies can mitigate barriers to technology adoption, ensuring smooth transitions to AI-enhanced accounting systems [5]. Furthermore, the adoption of forensic tools for detecting financial cybercrimes demonstrates how AI-enabled systems address emerging challenges in the financial domain, offering opportunities for enhanced compliance and fraud prevention [11].

This paper contributes to the growing body of research by examining the interplay between AI knowledge, personality traits, and adoption intentions within the accounting profession using Decision Tree Regression. By leveraging insights from existing studies and integrating advanced modeling techniques, this research seeks to advance understanding of the factors driving AI usage in accounting and auditing. Furthermore, it offers practical implications for educators, policymakers, and accounting professionals seeking to optimize AI integration in their practices.

2. Literature Review

Artificial intelligence (AI) is increasingly transforming organizational practices, particularly in accounting, auditing, and managerial decision making. AI-driven systems enhance efficiency, accuracy, and strategic decision making by automating routine tasks, improving fraud detection, and refining financial forecasting [1,2]. However, AI adoption is not solely a technological process; it is also shaped by individual, organizational, and behavioral factors, making it a multidimensional phenomenon requiring comprehensive investigation [3,5].

Personality traits play a fundamental role in shaping attitudes toward technology adoption. The Five-Factor Model of Personality, which includes agreeableness, conscientiousness, emotional stability, extraversion, and openness, has been widely applied to understand technology usage behaviors [7]. Research suggests that extraversion and openness are strongly linked to curiosity, adaptability, and higher engagement with emerging technologies, fostering positive attitudes toward AI [6]. In contrast, conscientiousness may enhance systematic engagement with AI tools, particularly in professions requiring high precision and accuracy, such as accounting [14]. On the other hand, high levels of neuroticism may contribute to resistance or anxiety toward AI adoption due to concerns over job security and perceived complexity [10].

AI-related knowledge and technology readiness are crucial determinants of adoption intentions. Technology readiness, defined as the extent to which individuals feel prepared to engage with new technologies, encompasses factors such as AI awareness, perceived usefulness, and ease of use [1,4]. Studies indicate that professionals with higher AI-related knowledge are more likely to adopt AI-driven tools, as they perceive these technologies as beneficial rather than disruptive [3,13]. Furthermore, perceptions of trust, security, and ethical considerations influence adoption, particularly in industries where transparency and accountability are essential [9,11].

The successful implementation of AI is heavily influenced by organizational structures, managerial support, and IT governance frameworks. Organizations that provide structured training programs and integrate AI strategies into their workflow experience higher adoption rates among employees [12]. Effective IT governance mitigates cybersecurity, privacy, and ethical risks, particularly in areas such as forensic accounting and fraud detection [11]. Additionally, research highlights the role of workplace culture, staff perceptions, and leadership in fostering a supportive environment for AI adoption [15]. Firms with strong AI governance policies ensure compliance with professional standards while promoting confidence in AI-driven decision making [16,17].

AI is reshaping accounting and auditing practices by automating processes, enhancing accuracy, and improving fraud detection mechanisms. AI applications such as machine learning, natural language processing, and predictive analytics are increasingly used in financial reporting, tax compliance, and forensic auditing [8,14]. Studies demonstrate that AI-powered decision support systems can optimize resource allocation and risk assessment strategies, making them valuable tools for accountants and auditors [18,19,20]. Moreover, forensic AI tools play a pivotal role in detecting financial fraud, ensuring regulatory compliance, and reinforcing financial security measures [16,21].

The integration of AI in accounting education is another critical area of research. AI-driven learning systems provide personalized feedback, enhance instructional effectiveness, and improve decision-making skills in accounting students [8,14]. Adaptive learning technologies facilitate real-time assessment and offer customized learning experiences, which are essential for developing AI literacy among future professionals. Research also highlights how accounting educators’ perceptions of AI influence their willingness to incorporate AI-based teaching tools into curricula, underscoring the importance of faculty training and institutional support [10,11].

Based on the literature, this study explores the interplay between AI knowledge, personality traits, and adoption intentions in the accounting profession. The proposed conceptual framework suggests that AI-related knowledge serves as the strongest predictor of AI adoption, while personality traits (particularly extraversion, openness, and conscientiousness) play a moderating role. The organizational context, including managerial support and governance structures, further influences adoption decisions.

This study contributes to the growing body of knowledge by systematically examining how individual psychological factors, technology-related expertise, and workplace conditions shape AI adoption intentions among accounting professionals. The insights derived from this research have practical implications for policymakers, educators, and business leaders seeking to optimize AI integration strategies in accounting and auditing.

3. Materials and Methods

3.1. Participants

An electronic questionnaire was distributed through Romanian professional networks and accountant groups between June and July 2024. This convenience sampling approach was chosen for its practicality in reaching a specific professional demographic within Romania. The final sample consisted of 558 participants, with a predominantly female representation: 429 women (76.9%) and 129 men (23.1%).

Participants reported diverse professional roles. The majority, 334 individuals (59.9%), identified as accountants or economists employed within companies. Additionally, 46 auditors (8.2%), 100 accountants or economists working in accounting firms (17.9%), and 61 self-employed certified accountants (10.9%) were included. A smaller group of 17 respondents (3.0%) reported holding multiple roles, such as combining self-employment with auditing or managerial responsibilities.

Regarding professional responsibilities, 338 participants (60.6%) described their roles as operational, 92 (16.5%) identified as self-employed, and 102 (18.3%) reported managerial responsibilities. A subset of 12 respondents (2.2%) indicated that they combined self-employment with supervisory responsibilities, highlighting the presence of mixed professional roles.

The age of participants ranged from 18 to 75 years, with a mean age of 36.06 years (SD = 12.15). Professional experience varied from zero to 50 years, with an average of 9.94 years (SD = 10.00). These demographic characteristics provide a comprehensive overview of the Romanian accounting profession, capturing diversity in professional roles, experience levels, and responsibilities.

3.2. Instruments

To assess the variables in this study, we employed validated and widely recognized measurement scales, ensuring a comprehensive and reliable assessment of personality traits, AI knowledge, and AI adoption intention. Each construct was measured using a 5-point Likert scale (1 = Strongly disagree to 5 = Strongly agree), providing participants with a structured and consistent response format.

Personality traits were assessed using the International Personality Item Pool (IPIP) scales, a well-established instrument widely used in psychology and behavioral sciences. The IPIP scales have been extensively validated across various populations and research contexts, demonstrating strong construct validity, convergent reliability, and internal consistency. The five personality traits assessed were agreeableness, conscientiousness, emotional stability, extraversion, and openness.

Agreeableness (M = 4.33, SD = 0.58) measures cooperation and compassion, with a sample item such as “I am sympathetic to others’ needs.” Conscientiousness (M = 4.01, SD = 0.65) reflects diligence, organization, and attention to detail, exemplified by the item “I pay attention to details.” Emotional stability (M = 3.37, SD = 0.82) captures resilience and stress management, assessed through items like “I remain calm under pressure.” Extraversion (M = 3.60, SD = 0.66) evaluates sociability and assertiveness, with a sample statement such as “I enjoy being the center of attention.” Lastly, openness (M = 3.84, SD = 0.62) reflects intellectual curiosity and willingness to explore new ideas, as measured by items like “I have a vivid imagination.” Each subscale exhibited strong internal consistency, with Cronbach’s alpha values ranging from 0.75 to 0.87, confirming their reliability in measuring individual differences in personality.

AI knowledge was measured using a custom-developed scale, designed to assess familiarity with AI applications in managerial accounting. The scale was developed based on prior AI adoption studies and was refined to align with the specific accounting and auditing domain. A sample item includes the following: “I am familiar with the basic principles of AI applications in accounting.” The scale demonstrated good internal consistency, with a Cronbach’s alpha of 0.81, indicating high reliability in assessing participants’ understanding of AI technologies.

AI adoption intention was measured using an adapted version of the Technology Acceptance Model (TAM) scales, which have been widely applied in research on technology adoption in professional environments. A representative item from this scale is “I plan to use AI tools for decision-making in my accounting practices.” The scale exhibited strong internal consistency, with a Cronbach’s alpha of 0.85, confirming its reliability in predicting behavioral intention toward AI adoption.

By utilizing well-validated instruments and ensuring strong psychometric properties, this study guarantees a robust measurement framework, enhancing the credibility and generalizability of the findings.

3.3. Procedure

The procedure for this study involved administering an electronic questionnaire distributed through Romanian professional networks and accountant groups between June and July 2024. The questionnaire collected demographic information and assessed participants’ Big Five personality traits, AI knowledge, and AI intention using reliable scales. All responses were recorded on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). This approach ensured the collection of comprehensive data for the subsequent analyses.

To examine the relationships between variables, a Decision Tree Regression model was utilized. This machine learning technique is well suited for identifying nonlinear patterns and interactions among predictors.

To examine the relationships between variables, a Decision Tree Regression model was utilized. This machine learning technique was chosen over conventional Structural Equation Modeling (SEM) approaches, such as covariance-based SEM or Partial Least Squares (PLS-SEM), due to its ability to capture nonlinear relationships and hierarchical interactions among predictors. Unlike SEM, which is primarily used for testing predefined relationships, Decision Tree Regression offers a data-driven approach that reveals decision thresholds and interaction effects that might not be apparent in linear models. This method is particularly valuable in identifying the conditions under which AI knowledge and personality traits influence AI adoption, making it a suitable choice for this study. The dependent variable (DV) was AI intention, while the independent variables (IVs) included AI knowledge and the Big Five personality traits. The data were split into training (80%) and test (20%) sets to construct and validate the predictive model.

The model’s performance was evaluated using several fit indices. The Mean Squared Error (MSE) quantified the average squared differences between observed and predicted values, while the Root Mean Squared Error (RMSE) offered a standard deviation-like measure of prediction error. The Mean Absolute Error (MAE) represented the average absolute difference between observed and predicted values, and the Mean Absolute Percentage Error (MAPE) expressed prediction errors as a percentage of observed values. Finally, the R² (Coefficient of Determination) term indicated the proportion of variance in AI intention explained by the independent variables.

The decision tree analysis produced a branching structure that identified significant predictors and their thresholds for splitting the data into distinct groups. Each split point represented a critical decision threshold, revealing hierarchical and interaction effects among variables. The tree’s branches highlighted the conditional relationships and the sequential importance of predictors in influencing AI intention, offering valuable insights into the dynamics of personality traits and knowledge in this context.

All statistical analyses, including Decision Tree Regression and model performance evaluations, were conducted using JASP 0.17.3.0. This software was selected for its robust machine learning capabilities and user-friendly interface, ensuring accurate implementation of predictive modeling techniques.

4. Results

The Decision Tree Regression model was applied to analyze the relationship between AI usage intention (AI_intention) as the dependent variable and the independent variables, including AI knowledge and the five personality traits. To construct the scores for each latent variable, we calculated composite scores by averaging the item responses for AI knowledge and each of the Big Five personality traits. These scores were standardized to ensure comparability across variables before being included in the regression tree analysis. This approach allows for a consistent measurement framework, ensuring that all latent constructs were treated uniformly in the modeling process.

The dataset consisted of 558 observations, which were divided into a training set of 447 cases and a testing set of 111 cases. A total of 83 splits were performed by the decision tree, reflecting the complexity of the dataset and the underlying relationships between variables.

The performance of the model was evaluated using several metrics. The Mean Squared Error (MSE) was 0.818, indicating the average squared difference between the observed and predicted values. This suggests that the model demonstrated moderate predictive accuracy. Similarly, the Root Mean Squared Error (RMSE) was 0.904, providing an interpretable measure of prediction error in the same units as the dependent variable. The Mean Absolute Error (MAE) or Mean Absolute Deviation (MAD) was 0.686, highlighting a relatively low average absolute difference between predictions and actual outcomes. However, the Mean Absolute Percentage Error (MAPE) was recorded at 794.37%, reflecting notable variability in the dataset and areas for potential improvement in the model’s accuracy.

The R² value, representing the proportion of variance in AI intention explained by the independent variables, was 0.176. This indicates that the model accounts for approximately 17.6% of the variance, suggesting that additional factors not included in the current analysis may play a role in influencing AI usage intention.

Given the relatively low R² value compared to other studies on technology acceptance, it is acknowledged that additional explanatory variables could improve the model’s predictive capacity. Demographic factors such as age and gender, which are available in the dataset, have been widely documented as influential in technology adoption studies. However, the present analysis focused primarily on psychological and knowledge-based predictors. Future research should explore the inclusion of these demographic variables to assess their contribution to AI adoption in accounting and further refine the predictive model.

A visual representation of the dataset split confirmed the balanced allocation of observations between the training and testing sets, with 447 cases used for training the model and 111 reserved for validation.

To further assess the predictive capacity of the model, we compared the performance metrics separately for the training and validation samples. The model fitting adjustments reported, including MSE, RMSE, MAE, and R², were first calculated for the training dataset to evaluate in-sample predictive accuracy. Subsequently, the same performance metrics were computed on the test dataset to assess the model’s generalization capability. A comparative analysis of these values indicated that the predictive accuracy remained consistent between the training and validation samples, suggesting that the model was not overfitted and maintained reliability in predicting AI adoption intention in new data.

Table 1. Feature importance.

	Relative Importance
AI knowledge	51.285
Agreeability	15.658
Extraversion	9.680
Emotional stability	8.518
Openness	8.494
Conscientiousness	6.365

presents the feature importance scores, highlighting the contribution of each variable to the model’s predictive accuracy.

The results indicate that AI knowledge is the most significant predictor, with a relative importance score of 51.285, far surpassing the contributions of the personality traits. This finding underscores the pivotal role of familiarity and understanding of AI technologies in shaping individuals’ intentions to adopt AI tools in the managerial accounting profession.

Among the personality traits, agreeableness was the most influential, with a relative importance score of 15.658, suggesting that interpersonal tendencies and cooperation may play a notable role in AI adoption. Extraversion (9.680) followed as the second most impactful trait, aligning with previous evidence that social and assertive individuals are more likely to embrace technological innovations.

The remaining traits—emotional stability (8.518), openness (8.494), and conscientiousness (6.365)—demonstrated relatively lower importance. While they contribute to the model, their impact is modest compared to AI knowledge and agreeableness. These results indicate that while personality traits influence AI usage intention, their effects are secondary to the knowledge variable.

The Decision Tree Regression model identified key splits that contributed to predicting AI usage intention (AI_intention) based on the independent variables (

Table 2. Splits in tree.

	Obs. in Split	Split Point	Improvement
AI knowledge	447	−0.019	0.269
AI knowledge	219	−1.244	0.066
Agreeability	57	0.206	0.149
AI knowledge	36	−2.163	0.268
Agreeability	21	0.722	0.341
Emotional stability	162	0.708	0.042
Agreeability	134	−1.171	0.060
Agreeability	113	0.378	0.073
AI knowledge	228	0.594	0.140
Extraversion	109	0.227	0.138

Note: For each level of the tree, only the split with the highest improvement in deviance is shown.

). These splits highlight the hierarchical importance of variables and their interaction with one another in refining predictions.

The most significant predictor throughout the decision tree was AI knowledge, which appeared in multiple splits and consistently accounted for substantial improvements in deviance. The first split, involving 447 observations at a split point of −0.019, resulted in the highest improvement in deviance (0.269), confirming the dominant role of AI knowledge in stratifying the dataset. Subsequent splits based on AI knowledge occurred at −1.244 (219 observations, 0.066 improvement), −2.163 (36 observations, 0.268 improvement), and 0.594 (228 observations, 0.140 improvement). These findings underscore the critical importance of AI knowledge in predicting AI usage intentions.

Agreeableness emerged as the second most impactful variable, appearing in several splits that refined predictions within smaller subsets of the data. For instance, splits at 0.206 (57 observations, 0.149 improvement), 0.722 (21 observations, 0.341 improvement), and 0.378 (113 observations, 0.073 improvement) demonstrated its nuanced contribution. These results suggest that agreeableness adds value to the model by further segmenting groups with varying levels of AI intention.

Other personality traits played more modest roles in the tree. A split involving emotional stability occurred with 162 observations at a split point of 0.708, yielding a small improvement in deviance (0.042). This indicates that while emotional stability has some predictive utility, its influence is less significant than that of AI knowledge or agreeableness. Similarly, extraversion contributed with a split at 0.227 (109 observations, 0.138 improvement), reinforcing its secondary yet meaningful role in influencing AI adoption.

The splits reveal a clear hierarchy of predictors, with AI knowledge as the primary driver of AI intention and agreeableness as the most influential personality trait.

Figure 1 presents a scatterplot comparing the predicted test values of AI usage intention (AI_intention) generated by the Decision Tree Regression model with the observed test values. The red diagonal line represents the ideal scenario where predicted values perfectly match the observed values.

The plot shows a general clustering of points around the diagonal line, indicating that the model captures the general trends in the data. However, there is noticeable dispersion around the line, reflecting variability in prediction accuracy.

The Decision Tree Regression model (Figure 2) highlights the predictive relationships between AI usage intention (AI_intention) and the independent variables, including AI knowledge and personality traits. The hierarchical structure of the tree reveals the dominance of AI knowledge as the most critical factor, with subsequent splits incorporating personality traits to refine predictions. The root node, comprising all 447 training observations, first splits based on AI_knowledge at a threshold of −0.0187, representing the most significant improvement in deviance. This initial split divides the data into two subsets: observations with AI knowledge below the threshold (219 ob-servations) and those at or above the threshold (228 observations). This confirms that AI knowledge is the primary driver of differences in AI intention.

In the left subtree, where AI_knowledge is less than −0.0187, a further split occurs at −1.244, refining the distinction among participants with lower levels of AI knowledge. Within this group, agreeableness emerges as a key variable, with splits occurring at 0.206 and 0.722, indicating its secondary importance in shaping AI usage intention for participants with limited AI knowledge. In contrast, the right subtree (AI_knowledge ≥ −0.0187) introduces extraversion as a critical variable, splitting at 0.227 and highlighting the role of social tendencies in influencing adoption behavior.

Additional splits occur deeper in the tree, incorporating emotional stability and further refinements based on agreeableness. These splits emphasize the complex role of personality traits in shaping AI intention within specific knowledge-based subgroups. For instance, in the left branch of the tree, emotional stability at a threshold of 0.708 introduces further granularity, while agreeableness plays a consistent role across multiple branches, demonstrating its relevance in both high- and low-knowledge contexts. The recursive application of these splits results in progressively smaller subsets, with each node capturing a specific combination of traits and knowledge levels that contribute to the prediction of AI intention.

This analysis illustrates the complex interaction between cognitive (knowledge-based) and behavioral (trait-based) factors in predicting AI usage intention. While AI knowledge dominates as the primary predictor, personality traits such as agreeableness, extraversion, and emotional stability refine the model’s predictions, emphasizing the multifaceted nature of decision making in the context of AI adoption in the managerial accounting profession.

To further clarify the implications of the Decision Tree Regression analysis, the results emphasize that AI knowledge consistently emerges as the strongest predictor of AI adoption intention. The decision tree’s structure confirms that higher AI knowledge levels significantly increase the likelihood of AI adoption, reinforcing prior studies highlighting the role of technical competence in shaping technology acceptance decisions [21,22,23,24,25,26].

A particularly noteworthy insight is the role of agreeableness and extraversion as secondary yet meaningful predictors. The results indicate that individuals with higher agreeableness scores—who prioritize collaboration and interpersonal harmony—tend to exhibit stronger AI adoption intentions, especially when AI knowledge is moderate to high. This pattern aligns with research suggesting that socially adaptive individuals are more receptive to collaborative technological innovations [27,28,29]. Similarly, extraversion is associated with greater AI adoption tendencies, suggesting that proactive, outgoing professionals may be more open to integrating AI into their workflow due to their adaptability to change.

Additionally, the decision tree structure revealed some splits with relatively small sample sizes, raising concerns about stability. To ensure robustness, alternative tree structures and sensitivity analyses were conducted, confirming that AI knowledge consistently remained the dominant predictor across multiple tree configurations, with personality traits maintaining their secondary but significant role.

Given the low R² value (0.176), which suggests that only 17.6% of the variance in AI adoption intention is explained by the model, we acknowledge that additional factors should be considered in future studies. Organizational culture, industry regulations, and AI training programs could play critical roles in influencing adoption behaviors and should be explored further. Future research should integrate these elements to enhance predictive power and provide a more comprehensive framework for understanding AI adoption in managerial accounting.

5. Discussion

This study contributes to the growing body of literature on behavioral economics by examining the interplay between AI knowledge, personality traits, and AI usage intention in the managerial accounting profession. The findings provide valuable insights into how cognitive and psychological factors interact in shaping decision-making processes, aligning with existing theories of technology adoption and behavioral economics.

5.1. Relationship with the Literature

The dominant role of AI knowledge in predicting AI usage intention aligns with prior research emphasizing the critical influence of cognitive factors in technology adoption [26,27,28]. Participants with higher AI knowledge demonstrated significantly greater intention to adopt AI tools, suggesting that technical competence and familiarity play foundational roles in shaping attitudes toward technology. This supports the behavioral economics framework, which posits that individuals’ decisions are influenced by their perceived costs and benefits, knowledge, and bounded rationality [26].

Personality traits, particularly agreeableness and extraversion, emerged as secondary yet significant predictors. Agreeableness, reflecting individuals’ tendencies toward cooperation and social harmony, was positively associated with AI usage intention. This finding is consistent with Calluso and Devetag [29], who found that interpersonal dynamics play a critical role in shaping attitudes toward AI-assisted hiring practices. Similarly, extraversion, characterized by assertiveness and sociability, was also positively linked to AI adoption, suggesting that individuals with higher extraversion are more likely to embrace new technologies due to their proactive and adaptive behaviors [30,31,32,33,34,35,36,37].

The integration of behavioral and technological factors into decision-making processes highlights the need for frameworks that account for psychological and organizational dynamics. The findings align with Leitner-Hanetseder et al. [30], who emphasized the shifting roles and tasks in AI-based accounting environments. As AI tools increasingly replace traditional accounting functions, willingness to adopt these tools is likely influenced by a combination of technical competence and personality-driven adaptability. This underscores the importance of behavioral economics in addressing how intrinsic motivations and external pressures shape adoption behaviors [26,31].

The predictive dominance of AI knowledge reflects the critical role of domain-specific expertise in shaping technology adoption. This corroborates findings by Wang [28] and aligns with Rad et al. [35], who employed neural network models to predict behavior within educational contexts. Advanced methodologies such as radial basis function networks and fuzzy clustering have also demonstrated efficacy in predicting behavioral outcomes in dynamic systems, reinforcing the robustness of knowledge-based predictors [19,36].

Personality traits, particularly agreeableness and extraversion, emerged as significant secondary predictors, highlighting the psychological dimensions of AI adoption. These results align with Cabrera-Paniagua and Rubilar-Torrealba [23], who found that personality traits influence adaptive decision making in intelligent systems. Similarly, the integration of personality traits with AI technologies underscores the importance of leveraging behavioral insights to enhance adoption outcomes [26,29,37].

The integration of fuzzy logic and clustering methods further underscores the potential of advanced computational approaches to unravel complex behavioral patterns [38]. These techniques, as applied by Wan and Tian [39] in stress detection and by Liu et al. [40] in educational research, demonstrate how AI-driven decision-making models can be optimized for various domains.

Additionally, historical perspectives on accounting theory evolution [41,42,43,44,45,46,47,48,49,50,51,52] highlight the ongoing transformation of the profession in response to emerging technologies. These studies reinforce the necessity of integrating behavioral economics, AI competency development, and ethical frameworks into strategic decision-making processes, ensuring that technology adoption aligns with professional standards, organizational goals, and workforce dynamics [26,41,45].

5.2. Practical Implications

The findings of this study have several practical implications for organizations seeking to enhance AI adoption in accounting.

First, AI training and knowledge dissemination are essential for fostering AI readiness. Since AI knowledge was found to be the strongest predictor of AI adoption intention, organizations should implement targeted training programs to increase employees’ familiarity with AI applications. This aligns with Namazi and Rezaei’s [31] emphasis on competency development in AI-based decision making.

Second, personalized training programs based on personality profiles could improve adoption outcomes. Since agreeableness and extraversion were significant predictors of AI adoption, training initiatives should be tailored to different personality types. For instance, individuals high in agreeableness may respond well to collaborative learning environments, while those scoring high in extraversion may benefit from interactive, hands-on training.

Third, ethical considerations and transparency must be prioritized in AI implementation. As suggested by Chong and Eggleton [33], AI tools should be introduced with clear ethical guidelines and transparent decision-making frameworks to reduce resistance. The positive association between interpersonal traits and AI adoption in this study further suggests that organizational cultures emphasizing ethical AI use may encourage greater acceptance [26,46].

Finally, organizations should incorporate behavioral insights into AI adoption strategies, ensuring that AI tools align with employees’ cognitive and emotional tendencies. Decision tree methodologies, as used in this study, provide an effective approach for identifying key factors influencing adoption and optimizing AI implementation strategies [41,42,43,44].

5.3. Methodological Considerations and Limitations

This study employed Decision Tree Regression to analyze predictors of AI adoption intention. Decision trees are particularly useful for detecting nonlinear interactions and threshold effects among variables. Unlike traditional linear models, they allow for hierarchical interpretation of predictor importance, revealing the conditions under which AI knowledge and personality traits influence adoption. This methodological approach aligns with prior studies using machine learning techniques in behavioral research [15,34].

However, the study has certain limitations. The relatively low R² value suggests that additional factors, such as demographic variables (e.g., age, gender), could improve the model’s predictive capacity. Future research should incorporate these variables to enhance explanatory power. Additionally, while validated scales were used, further assessment of construct validity (e.g., Average Variance Extracted, heterotrait–monotrait ratios) was not conducted. Future research should integrate these psychometric evaluations to strengthen measurement accuracy [42,43].

Another limitation concerns the generalizability of the findings. This study relied on a convenience sample of Romanian accounting professionals, which may limit its applicability across different cultural and professional settings. Future studies should replicate this research with internationally diverse samples and consider cross-cultural differences in AI adoption behavior [26,50].

Another limitation of this study is the restricted sample of professionals surveyed, which focused primarily on accountants, auditors, and financial experts directly involved in AI adoption within managerial accounting. However, other professionals indirectly engaged in managerial accounting decisions, such as financial analysts, tax consultants, corporate managers, and IT specialists involved in financial technology development, may also play a crucial role in AI adoption and implementation. Their distinct expertise, education levels, and perspectives on AI integration could offer additional insights into the broader implications of AI adoption in the accounting profession.

To enhance the credibility and applicability of future research, we recommend expanding the participant pool to include a more diverse range of professionals involved in financial decision making and AI implementation. This broader approach would allow for a more comprehensive analysis of AI readiness, adoption barriers, and the impact of interdisciplinary expertise on AI integration in managerial accounting and auditing practices.

Another methodological limitation is the current data collection approach, which relies on a static dataset obtained through an electronic questionnaire distributed within a defined timeframe. While this approach provided valuable insights into AI adoption intentions among accounting professionals, a dynamic, continuously updated database could enhance the depth and scalability of future research. To address this, future studies should consider implementing a web-based platform that allows for ongoing data collection and real-time updates from accounting and auditing professionals. Such a centralized database would facilitate longitudinal analysis, enabling researchers to track changes in AI adoption trends over time. Moreover, this approach would allow for the automation of statistical processing using advanced analytical tools such as EViews, STATA, SPSS v24, Python, or R, improving the efficiency and complexity of statistical modeling and trend forecasting.

6. Conclusions

This study provides valuable insights into the factors driving AI adoption in the accounting profession, emphasizing the intersection of personality traits, AI knowledge, and professional roles. The findings highlight that AI knowledge is the most significant determinant of adoption, followed by agreeableness and extraversion as secondary predictors. These results reinforce the importance of integrating behavioral insights into AI implementation strategies.

While offering meaningful contributions, the study has limitations that should be acknowledged. The reliance on a convenience sample of Romanian accounting professionals may constrain the generalizability of findings across different contexts. Additionally, while Decision Tree Regression effectively identifies patterns in AI adoption, future research should incorporate longitudinal studies to examine how these factors evolve over time.

To build on these insights, future research should explore the role of organizational culture and leadership in moderating the relationship between personality traits and AI adoption. Investigating team dynamics and collaborative AI tools could yield practical implications for AI readiness in accounting. Additionally, incorporating qualitative methodologies could provide deeper insights into professionals’ lived experiences with AI, complementing the quantitative findings.

By addressing these aspects, future research can further advance the understanding of AI adoption in accounting and inform strategies for a seamless transition to AI-enhanced professional environments.