Abstract
This study investigates the influence of artificial intelligence (AI) adoption on the effectiveness of audit planning in the dynamic Egyptian business environment, with a focus on the moderating role of audit firm size. Using a 2 × 2 factorial experimental design, we conducted a field experiment with a purposive sample of 130 auditors, comprising 65 participants from Big 4 firms and 65 from non-Big 4 firms. Our findings confirm a significant positive association between AI adoption and audit planning effectiveness, which was measured across dimensions such as risk identification, efficiency, and clarity. The results further reveal a significant moderating effect of audit firm size, where the positive impact of AI on planning effectiveness is notably more pronounced in large audit firms. These findings, corroborated by sensitivity analyses, provide empirical support for Contingency Theory and the UTAUT. The study contributes to the literature by offering evidence from an emerging economy. It provides critical implications for regulators, audit firms, and accounting educators regarding the integration and governance of AI in auditing.
Plain Language Summary
Our study looks at how new technology, specifically Artificial Intelligence (AI), is changing the way auditors plan their work. We conducted an experiment with 130 auditors in Egypt to see if using AI tools makes their planning more effective. We found that using AI significantly improves audit planning by helping auditors find risks more accurately, work more efficiently, and better focus on important tasks. Our research also shows that this positive effect is much stronger in large audit firms compared to smaller firms. This is likely because large firms have more resources to invest in and train their staff on these powerful new tools. These findings suggest that for auditors to stay relevant, they must embrace AI. They also have important messages for regulators, who need to create new rules for using AI, and for audit firms of all sizes, who should plan for this technological shift.
Introduction
Audit planning is a foundational component of any audit engagement, setting the strategy and scope necessary to achieve audit objectives. The first standard of fieldwork promulgated within the generally accepted auditing standards (GAAS) explicitly stipulates the auditor’s responsibility to adequately plan the audit engagement. This planning process entails a series of structured activities undertaken by the auditor to formulate the overarching audit strategy for the engagement, as prescribed by the International Standard on Auditing (ISA) 300 (IAASB, 2009). This study investigates the influence of artificial intelligence (AI) adoption on audit planning effectiveness, which we define as the extent to which the planning process facilitates the attainment of audit objectives with the highest degree of efficiency and quality (Fatahi et al., 2025).
Effective audit planning enables the auditor to ensure that salient audit areas receive appropriate scrutiny and resource allocation, anticipated challenges are identified proactively and addressed promptly, and audit tasks are efficiently distributed among audit team members. Moreover, robust planning facilitates the supervisory and review processes, thereby enhancing audit quality and contributing to the effective coordination of activities undertaken by both internal audit personnel and external specialists engaged by the audit firm (Ilori et al., 2024).
While traditional planning methodologies rely on historical data and professional judgment, the integration of AI technologies is profoundly transforming this process. AI is a revolutionary area within computer science dedicated to developing intelligent machines that can perform cognitive tasks like those of humans, including learning, problem-solving, and decision-making. In the field of auditing, the rapid development and adoption of AI technologies, from advanced data analytics to generative AI, have introduced a new paradigm for the auditing profession (Boritz & Stratopoulos, 2023; Eulerich et al., 2024). Emerging literature consistently highlights how AI tools augment the audit process by automating repetitive tasks, enhancing the accuracy and reliability of financial analysis, and processing vast datasets in a fraction of the time required by traditional methods (Aitkazinov, 2023; Fedyk et al., 2022; Kokina et al., 2025). Specifically, AI enhances audit planning by analyzing large volumes of transaction data to identify and assess material risks, detect anomalous patterns, and flag potential misstatements with greater precision and speed. In addition, the ability of AI to rapidly process and visualize data ensures that the audit plan can cover a broader and deeper range of financial information, thereby enhancing the scope of coverage and leading to a more comprehensive audit.
The escalating reliance on sophisticated data analytics and technological innovations in the business landscape has fundamentally reshaped the environment in which audit firms operate (Bose et al., 2023). Despite the global discourse on AI in auditing, empirical research addressing the Egyptian context, particularly concerning the complexities of AI adoption and its impact on critical audit processes, such as planning, remains limited. Furthermore, the varying capacities and structures of audit firms, particularly the distinction between large (Big 4) and smaller firms, suggest that the benefits and challenges of AI integration might not be uniformly experienced across the industry (Pham et al., 2025). This differential impact of AI, contingent on organizational characteristics, underscores the need for a nuanced investigation.
This evolution is particularly pertinent in emerging economies like Egypt, where the rapid adoption of FinTech and digital transformation initiatives is creating both opportunities and challenges for the auditing profession. Building on these critical observations, this study aims to investigate the influence of auditor adoption of AI technologies on the effectiveness of audit planning processes, with a particular emphasis on the moderating role of audit firm size in this relationship within the context of a developing country. Specifically, we seek to answer the following research questions: (1) Does the adoption of AI by auditors enhance the effectiveness of audit planning? (2) Does audit firm size moderate the relationship between auditors’ adoption of AI and the effectiveness of audit planning?
By conducting a field experiment involving 130 auditors employed by licensed accounting and auditing firms in Egypt and using the Wilcoxon Signed-Rank test to analyze paired data, the results indicate a statistically significant increase in the effectiveness of audit planning when auditors adopt AI in the Egyptian context. Furthermore, using the Mann–Whitney test, audit firm size was found to significantly moderate this relationship, strengthening the positive association between auditors’ adoption of AI and the effectiveness of audit planning.
Our study offers significant theoretical and practical contributions to the auditing field. Theoretically, it addresses the expectations gap by demonstrating how AI enhances audit planning effectiveness and expands knowledge on AI adoption in auditing by examining the moderating role of audit firm size, a gap in previous research. This contribution is threefold. Firstly, it provides empirical evidence supporting the relationship between AI adoption, firm size, and planning effectiveness. Secondly, by focusing on Egypt, it offers insights into AI adoption within a developing economy, enabling cross-national comparisons and a broader understanding of AI’s global impact on auditing. Finally, the use of a field experiment strengthens the research methodology, providing a template for future studies. Practically, by empirically demonstrating AI’s positive impact on audit planning, the study helps align public expectations of audit quality with the capabilities of modern auditing practices, effectively bridging the expectations gap.
The remainder of our study is structured as follows. Section “A Contextual Background” presents an overview of the Egyptian context. Section “Literature Review and Hypotheses Development” provides literature review and hypotheses development. Section “Research Design” details the research design. Section “Results and Discussion” discusses the findings. Finally, section “Conclusion” summarizes and concludes the paper.
A Contextual Background
The regulatory landscape in Egypt’s FinTech sector has evolved significantly, creating an ideal research setting. A key development was Law No. 5 of 2022, which established a new framework for non-banking financial activities, created a dedicated supervisory unit at the Financial Regulatory Authority (FRA), and aimed to increase financial inclusion. The FRA built upon this with two subsequent decisions: Decision No. 135 of 2022, which outlined grievance procedures and promoted the digitization of non-banking transactions by addressing technological infrastructure, digital identity, and record-keeping, and Decision No. 57 of 2024, which introduced Egypt’s first regulations for Robo-Advisors in its capital market, with a focus on data management and human supervision. These regulatory changes provide a unique opportunity to examine the effects of AI on auditing practices (Egyptian House of Representatives, 2022; FRA, 2022, 2024).
Regarding audit planning in the Egyptian audit market, Egyptian Auditing Standard (EAS) No. 300 provides the normative framework for planning financial statement audits, encompassing both recurring and initial engagements. This iterative process begins after the prior audit and continues through the current one, emphasizing the timing of audit activities. This process includes defining the audit scope (identifying the financial reporting framework and specific reporting requirements), determining materiality, and conducting risk assessments. The standard aims to ensure audit quality and enhance stakeholder confidence in financial statement reliability (MOI, 2008a).
Although Egyptian auditing standards lack a specific standard dedicated to audit firm size, its influence is implicitly recognized across several standards. For instance, EAS No. 300 requires auditors to consider available resources (personnel quantity and expertise) during planning, which directly correlates with firm size. Larger firms typically have more human and financial capital, allowing for specialized teams and more sophisticated methodologies and technologies. Similarly, larger firms are better equipped to understand and assess the risks of material misstatement, especially in large and complex organizations, which is guided by EAS No. 315 that stresses the importance of understanding the audit client’s internal controls and operational context (MOI, 2008b).
Literature Review and Hypotheses Development
AI Technologies and Their Applications in Auditing
Artificial intelligence (AI) encompasses various advanced computational techniques designed to simulate human intelligence. In the auditing domain, key AI technologies include advanced data analytics, machine learning, natural language processing (NLP), and, more recently, generative AI models like ChatGPT (Kokina et al., 2025). These technologies empower auditors to move beyond traditional manual analysis, leveraging computational power to process and derive insights from vast and complex datasets.
Specifically, advanced data analytics and machine learning enable auditors to scrutinize large datasets, identify complex patterns, detect anomalies, and make data-driven predictions crucial for risk identification and audit planning. NLP tools are instrumental in analyzing unstructured data, such as contracts, emails, and qualitative disclosures, enhancing the comprehensiveness of information gathering (Bose et al., 2023; Kokina et al., 2025). Furthermore, the advent of generative AI, exemplified by models like ChatGPT, has introduced capabilities for automated report generation, intelligent querying, and even high-level cognitive tasks relevant to auditing, demonstrating disruptive potential in the industry (Eulerich & Wood, 2025; Eulerich et al., 2024).
The adoption of these AI technologies offers a range of benefits to the auditing profession. They contribute to enhanced efficiency by automating repetitive analytical tasks, thereby accelerating audit procedures and optimizing resource allocation. Moreover, AI improves the accuracy and reliability of analyses, helps mitigate human error, and facilitates the identification of complex patterns that might be overlooked by traditional methods. These capabilities allow auditors to conduct more comprehensive and insightful assessments, fundamentally augmenting the audit process across its various stages (Boritz & Stratopoulos, 2023; Munoko et al., 2020).
For instance, tools often classified primarily for business intelligence, like Microsoft Power BI Desktop, while not deep learning platforms, embody practical applications of AI’s core capabilities. They facilitate sophisticated data aggregation and transformation, enable pattern recognition and anomaly detection through calculated measures and built-in “Quick Insights” features, and offer interactive data visualization for enhanced decision support. These functionalities directly contribute to automating routine data analysis, identifying risks more efficiently, and generating clearer insights—all central tenets of AI-driven auditing.
Theoretical Perspectives
The expected connection between auditors adopting AI and improved audit planning is grounded in multiple theories, most notably contingency theory and the unified theory of acceptance and use of technology (UTAUT) (Deniswara et al., 2023; Piosik & Karmańska, 2023). Contingency theory posits that a system’s effectiveness is a result of the interaction between an organization’s internal traits (like its size and structure) and external factors (such as environmental uncertainty). For AI in auditing, this means successful planning hinges on how resources are allocated for the technology. The theory emphasizes that technologies must be customized to specific situations. Therefore, AI adoption may be more effective in certain organizational structures, ultimately leading to better audit results and improved performance through enhanced planning and collaboration (Piosik & Karmańska, 2023).
Contingency theory suggests that the size of an audit firm significantly influences how effective AI tools are in audit planning. Larger audit firms are better equipped to use AI because they have more financial resources, advanced technology, and specialized staff. These factors allow bigger firms to invest in, implement, and make the most of AI. As a result, the positive effects of AI on audit planning are likely to be much stronger in larger firms. Additionally, larger firms benefit from their enhanced resource capacity and strong market reputation, which attracts clients looking for credibility. While smaller firms might not have the same resources, they can build their reputation by focusing on social responsibility, creating a competitive edge in the audit market (Zhoulie & Williams, 2025).
UTAUT extends the Technology Acceptance Model (TAM) by incorporating additional constructs, such as social influence and facilitating conditions. Social influence is the perceived belief of significant others (peers, supervisors, leadership) that one should use technology. Facilitating conditions are the available organizational and technical resources, support, and training needed for effective use. In auditing, UTAUT suggests that auditors’ AI adoption is positively influenced by perceived social support and adequate facilitating conditions (training, support, infrastructure). This increased adoption is hypothesized to enhance the effectiveness of audit planning (Deniswara et al., 2023).
AI Adoption by Auditors and Audit Planning Effectiveness
Audit planning involves establishing the audit strategy and developing a detailed plan, which is crucial for achieving audit objectives. Effective audit planning, as mandated by ISA 300, is fundamental to ensuring audit quality and requires a clear audit strategy with key indicators such as precise risk assessment, judicious resource allocation, and tailored audit procedures (Avlokulov, 2024; IAASB, 2009). The integration of AI tools is profoundly impacting this process. By automating procedures and enhancing the accuracy and reliability of financial statement analysis, AI augments the audit process and helps mitigate human errors and detect illicit activities like fraud and money laundering (Munoko et al., 2020).
Recent research consistently indicates a strong positive relationship between auditors’ adoption of AI technologies and the effectiveness of their audit planning (Aitkazinov, 2023; Fedyk et al., 2022; Ganapathy, 2023; Kokina et al., 2025; Onwubuariri et al., 2024). This positive effect is particularly notable in risk assessment, where AI’s ability to process massive amounts of data and recognize complex patterns allows for the identification and evaluation of material risks with greater precision and speed than traditional methods (Onwubuariri et al., 2024). The use of generative AI, for example, can further assist in the creation of risk-based audit plans by automating risk assessments and suggesting schedules (Eulerich & Wood, 2025). This enhanced risk identification ultimately enables auditors to develop more customized and effective procedures, thereby improving the overall quality and comprehensiveness of the planning phase.
Beyond risk assessment, AI also contributes to improved time and resource efficiency by automating data collection, preliminary analysis, and the generation of certain planning documentation (Boritz & Stratopoulos, 2023). This optimized use of resources aligns with audit objectives emphasizing efficiency and allows auditors to focus on more complex analytical and judgmental activities. Furthermore, AI helps identify the most important audit procedures and focus efforts on them by providing clearer insights into areas that require greater attention, thereby enhancing the overall clarity and understandability of the planning process (Eulerich & Wood, 2025). These tangible benefits collectively underscore AI’s pivotal role in transforming and enhancing the effectiveness of the audit planning process.
The burgeoning literature on AI in auditing posits a clear and significant role for technology in enhancing traditional audit practices (Aitkazinov, 2023; Fedyk et al., 2022). This relationship is primarily informed by the UTAUT, which suggests that an individual’s behavioral intention to use a technology is influenced by their performance expectancy—the belief that using the system will help them attain gains in job performance. In the context of audit planning, AI tools are designed to improve efficiency, accuracy, and risk assessment, directly aligning with auditors’ performance expectations. By automating data-intensive tasks and providing rapid insights, AI is expected to enable auditors to create more comprehensive and effective plans. Consequently, the first hypothesis is posited as follows:
The Moderating Effect of Audit Firm Size
The existing literature consistently suggests a link between audit firm size and audit planning effectiveness (Fernandez et al., 2024; Ganesan et al., 2024; Pham et al., 2025). These studies find that larger firms, with their superior resources, technological sophistication, and specialized workforce, are better positioned to provide higher-quality audits and more effective planning procedures (Zhoulie & Williams, 2025). This environmental advantage is particularly crucial in the context of technological adoption, where a firm’s characteristics can act as a critical contingency.
This relationship can be explained through contingency theory, which states that an organization’s effectiveness relies on the alignment of its internal and external characteristics. In the context of technology, internal factors like a firm’s size and resources can either maximize or limit the benefits of a new tool. For instance, large audit firms are in a strong position to use AI because they have more financial resources, extensive expertise, and a client base with large, complex engagements. This environment is perfect for AI, whose main strength is processing the kind of extensive data these clients produce. This allows larger firms to invest in sophisticated AI software and specialized staff, making the benefits of pattern and anomaly detection more pronounced. Conversely, smaller firms with fewer resources and simpler engagements may not be able to fully take advantage of AI’s potential.
This perspective is further supported by the UTAUT framework, which highlights that larger firms are better positioned to provide the essential facilitating conditions, such as comprehensive training and robust IT infrastructure, which reduce perceived effort and boost auditors’ confidence in AI’s benefits. Based on this theoretical foundation, we hypothesize that audit firm size acts as a moderator, influencing the strength of the relationship between AI adoption and audit planning effectiveness.
Therefore, we posit that the interaction between audit firm size and auditors’ adoption of AI can serve as a moderating variable, potentially exerting an influence on both the direction and/or the magnitude of the relationship between auditors’ adoption of AI and the effectiveness of audit planning processes. Based on the above discussion, the second hypothesis can be derived as follows:
Research Design
Sample Selection
Our study uses a field experimental design, a method often used in social sciences where participants are exposed to different treatments that correspond to changes in an independent variable (Saleh, 2023). This approach lets researchers examine people’s opinions and cognitive abilities in a realistic setting. By systematically changing the independent variable in a controlled environment and observing how the dependent variable changes, this type of design is excellent for understanding the relationship between variables and testing hypotheses about causal effects (Babbie, 2007). In line with this, our experiment is designed to test if Egyptian auditors’ professional judgments about the effectiveness of audit planning are significantly affected by the use of AI, and whether this relationship is influenced by the size of their audit firm.
The study’s target population, consistent with the broader conceptualization of relevant stakeholders proposed by Peytcheva (2013), encompassed both auditors and academics. The principal focus of the empirical analysis was directed toward auditors employed within accounting and auditing firms possessing the requisite licenses to conduct audits of joint-stock companies. A judgmental sample of 250 individual auditors was subsequently selected from this delineated population. The sample selection methodology was guided by a set of predetermined criteria designed to ensure the representativeness of the selected sample of the broader target population. These selection criteria included: proportionate representation of auditors registered with the FRA and those operating outside the purview of FRA registration; and balanced representation of auditors from both large and small accounting and auditing firms, encompassing both formally registered and unregistered entities (Abouelela et al., 2025).
Our study utilized a purposive sampling approach, specifically judgmental sampling, to select participants with specific professional qualifications. Unlike random sampling, judgmental sampling allowed us to deliberately select auditors with relevant expertise. This approach was essential to ensure that participants could effectively engage with the experimental case and that the results of our manipulation would be internally valid. Furthermore, we included a secondary sample of academic personnel as a sensitivity analysis. This group was not intended to serve as a direct proxy for practicing auditors but rather as a robustness check to assess whether the fundamental theoretical relationships observed in our primary sample would hold in a different but knowledgeable population. This approach strengthens the generalizability of our findings by demonstrating consistency across different expert populations.
Accordingly, a secondary study population was delineated, comprising academic personnel, specifically faculty members and assistant staff affiliated with accounting departments of the faculties of commerce at two Egyptian universities. A judgmental sample of 120 academics was chosen using a purposive sampling approach to evaluate their awareness of the research topic, acknowledging its complexity and novelty. The primary criteria for selection were holding a postgraduate degree (master’s or doctorate) and specializing in auditing or financial accounting. To ensure consistency, the researchers also ensured a high degree of homogeneity within each qualification level (Abouelela et al., 2025; Saleh, 2023).
The primary analysis employed a judgmental sample of 250 individual auditors, comprising audit managers and senior auditors employed within accounting and auditing firms licensed to conduct audits of joint-stock companies. Of this initial sample, 130 responses were deemed usable for subsequent analysis, resulting in a response rate of 52%. While this response rate is acceptable for field experiments, the subsequent robustness checks, including the sensitivity analysis with the academic sample, provide additional confidence in the stability of our findings. For the supplementary sensitivity analysis, a separate judgmental sample of 120 individual academics affiliated with faculties of commerce at two Egyptian Universities was selected. This supplementary analysis yielded a final sample of 100 usable responses, corresponding to a response rate of 83%.
Variables’ Measurement
Dependent Variable
In this study, audit planning effectiveness is the dependent variable, defined as how well the planning process helps auditors achieve their engagement goals as efficiently as possible (Fatahi et al., 2025). This concept includes several key dimensions: a plan that is comprehensive (covering all important aspects), realistic (based on the client’s specific context), and flexible (adaptable to new information). We measured this effectiveness using a multi-item scale on a structured questionnaire. The questionnaire asked participating auditors to describe the audit procedures they would perform during the planning phase under two conditions: once using traditional, non-AI methods, and once using AI tools. Their responses were captured on a 5-point Likert scale.
Independent Variable
This study’s independent variable is the auditors’ adoption of AI, which we define as the use of AI tools to improve the efficiency and effectiveness of audit procedures. This integration is meant to free up auditors’ cognitive resources so they can focus on more complex, judgmental tasks and add more value to their work (Bui et al., 2025). To measure this, we designed a comparative experiment with two distinct case studies, each containing questions about the effectiveness of audit planning. The participating auditors were required to respond to these questions twice: once when they created audit plans using traditional, non-AI methods, and a second time when they used the Microsoft Power BI Desktop application to leverage its functionalities. Thus, while Power BI Desktop was utilized for its practical applicability in the experimental setting, it effectively represents the broader application of AI’s core capabilities (e.g., advanced data processing, pattern recognition, anomaly detection, and automated insights) that are transferable across various AI tools and crucial for enhancing audit planning.
For this study, Microsoft Power BI Desktop was selected as the specific AI technology to be used. The tool was chosen not merely for data visualization but for its advanced functionalities that are central to modern audit analytics. Its capabilities go beyond basic reporting to include automated data transformation, sophisticated anomaly detection, and predictive analysis. These features are considered core functions of AI in auditing, as they allow auditors to focus on more complex, judgmental tasks. For instance, Power BI’s Power Query feature automates complex data cleaning from various sources, essentially serving as a form of Robotic Process Automation (RPA). Additionally, its advanced DAX (Data Analysis Expressions) functions and built-in machine learning models were leveraged to find unusual transactional patterns that would be missed with traditional methods—a critical part of AI-enhanced risk assessment. These specific features justified the tool’s use as a representative example of an AI intervention for our study.
Power BI significantly enhances audit planning effectiveness by improving risk assessment, automating data handling, and providing powerful visualizations for better resource allocation. For example, its capabilities allow auditors to quickly build reports that highlight unusual spikes in expenses, unexpected revenue variations, or irregular transaction volumes, which are key indicators of potential fraud or material misstatements. In our hypothetical case, this meant auditors could use Power BI to rapidly identify discrepancies between recorded sales and shipping data, or to flag transactions with unusually high discounts. Furthermore, the tool automates the process of pulling and integrating data from multiple sources, freeing auditors from time-consuming manual reconciliation. This enables them to focus on more judgmental tasks and provides a comprehensive dataset for planning. Finally, Power BI’s visualization features allow auditors to see a clear breakdown of key data, such as accounts receivable, making it easier to spot problematic concentrations and allocate audit procedures more effectively.
Moderating Variable
Audit firm size is used as a moderating variable within the present study. This construct represents the scale of the accounting firm or company responsible for providing financial audit services to client entities (Zhang et al., 2025). The size of the audit firm is widely acknowledged as a salient contextual factor that can exert a significant moderating influence on the effectiveness of audit planning processes. For empirical analysis, this moderating variable was operationalized as a dichotomous (binary) variable, assigning a value of 1 to large audit firms (specifically, those classified as Big 4 firms) and a value of 0 to small audit firms (those not classified as Big 4) (Pham et al., 2025). This classification was based on participants’ self-reported audit firm affiliation provided in the demographic section of the questionnaire. The hypothesized relationships among the research variables are visually represented in Figure 1.
The research model.
The Experimental Study
Experimental Study Procedures and Tools
Our study used a within-subjects experimental design to test our hypotheses. This approach involved exposing a single group of auditors to two distinct conditions: one without AI and one with AI. This design is highly effective because it minimizes the impact of individual differences by comparing the performance of the same people under different circumstances (Charness et al., 2012). For H1, which proposes a direct effect of AI adoption on audit planning, we performed a within-subjects analysis. This involved comparing participants’ responses from the pre-test condition (Case 1: no AI) to the post-test condition (Case 2: with AI). For H2, which suggests that audit firm size moderates this relationship, we conducted a between-groups analysis. We compared the responses of two independent groups: auditors from large firms (Group A) and those from small firms (Group B). This allowed us to determine if firm size interacts with AI adoption to influence professional judgments on audit planning effectiveness.
The experimental protocol was structured as follows. Case 1 (Absence of AI), a homogeneous group of auditors received information for a hypothetical audit case and were instructed to plan audit procedures using only conventional methods. Case 2 (Presence of AI), 1 week later, the same auditors were given the same hypothetical case but were allowed to use an AI-driven tool, the Microsoft Power BI Desktop app, to assist in their planning. Before this condition began, participants received a dedicated training session to familiarize them with the app’s functionalities. Crucially, before the commencement of this condition, participants underwent a dedicated training intervention designed to familiarize them with the app, including instruction on its relevant functionalities and application within the context of audit planning, a week after exposure to the first condition. This training aimed to establish a baseline level of familiarity with the AI tool and mitigate potential performance differences stemming from varied prior experience. While conducted in actual audit firms to enhance ecological validity, the use of hypothetical cases allows for controlled manipulation of variables and standardization of the audit scenario, which is essential for internal validity in experimental research.
Data was collected through multiple field visits to a selection of accounting and auditing firms between mid-June and July 2024. During these visits, we distributed and collected structured questionnaires, which were the primary tool for measuring the effectiveness of audit planning in our experimental setting. This approach is justified to enhance data quality and internal validity by mitigating participant fatigue and preventing bias from a lengthy questionnaire. The on-site visits also allowed us to collect demographic information and provide participants with an opportunity to ask questions, which enhanced data quality and internal validity.
As explained in the Appendix, for Case 1 (Traditional Audit Planning), auditors were given a 3-hour time limit to create an audit plan for a fictional Egyptian electronics company, XYZ Inc., using only conventional methods. The provided materials included the company’s 2023 financial statements, a summary of its business activities, and past audit reports showing a clean history. The company operates in a dynamic environment, dealing with global supply chains and technological changes.
For Case 2 (AI-Assisted Audit Planning), the same auditors were given the same case information but were also trained on and allowed to use the Power BI app. They used the app to conduct deeper data analysis and more accurate risk assessments, specifically to detect anomalies related to a case of fraudulent misstatements in net sales. Like in Case 1, they had 3 hr to complete the plan. Afterward, they completed a questionnaire that measured their perceptions of AI’s impact on the planning process, including its effect on accuracy, risk identification, and time efficiency.
The measurement of audit planning effectiveness was conducted using a 5-point Likert-type scale, with response options ranging from 1 (strongly disagree) to 5 (strongly agree). The questionnaire instrument was designed to assess four distinct dimensions of planning effectiveness: (a) Comprehensiveness of planning assessed through “Your planning included complete identification and assessment of material risks”; (b) Efficiency of planning assessed through “Your planning was effective in terms of using available time and resources”; (c) Accuracy of planning assessed through “The planning helped you identify the most important procedures, focus efforts on them and avoid unnecessary actions”; and (d) Clarity of planning assessed through “Your planning was clear and easy to understand.” Additionally, four specific items were designed to assess the perceived direct impact of AI on the planning process as an additional analysis. These items included statements such as “The use of AI improved the accuracy of your planning,”“AI helped you better identify risks,”“Your planning covered all important aspects of the audit,” and “Using AI saved your time in the planning process.”
Experimental Design, Treatments, and Comparisons Experiment
The experimental design, presented in Table 1, employed a 2 × 2 factorial design, yielding four distinct experimental treatment conditions. These conditions were derived from the combination of two independent variables: AI-assisted audit planning (with two levels: present/absent) and audit firm size (with two levels: large/small). The four resultant treatment conditions were identified as follows: participants employed by large audit firms and performing audit planning without the utilization of AI tools; participants employed by large audit firms and performing audit planning with the utilization of AI tools; participants employed by small audit firms and performing audit planning without the utilization of AI tools; and participants employed by small audit firms and performing audit planning with the utilization of AI tools. Under all four treatment conditions, participants were required to assess the effectiveness of audit planning using the previously described measurement instrument.
Experimental Design, Treatments, and Comparisons.
Source. Created by the authors.
It is important to acknowledge that all experimental research designs face potential threats to both internal and external validity. To reduce these threats in our study, we implemented several methodological safeguards (Babbie, 2007). To prevent instrumentation effects, we used a standardized questionnaire with the exact same items across all experimental conditions. This ensured consistent measurement and minimized potential errors. To further improve internal validity, we introduced a 1-week gap between the two experimental cases, which helped to reduce the influence of other variables. In addition to these controls, we also conducted supplementary statistical analyses. To improve the external validity of our findings and make them more generalizable, we conducted the experiment as a field study within real accounting and auditing firms. As a further measure, we performed a sensitivity analysis by running the same experiment with a supplementary sample of academic personnel who assumed the role of auditors in a simulated setting. This helped to confirm the robustness of our findings.
Results and Discussion
Descriptive Statistics
The internal consistency reliability of the data collection instrument was assessed using Cronbach’s alpha (α), a widely employed statistical measure of internal consistency (Abouelela et al., 2025). Cronbach’s alpha quantifies the degree to which multiple items within a measurement scale measure the same latent construct. The resulting coefficient ranges from 0 to 1, with values approaching 1 indicating greater internal consistency and, consequently, higher reliability. A conventional criterion for acceptable internal consistency is a coefficient exceeding 0.5. In our study, the computed Cronbach’s alpha coefficient was .94 for the first experimental case (absence of AI intervention) and .93 for the second experimental case (presence of AI intervention). The obtained alpha values in both cases indicate an acceptable level of internal consistency, suggesting that the constituent items of the questionnaire instrument effectively measure the intended construct of audit planning effectiveness.
To provide a comprehensive understanding of the sample’s representativeness, our sample consisted of 130 auditors, with an equal representation of 65 participants from Big 4 firms and 65 from non-Big 4 firms. The professional experience of the auditors varied, with 40% having 0 to 5 years of experience, 35.4% having 6 to 10 years, and 24.6% having more than 10 years. In terms of seniority, junior auditors comprised 34.6% of the sample, while senior auditors made up 42.3%, and managers/partners accounted for 23.1%. This breakdown ensures a diverse range of experience levels is represented in the study.
To confirm the effectiveness of experimental manipulations and participant comprehension of the assigned cases, several manipulation checks were implemented. These checks indicated a strong understanding of the provided materials among participants. Specifically, participants affirmed their utilization of the provided AI audit planning tool for the analysis of the financial data within the context of the experimental study.
Main Findings
Table 2 presents a comparative analysis of participants’ perceptions regarding various facets of audit planning under two distinct experimental cases: Without AI (representing the application of traditional audit planning methodologies) and With AI (representing AI-assisted audit planning). The table presents, for each measured aspect of audit planning under both experimental cases, descriptive statistics, including the mean, standard deviation (SD), median, minimum, and maximum values. For every aspect of audit planning under consideration, both the mean and median scores exhibited a notable increase in the With AI case relative to the Without AI case. This consistent pattern provides convincing evidence suggesting a positive influence of AI adoption on participants’ perceptions of the audit planning process.
Descriptive Statistics and Results of the Wilcoxon Signed-Rank Test of H1 (n = 130).
Source. Created by the authors.
Note. This table presents the descriptive statistics and results of the Wilcoxon Signed-Rank test for H1. The test compared the median of the effectiveness of audit planning for all four measured dimensions in two conditions: without and with the use of AI.
The most substantial improvement was observed in the dimension of risk assessment, with the mean score increasing from 2.44 (SD = 0.92) in the “Without AI” case to 3.80 (SD = 0.80) in the “With AI” case, accompanied by a corresponding shift in the median score from 3 to 4. This finding indicates that AI is perceived as a particularly valuable tool for enhancing the identification and assessment of material risks during the planning phase of the audit. Furthermore, notable improvements were also evident in participants’ perceptions of time and resource efficiency, as indicated by the increase in the mean from 2.42 to 3.79, focus on key audit procedures (mean increasing from 2.55 to 3.83), and the clarity and understandability of the planning process (mean rising from 2.65 to 3.91).
The statistical significance of the observed differences between the “Without AI” and “With AI” cases was evaluated using Wilcoxon signed-rank tests, a non-parametric approach suitable for our paired data. As shown in Table 2, participants’ assessments of audit planning effectiveness varied significantly between the two experimental cases, with p-values consistently below .05. To provide a comprehensive interpretation of the findings, the effect size (r) was also calculated for each dimension. This measure indicates the practical significance of the results, a measure analogous to Cohen’s d or partial η2 for non-parametric tests. The calculated effect sizes for all four dimensions ranged from r = −.85 to r = −.87, which suggests a very large practical significance. The negative sign indicates the direction of the difference, showing that the scores for audit planning effectiveness were significantly higher in the “With AI” condition compared to the “Without AI” condition. This robust evidence supports H1, demonstrating that the integration of AI tools within the audit planning process leads to statistically and practically significant enhancements in key dimensions of audit planning, including risk assessment, resource efficiency, procedure prioritization, and clarity.
Our finding that AI adoption enhances audit planning effectiveness is consistent with the field evidence from developed markets presented by Kokina et al. (2025) and aligns with the performance potential of generative AI demonstrated by Eulerich et al. (2024). Furthermore, this finding aligns with the extant literature (Aitkazinov, 2023; Fedyk et al., 2022; Ganapathy, 2023; Onwubuariri et al., 2024) that highlights the positive impact of AI in various audit domains. This, in turn, frees resources for more complex analytical and judgmental activities and enables them to process and analyze larger volumes of data and identify complex patterns and anomalies that may not be readily discernible through traditional audit methodologies.
As detailed in Table 3, this study examined the perceived impact of AI on audit planning effectiveness, analyzing the moderating role of firm size. The research involved 65 participants in each subgroup (small firms, large firms, with AI, and without AI) and used a 5-point Likert scale to measure their perceptions. The overall findings indicate that all groups perceive a positive impact from AI on audit planning, with high mean and median scores. However, a key finding is that while both small and large firms saw benefits, larger firms perceived the improvements to be more significant. This suggests that larger firms, with their greater resources and complex audits, may be better equipped to leverage AI. For example, when it came to risk assessment, AI usage significantly improved perceived effectiveness in both firm types. Yet, the mean score for larger firms rose dramatically from 2.85 to 4.29, while smaller firms saw a more modest increase from 2.03 to 3.31. A similar trend was observed for the clarity of audit plans, where the mean score in large firms increased from 3.02 to 4.26, a more substantial jump than the increase from 2.29 to 3.55 seen in smaller firms.
Comparison of Audit Planning Aspects by Audit Firm Size and AI Usage (n = 65 Per Subgroup).
Source. Created by the authors.
Table 4 presents the findings of the Mann–Whitney U test, employed to investigate the influence of audit firm size (small versus large) on auditors’ perceptions of audit planning effectiveness, with the analysis conducted separately for cases with and without the utilization of AI. The p-values within the table (all are less than .001) reveal that statistically significant differences exist between small and large audit firms regarding their perceptions of planning effectiveness across all four assessed aspects, even in the absence of AI intervention, suggesting that firm size inherently impacts these perceptions. The effect sizes (r) for these comparisons ranged from −.45 to −.63, indicating large practical significance. This observation suggests that firm size moderates the impact of AI on perceived audit planning effectiveness, implying that larger audit firms may derive greater benefits from AI implementation in terms of enhancing their perception of planning effectiveness compared to smaller audit firms. These findings collectively provide support for the acceptance of H2.
Results of the Mann–Whitney U Test for H2 (Small vs. Large Audit Firm Size).
Source. Created by the authors.
This finding aligns with studies (e.g., Fernandez et al., 2024; Ganesan et al., 2024; Pham et al., 2025) that larger firms tend to exhibit superior audit planning practices, which can be attributed to their enhanced access to resources, specialized expertise, and advanced technological infrastructure. Our results extend this body of knowledge by demonstrating the moderating role of audit firm size on the relationship between auditor adoption of AI and audit planning effectiveness. Specifically, we find that the positive impact of AI on audit planning effectiveness is significantly more pronounced within larger audit firms compared to their smaller counterparts, thereby corroborating our hypothesized moderating effect.
Based on contingency theory, this result suggests that the effectiveness of AI is highly dependent on the organizational context in which it is implemented. Larger firms, particularly the Big 4, have the necessary financial resources, specialized personnel, and mature technological infrastructure to successfully implement AI tools. Additionally, the large and complex data from their client portfolios provides an ideal environment for AI to demonstrate its analytical strengths. In contrast, smaller firms’ resource constraints limit their ability to fully capitalize on AI’s potential, highlighting that the benefits of AI are not universal but are contingent on a firm’s strategic capabilities.
Our study provides strong empirical evidence supporting both contingency theory and the UTAUT, thereby enhancing the theoretical understanding of AI adoption in auditing. The significant moderating effect of audit firm size on the relationship between AI adoption and audit planning effectiveness, which was the focus of our second hypothesis (H2), directly supports contingency theory. This theory argues that a technology’s effectiveness depends on how well it aligns with specific organizational and contextual factors. Our findings show that although AI improves audit planning (H1), its positive impact is much more pronounced in larger audit firms. This suggests that the greater resources, specialized expertise, and more complex engagements found in large firms create an environment where AI tools can be used more effectively, leading to bigger improvements in planning outcomes compared to smaller firms (Kokina et al., 2025). This finding underscores the importance of considering the organizational context when new technologies are introduced in professional settings.
The strong evidence that auditors perceive AI as significantly enhancing audit planning effectiveness (H1, Table 2) provides a solid empirical foundation for the UTAUT. These results directly align with two of UTAUT’s central ideas: performance expectancy and effort expectancy. The observed improvements in risk identification, time efficiency, and the prioritization of procedures directly support the idea that auditors believe using AI will enhance their job performance. Furthermore, our experimental design, which included a training session on the AI tool, inadvertently acted as a facilitating condition, another key construct of the UTAUT. The positive results seen after this training suggest that providing sufficient resources, support, and education is vital for successful technology adoption. This reinforces the theoretical premise that a supportive environment is essential for effective technology integration and for auditors to view AI as a valuable tool that boosts their professional effectiveness.
Robustness Checks
To assess the robustness of our main findings, we conducted a sensitivity analysis. In research, this type of analysis evaluates whether the results remain stable when the conditions of the study are changed. For our study, we re-tested the hypotheses on a different sample: academic personnel from two Egyptian Universities. This new sample included faculty members and postgraduate students with expertise in accounting or finance (specifically those with graduate diplomas, MBAs, or Master’s/Doctoral degrees). We administered the same two experimental cases from our primary analysis—one with AI and one without—to this academic group. The data collection process for the sensitivity analysis was identical to the primary study. We hand-delivered the experimental materials, later retrieved them, and gave participants a chance to talk with the researchers and ask questions. This approach ensured that our initial findings were not limited to a single population. This included comparable instruction on the use of Microsoft Power BI Desktop to ensure a consistent baseline understanding for their participation in the AI intervention case, mirroring the training provided to the primary auditor sample. Participants assessed audit planning effectiveness within both cases. The sensitivity analysis results strongly aligned with the primary analysis findings, providing robust support for generalizing the initial conclusions beyond the primary sample of practicing auditors.
Moreover, H1 is retested using the overall effectiveness as a single composite score by averaging the Likert scale scores for the four items for each participant. Using Wilcoxon signed-rank tests, participants’ assessments of the effectiveness of audit planning varied significantly between the two experimental cases (p < .001; Z = −9.827), supporting H1. Similarly, the Mann–Whitney U test is employed to retest H2 to measure the overall effectiveness. Results of the test without the utilization of AI (p < .001; Z = −5.795), and with the utilization of AI (p < .001; Z = −7.625) both support the main findings.
Furthermore, Table 5 presents the results of a Multivariate Analysis of Covariance (MANCOVA) examining the interaction between audit firm size and AI usage (AI_usage × Au_firm_size) on four dependent variables (Item1, Item2, Item3, and Item4), representing aspects of audit planning effectiveness. The analysis employed Pillai’s Trace as the multivariate test statistic due to its robustness. The results reveal statistically significant main effects of both audit firm size (Au_firm_size) and the interaction term (AI_usage × Au_firm_size) across all four dependent variables (p < .001). Specifically, the F-statistics for Au_firm_size ranged from 58.249 to 70.960, with partial η2 values between .185 and .216, indicating medium to large effect sizes. The interaction effect demonstrated even stronger effects, with F-statistics ranging from 82.945 to 94.209 and partial η2 values between .244 and .268, indicating large effect sizes. Estimated marginal means show a consistent increase in the mean scores for all four items when AI is used (With AI) compared to when it is not (Without AI) (e.g., Item1: 3.481 vs. 2.758). Pairwise comparisons confirmed statistically significant mean differences (ranging from −.623 to −.731, all p < .001) between the With AI and Without AI cases for each item.
Summary of Multivariate and Univariate Test Results for AI Usage and Audit Firm Size.
Source. Created by the authors.
Note. This table presents the results of a MANCOVA investigating the effects of the interaction between audit firm size and AI usage (AI_usage × Au_firm_size) on audit planning effectiveness. The table includes Multivariate Tests using the test statistics Pillai’s Trace because of its robustness and reliability in multivariate testing. The p-value indicating values below .05 are typically considered statistically significant. Partial eta-squared (η2) as a measure of effect size, with values of .01, .06, and .14 representing small, medium, and large effects, respectively. There are two groups for audit firm size (small and large) with a sample size of 65 in each subgroup, the total sample size is 260.
As depicted in Table 5, the overall multivariate tests (Pillai’s Trace) revealed a significant overall effect of the interaction on the set of dependent variables (F = 26.778, p < .001, partial η2 = .297). These findings indicate that AI usage significantly improves audit planning effectiveness across multiple dimensions, with audit firm size acting as a key moderator. The significant interaction between AI usage and audit firm size suggests that the positive impact of AI varies according to firm size, underscoring the importance of AI integration in audit planning for enhanced outcomes.
Additional Analysis
Alternative Regression Models
Our study employed an additional analysis phase to investigate the core relationships further. This involved modifying the initial research model by introducing new variables to account for external influences. In our study, initially treated as a moderator, the audit firm size was then re-analyzed as a control variable. This re-examination aimed to determine if this variable has an independent influence on the assessment of audit planning effectiveness, separate from the influence of AI adoption. While demographic variables, such as auditor experience or education, are important and some were collected, the regression models presented below primarily focus on isolating the effects of AI usage and audit firm size on audit planning effectiveness, as per our core hypotheses.
As shown in Table 6, the regression analysis was repeated under different conditions to confirm the relationships under study further, as follows. The effect of AI usage as an independent variable on the assessment of audit planning effectiveness. Audit firm size was introduced as another independent variable in the model. The effect of both AI usage and audit firm size as two independent variables. This approach allowed for a direct comparison of the results obtained with audit firm size as a moderator and as a control variable, providing a deeper understanding of its impact on the overall findings.
Results of the Additional Regression Models (n = 260).
Source. Created by the authors.
Note. This table summarizes regression analyses conducted under three different models to further validate the relationships. AI_usage is the auditor AI adoption as a dummy variable (with vs. without), and Au_firm_size as a dummy variable (large vs. small audit firms). The dependent variable is the overall perceived audit planning effectiveness as a single composite score by averaging the Likert scale scores for the four items for each participant. Coefficient estimates are presented, with associated t-statistics provided parenthetically.
Statistical significance at the α = .05 level is denoted by an asterisk.
Table 6 summarizes three regression models analyzing the relationship between audit planning effectiveness (the dependent variable), auditor AI adoption (AI_usage), and audit firm size (Au_firm_size). All models demonstrate statistical significance (p < .05) as evidenced by the F-statistics. Model (1) reveals a positive association between AI usage and audit planning effectiveness. In this model, AI usage accounts for 44.1% of the variance in audit planning effectiveness, indicating a robust explanatory power after accounting for the number of predictors. Model (2) similarly indicates a positive relationship between audit firm size and audit planning effectiveness (p < .05, t = 7.330) with a positive coefficient (0.821). This model explains 16.9% of the variance in audit planning effectiveness, demonstrating that firm size alone has a notable independent influence. Finally, Model (3), incorporating both AI usage and audit firm size, demonstrates positive effects for both variables and a substantially higher adjusted R2 value (61.3%) compared to the individual models. Specifically, this comprehensive model accounts for 61.3% of the variance in audit planning effectiveness, highlighting a significantly greater explanatory power when both factors are considered simultaneously.
The Perceived Impact of AI Usage on Audit Planning Effectiveness
After exposure to both experimental cases, participants completed a questionnaire of four items designed to elicit their perceptions regarding the impact of AI on their planning process. Using Cronbach’s alpha (α) to assess internal consistency reliability, the computed coefficient was 0.82 which indicates an acceptable level of internal consistency, suggesting that the constituent items effectively measure the intended construct of participants’ perception of using AI in audit planning.
Table 7 presents the results of a one-sample Wilcoxon signed-rank test (n = 130), which was used to determine if the median perceived impact of AI on audit planning was significantly different from a neutral score. Participants rated four statements on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). The findings indicate a consistently positive perception, with all mean scores exceeding 4.09 and medians at 4 or 5. The statistical analysis provides strong evidence for this positive view, as all four items had significant p-values (p < .05) and negative test statistics. These results conclusively show that the median response for each statement is significantly above the neutral midpoint, leading to the conclusion that respondents perceive AI as positively influencing multiple aspects of audit planning, such as accuracy, risk identification, audit coverage, and time efficiency. Table 7 further validates UTAUT, as participants’ direct perceptions of AI’s benefits overwhelmingly provide even stronger evidence for the performance expectancy construct.
One-Sample Wilcoxon Signed Rank Test of the Perceived Impact of AI on Audit Planning Effectiveness (n = 130).
Source. Created by the authors.
Note. Z-statistics are the standardized test statistics for the Wilcoxon Signed-Rank test; a significant Z-statistic and a p-value less than .05 indicate a statistically significant perceived improvement in the audit planning effectiveness attribute for that item. Mean and Median values are based on a 5-point Likert scale, where 1 = Strongly Disagree and 5 = Strongly Agree.
Conclusion
Our study examined the impact of auditor adoption of AI technologies on the effectiveness of audit planning processes, with a specific focus on the moderating role of audit firm size. The findings from our study confirmed both of our hypotheses. First, we found robust support for H1, demonstrating a statistically significant positive association between auditor adoption of AI technologies and the effectiveness of audit planning. This positive impact was observed across multiple dimensions, including enhanced accuracy, improved risk identification, and greater time efficiency. Second, our results also support H2, indicating that audit firm size significantly moderates this relationship. The positive effects of AI were found to be notably more pronounced within larger audit firms compared to smaller counterparts, thereby corroborating our hypothesized moderating impact. The robustness and generalizability of these primary findings were subsequently corroborated through the implementation of a rigorous sensitivity analysis.
This study contributes to the literature on audit planning and AI integration in auditing by providing evidence from a developing economy—the Egyptian audit market. Furthermore, this study investigates how audit firm size moderates this relationship, providing insights into the varying effects of AI adoption across different organizational contexts. Methodologically, we employed a 2 × 2 factorial experimental design conducted within actual audit firms, incorporating a sensitivity analysis with academics to enhance the validity and generalizability of the findings. By doing so, this study develops a theoretically grounded model linking AI adoption, firm size, and audit planning effectiveness in the Egyptian context, enabling a deeper understanding of these relationships and facilitating cross-national comparisons.
Our findings highlight the significant implications of AI for the auditing profession. At the professional level, it emphasizes the growing importance of AI and the need for adaptation, noting its potential to improve audit quality and efficiency while raising implementation, governance, and oversight concerns. Regulators are urged to develop frameworks for AI use in auditing, including standards for implementation, auditor competency, data privacy, and market competition. Auditors must acquire new AI-related skills, including understanding AI tools, interpreting their outputs, and integrating them into audit methodologies, requiring a shift towards analytical reasoning and professional judgment. For smaller audit firms, it might benefit from starting with specific, targeted AI applications like those for data extraction or anomaly detection, or by collaborating to pool resources for AI infrastructure. It is also important to invest in targeted training programs that are scalable for firms of all sizes. Finally, accounting and auditing educators should integrate AI topics into the auditing curricula, addressing its benefits, limitations, ethical considerations, and practical implications.
However, our study is subject to some limitations. Firstly, while purposive sampling allowed us to target auditors with specific expertise, it inherently limits the ability to generalize the results to the entire population of auditors in Egypt or to other national contexts. Furthermore, the experimental design utilized a specific, single AI tool (Microsoft Power BI Desktop) for operationalization. While this approach enabled controlled manipulation, it means the generalizability of our findings regarding AI’s impact is tied to the capabilities and functionalities of this specific platform. Secondly, while the sensitivity analysis using academic personnel provides some corroborative evidence, it is important to recognize the potential divergence between academic perceptions and the realities of professional practice. Thirdly, while the study was conducted in real audit firms, the use of hypothetical audit cases, though necessary for experimental control, inherently limits the direct ecological validity compared to real-world, complex audit engagements. Fourthly, although participants received training on Microsoft Power BI Desktop, individual variations in prior familiarity with AI tools and overall auditor experience were not explicitly included as controls in our main analyses. Moreover, the sample was restricted to auditors in private accounting and auditing firms, which precluded direct inferences regarding the applicability to the governmental sector. Finally, while measuring audit firm size as a dichotomous variable was necessary for our 2 × 2 experimental design, and it is a widely used proxy in the auditing literature, it limits our ability to capture the full spectrum of firm-size effects.
Based on your study’s findings and limitations, future research should focus on several key areas. To begin, it would be beneficial to replicate this research with a larger and more diverse auditor population, including cross-national comparisons, and to use longitudinal or field-based methodologies to study AI that is already integrated into real-world audit processes. Future studies should also investigate the impact of other AI platforms or simulate a broader range of AI capabilities to enhance the generalizability of the findings. More research is needed to understand how AI tools affect auditor judgment and decision-making, particularly the potential for bias and over-reliance. It would also be valuable to explicitly account for factors like auditors’ prior experience with AI and their overall professional experience, possibly by including them as control variables. Finally, future research could explore the perspectives of other stakeholders—such as regulators, clients, or internal auditors—and use more refined, continuous proxies for firm size (like annual revenue, number of employees, or client portfolio size) to gain a more nuanced understanding of its moderating role, while also examining the crucial ethical implications of AI in auditing, including data privacy and accountability.
Footnotes
Appendix: The Experimental Package
Acknowledgements
The authors would like to thank Alexandria University and Prince Sultan University for their support.
Ethical Considerations
The study was conducted in full compliance with internationally recognized ethical standards, including the principles outlined in the 1964 Helsinki Declaration and its subsequent amendments, as well as the regulations of the relevant institutional and/or national research ethics committees. In addition, all procedures strictly followed the ethical guidelines of the Faculty of Business, Alexandria University. Formal approval from the institutional research ethics committee was obtained on April 15, 2024, prior to the commencement of data collection.
Consent to Participate
All participants were fully informed of the study’s objectives, procedures, and their rights, including the right to withdraw at any stage without penalty. Written informed consent was obtained from each participant prior to their inclusion in the study. All data were anonymized and handled with the utmost confidentiality, and were used solely for the purposes outlined in the approved research protocol by the institutional ethics committee.
Author Contributions
Conceptualization, Safaa Saleh and Osama Abouelela; Methodology, Safaa Saleh and Osama Abouelela; Project administration, Safaa Saleh and Ahmed Diab; Writing—original draft, Safaa Saleh and Osama Abouelela; Writing—review and editing, Ahmed Diab.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author.