The research utilized a quantitative survey methodology to gather data on using AI technologies in tax compliance and planning, this approach allows for the collection of structured and measurable data, facilitating the analysis of trends, perceptions, and experiences related to AI-powered tax software ( Lim, 2024 ).

A structured questionnaire was developed using Google Forms, comprising multiple-choice questions, 5-Likert scale statements, and Yes-No questions ( Darnton, 2023 ). The survey is divided into six sections: a cover letter, demographic information, awareness and usage of AI in tax compliance, perceived benefits of AI in tax compliance and planning, challenges and concerns, and future perspectives.

This study employed a purposeful sampling technique. The survey targeted a diverse audience, including students, employed individuals in tax-related and non-tax-related fields, self-employed, and retirees. This broad sampling strategy is designed to gather perspectives from a wide range of respondents, ensuring the generalizability of the findings and the increase in sample size.

The survey was distributed electronically via email, WhatsApp, and LinkedIn to reach 980 broad audience. Participants were invited to complete the survey voluntarily. Electronic distribution ensures efficient data collection and the ability to reach respondents across different geographic locations using a cross-sectional descriptive study design. Quantitative survey responses were collected at one time point only, and 109 valid respondents were captured.

Quantitative data collected from the survey was analyzed using Excel and google forms statistical software. The Likert scale responses to measure the level of agreement or disagreement were summarized, and the survey data were coded into numerical values, ranging from 1 to 5 for the Likert scale statements and 1 to 6 for the age interval variable. Descriptive statistics, such as means, frequencies, and percentages, were used to summarize the data and check the normality and reliability of the data, and inferential statistics tests were utilized to test the twelve hypotheses and to identify common themes and insights. Relative importance index (RII) technique was utilized to rank the results based on the perception of AI usage in tax-driven solutions from the highest to the lowest influence to answer the research question RQ5 ( Ammoury, 2023 ). A correlation test was utilized to verify the correlation between age and future implementation of AI-driven tax solutions and test H12₀ ( Lu & Yuan, 2024 ). A two-tailed one-sample t-test was conducted to identify the existence of a relation between the eleven identified independent variables and the AI in tax compliance and planning ( Kronthaler, 2024 ), to test H1₀, H2₀, H3₀, H4₀, H5₀, H6₀, H7₀, H8₀, H9₀, H10₀ and H11₀. The nineteen variables in the dataset are the were used for descriptive analysis and to test the hypotheses and the conceptual diagram: Independent Variable (IV), Dependent Variable (DV).

Participation in the survey was voluntary, and respondents were assured of the confidentiality and anonymity of their responses. Informed consent was obtained by reading the cover letter and proceeding to the next section of the survey. Participants were also informed about the purpose of the research, the use of the data collected, and their right to withdraw from the study at any time.

The study acknowledges potential limitations, such as self-selection bias, as participants more interested in AI or tax compliance might be more likely to respond. Additionally, the sample size is small of 109 respondents. Lastly, the rationale for including business and finance major’ students is justified because tax professionals would have a background of business and finance major, but this mixed sample in a study on professional tax practice might affect the results’ generalizability and hence another limitation of the study.

The survey received responses from a diverse group of demographic participants as follows. Figure 2 shows that the age is diverse with the majority of respondents being under 25 (38.5 percent), between 25 - 34 years old (10.1 percent), 35 - 44 years old (13.8 percent), a higher proportion of respondents are between 45 - 54 years old (26.6 percent), which indicates that the respondents are relatively young. Figure 3 shows that most participants have a bachelor’s degree (44 percent) and additionally a master’s degree (32.1 percent), a significant number of respondents held a Doctorate (6.5 percent), while a few had only completed high school or currently enrolled in university (11 percent), which indicates that most of the respondents have a good education level. Figure 4 shows that majority of the participants included students (33.9 percent), self-employed individuals (10.1 percent), retired (7.3 percent), and employed individuals both in tax-related (15.6 percent) and non-tax-related fields (32.1 percent), which indicates that most respondents are working professionals and 15.6 percent are specifically working in tax related fields. Figure 5 shows that respondents have varied levels of experience in their fields, with the largest group having more than 10 years of experience (51.4 percent), followed by those with less than 1 year (24.8 percent), 1 - 3 years (14.7 percent), which indicates that the majority have good work experience. Figure 6 shows that the familiarity with AI technologies among respondents varied, only 13.8 percent of respondents reported being very familiar with AI, about 33.9 percent were somewhat familiar, while 32 percent had moderate familiarity, 14.7 percent were slightly familiar, and 8.3 percent were not familiar at all, which indicates that most respondents have moderate to somewhat familiarity about AI in general. Figure 7 shows that the majority (67 percent) are not aware of AI technologies in tax compliance, 33 percent of respondents were aware of AI technologies being used in tax compliance and planning. Figure 8 shows that the usage of AI-powered tax software among respondents was limited to only 2.8 percent, indicating a significant opportunity for growth in this area. This means that among the seventeen tax related professionals (15.6 percent, Figure 4 ), three of them use AI power tax software. Figure 9 indicates that most respondents agree to the perceived benefits of AI in tax, that AI tax software simplifies the tax preparation process, helps in accurate filing tax returns, reduces the time spent on tax compliance, identifies tax savings opportunities and provides better compliance with changing tax regulations. Figure 10 shows classifies the most perceived benefits of AI in tax compliance among respondents: time savings in tax preparation, filing, and reporting (79.8 percent) was the highest chosen benefit, better improved data analysis and reporting (48.6 percent), improved accuracy in tax calculations and compliance (40.4 percent), cost reduction (39.4 percent), enhanced ability to identify tax-saving opportunities (27.5 percent), enhanced compliance with regulations (26.6 percent). Figure 11 shows the participants’ agreement to the highlighted challenges and concerns related to AI in tax compliance: data privacy and security concerns were the most agreed-to, followed by the perception that AI cannot fully replace the expertise of human tax professionals and then by lack of trust in AI, which indicates the resistance to change to AI in tax. Figure 12 shows the classification of the most important features that should be present in future AI-tax software, the majority of respondents (60.6 percent) think that AI-powered tax software should have regular updates on tax regulations, (57.8 percent) think it should have more robust data security, (55 percent) think that it should have better integration with financial software, (50.5 percent) think it should have more personalized tax savings recommendations, (47.7 percent) think it should have improved user interface and (42.2 percent) think it should have enhanced customer experience. Figure 13 shows that most respondents (37.6 percent) think that it is very likely that AI-powered tax software will be used in the future, and (36.7 percent) think it is likely, which means that the majority (74.3 percent) think that AI-powered tax software will be used in the future. Figure 14 shows that despite the resistance to change to AI, most respondents (48.6 percent) believe that AI would be a useful tool but not essential, followed by (40.4 percent) who thought that AI will become essential for all tax compliance and planning tasks, 11 percent were skeptical about the widespread adoption of AI in this field.

The survey was shared with two professors for validity prior to distribution. For the reliability of the datasets, Cronbach’s alpha was calculated and found to be 0.6, therefore the data is questionable ( Tavakol & Dennick, 2011 ).

Table 1 shows the calculated means, median, variance, standard deviation, range, minimum, maximum and sum for the fourteen variables (related to fourteen statements), the results are sorted in the order of agreement of the respondents from highest agreement to lowest agreement with the fourteen statements (corresponding variables). Notably, the results show that most respondents found that Time reduction (M = 4.22, SD = 0.77) is the highest effect of AI usage in tax-driven solutions, followed by Simplicity (M = 3.92, SD = 0.76), Compliance with changing tax regulations (M = 3.92, SD = 0.82) and Security and privacy issues (M = 3.92, SD = 0.95), all three with equal mean in second place and the least effect of AI using in tax-driven solutions is Difficulty to use (M = 2.88, SD = 0.89). Familiarity with AI in tax and Usage of AI in tax were yes/no statements with 2 responses; therefore, they will not be considered in future analysis.

Table 1 also shows that the means values of the six variables are very close to the median values, in addition the Kurtosis is between −7 and +7 for all variables except for the variable “usage of AI in tax”, and the skewness is between −2 and +2 for all variables, except for the variable “usage of AI in tax”, this means that the data follows a normal distribution or all thirteen other variables, so we can proceed with parametric tests for these variables.

To answer to the research question RQ5 “What is the rank and level of importance of the results of AI usage in tax-driven solutions in order of importance”, the researcher conducted an analysis to rank the results based on the perception of AI usage in tax-driven solutions from the highest to the lowest influence. Therefore, the most important factor was ranked “1” and the least important factor was ranked “10”. RII was calculated to determine the weighted value of each of the eleven variables that affect the use of AI in tax-driven solutions. It was computed by applying the formula described by Aziz et al. (2016):

$\left(\text{RII}\right)=\sum \text{W}/\left(\text{A}\times \text{T}\right)$

W = the weight given by every respondent to the specific variable, ranging from 1 to 5.

A = the maximum possible weight, which is 5.

T = the total number of acquired valid questionnaire responses, which is 109.

The value of RII ranges between zero and one. The higher the RII value, the more the variable is important in influencing results of AI usage in tax-driven solutions in order of importance.

Table 2 shows that the variable Time reduction has the highest RII value of 0.84, which is interpreted as very important ( Aziz et al., 2016 ) and therefore ranked as 1; this indicates that time reduction is the most perceived benefit of using AI in tax. The second most important variables, all with the same RII value of 0.78, were simplicity, compliance with changing tax regulations, and security and privacy issues, and all three were ranked second. The variable Tax-saving opportunities and strategies ranked fifth in order of importance, with an RII value of 0.779, and the variable Replacing human expertise ranked Sixth with an RII score of 0.75. Notably, all these six variables had a rounded RII value between 0.75 and 0.89 ( Aziz et al., 2016 ), indicating that they should all be interpreted as very important. Accuracy ranked seventh with RII score of 0.74, followed by the variable Lack of Trust with RII score of 0.68 ranking eighth, then High cost ranked nineth with RII score of 0.67, these last three variables are somewhat important ( Aziz et al., 2016 ). The tenth-ranked variable with the importance of neutral was Difficulty to use with RII 0.58.

The correlation test was conducted to test the below hypotheses:

H12_a: Younger generations are more receptive to the future implementation of AI-driven tax solutions, suggesting a demographic trend in AI technology adoption (alternate hypothesis).

H12₀: There is no demographic trend related to age in AI technology adoption (null hypothesis).

Table 3 shows that Pearson correlation coefficient is r (107) = 0.07, which indicates a very weak to non-existing correlation. Therefore, there is no correlation between age and future implementation of AI-driven tax solutions. So, H12_a will be rejected and H12₀ will be accepted ( Table 4 ).

A two-tailed one-sample t-test was conducted to identify the existence of relation between the eleven identified independent variables and the AI in tax compliance and planning by testing the mean of the eleven independent variables with the neutrality value of the scale μ = 3 because the scale is 1 to 5 in the 5-point Likert scale. Table 5 shows the results of the one-sample t-test; it indicates that the mean scores for ten variables are higher than the midpoint of 3, but for the variable difficulty to use, it is less than 3. The results of the one sample t-test is t (109) > 1.98, p < 0.05, is statistically significant for the following ten variables, sorted by the highest mean to the lowest: Time reduction (M = 4.22, SD = 0.77) which had the highest RII and ranking (refer to Table 2 ), Simplicity, Compliance with changing tax regulations, Tax-saving opportunities and strategies, Security and privacy issues, Accuracy, Replacing human expertise, Future implementation of AI in tax, High cost and Lack of trust. Therefore, the corresponding null hypotheses are rejected H1₀, H2₀, H3₀, H4₀, H5₀, H6₀, H7₀, H8₀, H9₀ and H10₀ and the corresponding alternative hypotheses are accepted ( Table 4 ). For the variable Difficulty to use (M = 2.88, SD = 0.89), M < μ and t (109) < 1.98, p < 0.05, is not statistically significant, therefore, H11₀ was accepted ( Table 4 ).

Below are the corresponding eleven null hypotheses that were tested:

H1_a: AI technologies reduce the time required for tax compliance.

H1₀: AI technologies don’t reduce the time required for tax compliance.

H2_a: AI increases the accuracy in filing tax returns.

H2₀: AI doesn’t increase the accuracy in filing tax returns.

H3_a: AI tax software simplifies the tax preparation process.

H3₀: AI tax software doesn’t simplify the tax preparation process.

H4_a: AI technologies create concerns regarding security and privacy issues in tax compliance and planning processes than traditional methods.

H4₀: AI technologies don’t create concerns regarding security and privacy issues in tax compliance and planning processes than traditional methods.

H5_a: AI technologies can replace the expertise of human tax professionals.

H5₀: AI technologies cannot replace the expertise of human tax professionals.

H6_a: There is a lack of trust in AI’s recommendations in tax compliance and planning.

H6₀: There isn’t a lack of trust in AI’s recommendations in tax compliance and planning.

H7_a: AI technologies are too expensive to be used in tax compliance and planning processes than traditional methods.

H7₀: AI technologies aren’t too expensive to be used in tax compliance and planning processes than traditional methods.

H8_a: AI-powered tax software enhances the identification of tax-saving opportunities, leading to optimized tax strategies.

H8₀: AI-powered tax software doesn’t enhance the identification of tax-saving opportunities, leading to optimized tax strategies.

H9_a: AI-powered tax software provides better compliance with changing tax regulations.

H9₀: AI-powered tax software doesn’t provide better compliance with changing tax regulations.

H10_a: AI-powered tax software will be widely used in the future.

H10₀: AI-powered tax software won’t be widely used in the future.

H11_a: AI-powered tax software is difficult to use.

H11₀: AI-powered tax software is not difficult to use.

H12₀: Younger generations aren’t more receptive to the future implementation of AI-driven tax solutions, there is no evidence of a demographic trend in AI technology adoption.

This research contributes significantly to the academic discourse on AI in tax compliance and planning. It underscores the transformative potential of Artificial Intelligence (AI) in tax compliance and planning, with significant advantages in time savings, accuracy, and regulatory compliance. It also identifies challenges, particularly trust, security, and privacy concerns, while confirming that demographic factors, such as age, do not influence AI adoption. By ranking the effects of AI adoption in terms of importance, this research offers a structured perspective, providing a roadmap for prioritizing benefits and addressing limitations.

While this study provides meaningful contributions, a holistic and multidisciplinary approach is needed to address the evolving role of AI in tax compliance and strategy optimization. Future research should explore the following dimensions: Investigate how AI-powered tax solutions can be seamlessly integrated with Enterprise Resource Planning (ERP) systems and other financial tools, enabling comprehensive financial management and alignment with organizational objectives; Evaluate the usability and accessibility of AI tax systems, focusing on user experience across diverse demographics. Research should identify the training needs of tax professionals to enhance their engagement and trust in AI-based recommendations; Examine the customization and scalability of AI tax systems to meet the unique needs of businesses of various sizes and sectors, including multinational organizations with complex regulatory frameworks; Conduct in-depth studies comparing AI-powered tax solutions to traditional human expertise. Such comparisons should focus on efficiency, accuracy, cost-effectiveness, and areas where hybrid approaches may yield optimal results; Explore how AI adoption in tax compliance influences taxpayer behavior, compliance attitudes, and perceptions of fairness. Ethical dimensions, including accountability, data privacy, and equity, must also be a central focus of future investigations; Research the development of adaptive AI algorithms capable of responding to rapidly changing tax laws and economic conditions, ensuring continued compliance and proactive identification of tax risks and opportunities; Documenting detailed case studies of organizations successfully implementing AI in tax systems will provide actionable insights, foster public trust, and illustrate best practices for adoption across diverse industries.

The findings of this study highlight the need for collaborative efforts among researchers, policymakers, and industry leaders to harness AI’s full potential in tax compliance and planning. Future research should bridge the gap between technological innovation and real-world application, ensuring that AI solutions are efficient, secure, and ethically sound. By addressing these challenges and opportunities, AI can transform tax systems into more transparent, equitable, and dynamic tools that support the evolving needs of modern economies.