Perceptions of South African Accountants on Factors with a Role in the Adoption of Artificial Intelligence in Financial Reporting

Abstract

Purpose—The objective of this study was to conduct a detailed South African study that sought to explore and analyse the views of South African accountants regarding the factors that affect the adoption of AI in financial reporting. In other words, this study aimed to understand what accountants in South Africa think about the use of AI in their field, especially concerning its integration into financial reporting practices. Three main theories underpinned the study, namely, the diffusion of innovation, technology, organisation, and environment framework, and the institutional theory. In essence, the study sought to determine the perception of South Africa’s accountants on these factors. Design/methodology/approach—This study adopted the quantitative research method and descriptive design. In this regard, positivism as a philosophy was preferred. An online survey was developed to collect information from the participants. Participants were recruited based on their affiliation with the four IFAC-recognised accounting bodies in South Africa: SAICA, SAIPA, CIMA, and ACCA. Findings—Th study found that, overall, South African accountants believe that organisational, technological, and environmental factors play a role in adopting artificial intelligence in financial reporting. Originality/value: This study contributes by enriching the understanding of South African accountants’ perceptions of the adoption of artificial intelligence in financial reporting through the lenses of the selected theories.

1. Introduction

Russell and Norvig (2016) and Metz (2016) indicate that artificial intelligence’s (AI’s) theoretical and technological foundation was developed in the 1950s and has been evolving since then. However, although the academic literature on AI has evolved in the last six decades (Buchanan 2005), few studies demonstrate the engagement with AI in the Southern Hemisphere, particularly in the South African context.

According to Kommunuri (2022), there are a handful of academic conversations on using AI in accounting, particularly the published work in peer-reviewed journals. In Kommunuri’s (2022) review of the literature in New Zealand, it was found that there were few articles published on the AI topic specifically in business and professional accounting magazines. In France, a qualitative study that focused on auditing and digitalisation found that it had the potential to improve audit relevance and audit quality mainly by analysing all customers’ data (Manitaa et al. 2020).

South Africa has seen several studies emerging on AI. However, none of these studies have focused on financial reporting. Some of these studies include Qwabaza (2022) who used an online survey to explore the success factors for the adoption of artificial intelligence by South African financial services companies.

Further, through a combination of diffusion of innovation (DOI), institutional theory, and technology, organisation, and environment (TOE) frameworks, Rao (2017) sought to determine the factors that are critical to the adoption of AI by South African organisations. Using the TOE framework, Mariemuthu (2019) investigated factors influencing AI adoption by South African banks. Finally, in the African context, Mbizi et al. (2022) sought to determine the skills required by African accountants in the Fourth Industrial Revolution.

A study on the perception of accountants regarding the adoption of financial reporting is important as accountants collect, prepare, analyse, and present data to the relevant platforms. The data collected, prepared, analysed, and presented culminates in financial reports, which are audited and consumed by stakeholders for decision making.

Given the importance of financial reports, the International Accounting Standards Board (IASB) promulgated the qualitative characteristics of financial reporting. Accordingly, the qualitative characteristics (enhancing) increase financial reports’ decision usefulness and stewardship when the fundamental characteristics have been established.1 According to the IASB (2018)2, there are two sets of qualitative characteristics of financial reports, namely, the fundamental characteristics, which include relevance and faithful representation, as well as the enhancing characteristics, which include comparability, verifiability, timeliness, and understandability 3.

The promulgation of the qualitative characteristics of financial reports demonstrates how important the financial reports are in the life of each organization. It also makes the role of accountants important within an organization as they are the main implementers of the qualitative characteristics. Understandably, the IASB will emphasize the importance of the qualitative characteristics of financial reporting given the corporate failures such as Enron, AOL, Tyco, Aldephia Communications, and Global Crossing in the early 2000s (Naidoo 2002; Moloi 2008) and Tongaat Hullet and Steinhoff (Day 2020; Edmans 2020; Rossouw and Styan 2019), all recently stemming from the manipulation of financial reports.

It is clear in the brief assessment of the existing literature above that no study has been undertaken or has even come close to determining the key attributes that drive the adoption of artificial intelligence tools in financial reporting in the South African context.

2. Literature Review and Theoretical Framework

According to Kok et al. (2009), the exact definition of AI is a topic of considerable discussion, with more definitions speaking of AI as “imitating intelligent human behavior” (Kok et al. 2009). In this regard, Marwala (2007, 2009) indicates that AI can be viewed as a “technique that is used to make computers intelligent”. However, how do we measure if computers are intelligent? It is when computers can conduct the tasks that would ordinarily be reserved to be conducted by human beings. In this regard, the measure will be a human being. Essentially, AI can be viewed as the art of humanising machines.

In the recent past, studies in AI tended to focus on areas such as communication, technology, engineering, and manufacturing (Haugeland 1985; Chu and Wang 1988; Connolly 2008; Russell and Norvig 2016).

Large international accounting conglomerates such as PriceWaterhouseCoopers have also argued that AI subsystems such as assisted intelligence, augmented intelligence, and autonomous intelligence have the propensity to solve some of the world’s most complex challenges (Price Waterhouse Coopers 2017). An example of the propensity to solve some of the world’s most complex challenges is provided by van Bommel and Blanchard (2017). van Bommel and Blanchard (2017) argue that big organisations such as banks that adopt and harness AI technology are poised to benefit from faster digitization. They are also poised to offer customers omnichannel, customer-centric products and services timeously compared to those that do not. As a result, AI is now widely considered a “general-purpose technology” (Mantas 2019).

AI has moved from purely being a technical domain, and large conglomerates and business leaders are adopting AI to derive efficiency. Manitaa et al. (2020), for example, found that it had the potential to improve audit relevance and audit quality. Accordingly, this could be achieved mainly by analysing all data rather than sampling. Even with this finding, it is essential to indicate that we still do not know much about the utilisation of AI in general accounting functions and, more so, in financial reporting.

According to Khanzode and Sarode (2020), the main advantages of deploying AI include the fact that it expedites tasks, it completes complex tasks easily, it can multitask, its success ratio is high, it results have few errors and defects, it results in more efficiency in a short time, it can calculate long-term and complex situations faster, and it is good for discovering unexplored things. Given these advantages and the fact that the financial reporting process is complex, tedious, and risky, it is concerning that we do not know much about the prevalence (adoption and practice) of AI in accounting, particularly in financial reporting. Unlike in the other fields, where the role of AI is understood, there are no studies on the key attributes essential in adopting AI in financial reporting.

With regard to the theoretical framework, various theories of technology adoption exist. The theories focused on adopting and utilising technologies at an organizational level include the TOE developed in the 1990s by Tornatzky (Tornatzky et al. 1990). In addition, these theories include the DOI (Rogers 2003) and the institutional theory (DiMaggio and Powell 1983). For this study, the factors underpinning these theories were consolidated to form harmonised factors of the TOE, DOI, and institutional theory. This is because, looking at theory in isolation, these factors are not comprehensive. Table 1 below shows the harmonised factors.

Table 1. Consolidated factors with a role in the organization’s ability to adopt an innovation or technology.

Diffusion of Innovation Theory (Rogers 2003)	Variable in the Framework	Technology Organization and Environment Framework (Tornatzky et al. 1990)	Harmonized Factors	Denotation	Output
Organizational characteristics	Organization	Equivalence of Organizational Characteristics in the TOE	Organizational Factors	O	Determination of the key attributes that drive the adoption of artificial intelligence tools in financial reporting in the South African context.
Centralization		No equivalence	Centralization
Complexity		No equivalence	Complexity
Formalization		No equivalence	Formalization
Interconnectedness		No equivalence	Interconnectedness
Organizational slack		No equivalence	Organizational slack
No equivalence		Organizational innovativeness	Organizational innovativeness
No equivalence		Top management support	Top management support
Innovation characteristics	Technology	Equivalence of Innovation Characteristics in the TOE	Technology Factors	T
Compatibility		Technical integration	Technical integration
Complexity		Technical complexity	Technical complexity
No equivalence		Technical readiness	Technical readiness
Institutional Theory (DiMaggio and Powell 1983)	Environment	Equivalent in the TOE framework	Environment Factors	E
Mimetic pressure		Competition intensity	Mimetic pressure
Coercive pressure		No equivalence	Coercive pressure
Normative pressure		No equivalence	Normative pressure

Source: Author’s conceptualization.

As such, questions were conceptualised in such a way that they reflected the organisational factors (centralisation, complexity, formalisation, interconnectedness, organisational slack, organizational innovativeness, and top management support), technological factors (technical integration, technical complexity, and technical readiness), and environmental factors (mimetic pressure, coercive pressure, and normative pressure).

3. Methodology

Participants were recruited based on their affiliation with the four main accounting bodies in South Africa, namely, the South African Institute of Chartered Accountants (SAICA), the South African Institute of Professional Accountants (SAIPA), the Chartered Institute of Management Accountants (CIMA), and the Association of Certified Chartered Accountants (ACCA). These professional bodies operate in South Africa and are recognized by the International Federation of Accountants (IFAC). The total population of accountants in South Africa is not known. As of March 2022, SAICA had close to 45,000 members (SAICA 2022). SAIPA had close to 15,000 members (SAIPA 2022). CIMA and ACCA are British accounting bodies with a presence and recognition in South Africa. Both these accounting bodies have offices in South Africa that advocate for their members’ interests in the country. Globally, ACCA has more than 233,000 members (ACCA 2022) and CIMA has a membership of more than 150,000 globally and represents more than 3000 members in South Africa (CIMA 2022).

An online survey was developed to collect information from the participants. Participants were recruited based on their affiliation with the four main accounting bodies in South Africa: SAICA, SAIPA, CIMA, and ACCA. Essentially, this study follows a deductive approach with a nonexperimental research design. A convenient sampling method, which is a method that selects participants based on their accessibility and availability to the researcher, was followed.

Initially, a five-point Likert scale was scaled using disagree, strongly disagree, neutral, agree, and strongly disagree to gauge the views of South African registered accountants. For dimension reduction in the presentation of the results, the scales were aggregated to disagree/strongly disagree, neutral (no aggregation here), and agree/strongly disagree. Morgan et al. (2017) support the reduction in dimension to reporting data.

To encourage accountants in South Africa to complete the questionnaire, a process had to be developed. The process followed was as follows:

• Search for Associate of Certified Chartered Accountants/ACCA, Associate Chartered Management Accountant/ACMA, Chartered Global Management Accountant/CGMA, Chartered Accountants South Africa/CA(SA), and Professional Accountant (SA)/PA(SA).
• Invite the members to be a connection on LinkedIn. The total number of invitations sent was 415. A total of 200 accountants accepted the invitation to be a connection on LinkedIn.
• For those who accepted to be a connection on LinkedIn, an invitation letter together with the letter of informed consent inviting them to participate in the study was sent. Once they consented to participate, the online survey was shared with them.
• A total of 172 accountants completed the survey which was sent to them via a link.

Following the data cleaning, it was clear that 22 respondents did not complete the online survey in full. These respondents missed some of the questions on the online survey. After the cleanup of the responses, only 150 were deemed to be valid. As such, the valid response rate for this study was 87.2%.

4. Findings

The following sections illustrate and discuss the findings of the study.

Results

Table 2 below depicts the constructed custom table for organisational factors using the aggregated scale of disagree/strongly disagree, neutral (no aggregation here), and agree/strongly disagree.

Table 2. Organisational factors.

Organisational Factors		Disagree/Strongly Disagree	Neutral	Agree/Strongly Agree	Mean	SD
OF1	Count	16	25	108	3.95	1.141
	Row n%	10.7	16.8	72.5
OF2	Count	15	8	126	4.23	1.123
	Row n%	10.0	5.4	84.5
OF3	Count	10	19	120	4.11	1.001
	Row n%	6.7	12.8	80.5
OF4	Count	16	19	111	3.91	1.095
	Row n%	11.0	13.0	76.0
OF5	Count	18	11	119	4.12	1.130
	Row n%	12.1	7.4	80.4
OF6	Count	13	12	120	4.28	1.135
	Row n%	9.0	8.2	82.8
OF7	Count	16	20	113	3.97	1.099
	Row n%	10.8	13.4	75.8

Note: OF refers to organisational factor—see Table A1 of Appendix A for OF descriptions.

OF1 represents the organisational factor which tested the statement that the centralisation of decision making in the organisation affects the acquisition of AI technology or was a determinant in consideration of the acquisition of AI tools for financial reporting in an organisation. A total of 72.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 16.8% of respondents indicated that they were neutral and 10.7% indicated that they strongly disagreed/disagreed with the statement. The mean for this organisational factor was 3.95 and the standard deviation was 1.141. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

OF2 sought to gauge the respondents’ views on the statement, which sought to test whether the availability of more individuals with complex sets of skills in the organization (e.g., understanding newer technologies and their abilities) would make it easier to sell the benefits of acquiring AI tools in the financial reporting process/made it easier to sell the benefits of AI tools for the financial reporting process. The findings are that 84.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 5.5% of respondents indicated that they were neutral and 10.0% indicated that they strongly disagreed/disagreed with the statement. The mean for OF2 was 4.23 and the standard deviation was 1.123. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

OF3 sought to gauge the respondents’ views on the statement which sought to test whether the finance functions (accountant’s) depth and pervasiveness of interpersonal networks within an organization, which enables the flow of information, would make it easier to sell the benefits of acquiring AI tools for the financial reporting process/made it easier to sell the benefits of AI tools for the financial reporting process. The findings point to the fact that 80.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 12.8% of respondents indicated that they were neutral and 6.7% indicated that they strongly disagreed/disagreed with the statement. The mean for OF3 was 4.11 and the standard deviation was 1.001. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

OF4 sought to gauge the respondents’ views on the statement which sought to test whether the finance functions (accountant’s) ease of access to uncommitted resources which could then be utilized to acquire things such as newer technologies would affect the decision of whether to acquire AI tools for the financial reporting process/was a determinant in the consideration for adopting AI in the financial reporting process. The findings of this statement indicate that 76% of surveyed accountants strongly agreed/agreed with this statement, whereas 13% of respondents indicated that they were neutral and 11% indicated that they strongly disagreed/disagreed with the statement. The mean for OF4 was 3.91 and the standard deviation was 1.1095. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

OF5 sought to gauge the respondents’ views on the statement which sought to test whether the propensity of the organization to push for innovation in its function/operations would make it easier to sell the benefits of acquiring AI tools for the financial reporting process/made it easier to sell the benefits of acquiring AI tools for the financial reporting process. The study found that 80.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 7.4% of respondents indicated that they were neutral and 12.1% indicated that they strongly disagreed/disagreed with the statement. The mean for OF5 was 4.12 and the standard deviation was 1.1130. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

OF6 sought to gauge the respondents’ views on the statement which sought to test whether the top-level management support of innovativeness within the organization would make it easier to sell the benefits of acquiring AI tools for the financial reporting process/made it easier to sell the benefits of acquiring AI tools for the financial reporting process. The study found that 82.8% of surveyed accountants strongly agreed/agreed with this statement, whereas 8.2% of respondents indicated that they were neutral and 9% indicated that they strongly disagreed/disagreed with the statement. The mean for OF6 was 4.28 and the standard deviation was 1.135. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

Lastly, OF7 sought to gauge the respondents’ views on the statement which sought to test whether the more emphasis the organization places on procedural conformance and higher level (executive) control would affect the decision of the finance function (accountants) whether to push for the acquiring of AI for the financial reporting process/was a determinant in the consideration for adopting AI in the financial reporting process. The study found that 75.8% of surveyed accountants strongly agreed/agreed with this statement, whereas 13.4% of respondents indicated that they were neutral and 10.8% indicated that they strongly disagreed/disagreed with the statement. The mean for OF7 was 3.97 and the standard deviation was 1.095. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

Having analysed the information relating to the organisational factors, Table 3 below depicts the constructed custom table for technological factors using the aggregated scale of disagree/strongly disagree, neutral (no aggregation here), and agree/strongly disagree.

Table 3. Technological factors.

Technological Factors		Disagree/Strongly Disagree	Neutral	Agree/Strongly Agree	Mean	SD
TF1	Count	16	25	108	4.03	1.117
	Row n%	10.7	16.8	72.5
TF2	Count	15	8	126	4.07	1.140
	Row n%	10.0	5.4	84.5
TF3	Count	10	19	120	4.14	1.160
	Row n%	6.7	12.8	80.5
TF4	Count	16	19	111	4.12	0.976
	Row n%	11.0	13.0	76.0
TF5	Count	18	11	119	4.28	0.932
	Row n%	12.1	7.4	80.4
TF6	Count	13	12	120	4.30	1.063
	Row n%	9.0	8.2	82.8
TF7	Count	16	20	113	3.95	1.208
	Row n%	10.8	13.4	75.8

Note: TF refers to technological factors—see Table A2 of Appendix A for TF descriptions.

Table 3 above tested the statement relating to the technological factors as attributes for the adoption of AI in financial reporting. In this regard, TF1 represents the technological factor that tested the statement that the perceived easiness of integrating AI technologies across other organizational systems will affect the decision of whether to acquire AI in the financial reporting process/was a determinant for acquiring AI in the financial reporting process. A total of 72.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 16.8% of respondents indicated that they were neutral and 10.7% indicated that they strongly disagreed/disagreed with the statement. The mean for TF1 was 4.03 and the standard deviation was 1.117. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

TF2 sought to gauge the respondents’ views on the statement which sought to test whether the perceived easiness of utilizing AI technologies by a team (accountants in the division) will affect the decision of whether to acquire AI in our financial reporting process/was a determinant for adopting AI in the financial acquiring process. The findings were that 84.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 5.5% of respondents indicated that they were neutral and 10.0% indicated that they strongly disagreed/disagreed with the statement. The mean for TF2 was 4.07 and the standard deviation was 1.140. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

TF3 sought to gauge the respondents’ views on the statement which sought to test whether the teams (accountants’) skills and knowledge of AI tools would affect the decision of whether to acquire AI in our financial reporting process/was a determinant for acquiring AI in the financial reporting process. The findings point to the fact that 80.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 12.8% of respondents indicated that they were neutral and 6.7% indicated that they strongly disagreed/disagreed with the statement. The mean for TF3 was 4.14 and the standard deviation was 1.160. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

TF4 sought to gauge the respondents’ views on the statement which sought to test whether the market availability of AI tools will affect the decision of whether to adopt AI in our financial reporting process/was a determinant for acquiring AI in the financial reporting process. The findings of this statement indicate that 76.0% of surveyed accountants strongly agreed/agreed with this statement, whereas 13.0% of respondents indicated that they were neutral and 11.0% indicated that they strongly disagreed/disagreed with the statement. The mean for TF4 was 4.12 and the standard deviation was 0.976. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

TF5 sought to gauge the respondents’ views on the statement which sought to test whether the availability of training on AI tools will affect the decision of whether to acquire AI in our financial reporting process/was a determinant for acquiring AI in the financial reporting process. The study found that 80.4% of surveyed accountants strongly agreed/agreed with this statement, whereas 7.4% of respondents indicated that they were neutral and 12.1% indicated that they strongly disagreed/disagreed with the statement. The mean for TF5 was 4.28 and the standard deviation was 0.932. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

TF6 sought to gauge the respondents’ views on the statement which sought to test whether the availability of technical support from suppliers of AI tools will affect the decision of whether to acquire AI in our financial reporting process/was a determinant for acquiring AI in the financial reporting process. The study found that 82.4% of surveyed accountants strongly agreed/agreed with this statement, whereas 8.2% of respondents indicated that they were neutral and 9.0% indicated that they strongly disagreed/disagreed with the statement. The mean for TF6 was 4.30 and the standard deviation was 1.063. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

Lastly, TF7 sought to gauge the respondents’ views on the statement which sought to test whether the staff profile (in terms of age group) in the finance/accounting function will affect the decision of whether to acquire AI tools in our financial reporting process/was a determinant for acquiring AI in the financial reporting process. The study found that 75.8% of surveyed accountants strongly agreed/agreed with this statement, whereas 13.4% of respondents indicated that they were neutral and 10.8% indicated that they strongly disagreed/disagreed with the statement. The mean for TF7 was 3.95 and the standard deviation was 1.208. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

Having analysed the information relating to the organisational factors, Table 4 below depicts the constructed custom table for technological factors using the aggregated scale of disagree/strongly disagree, neutral (no aggregation here), and agree/strongly disagree.

Table 4. Environmental factors.

Environmental Factors		Disagree/Strongly Disagree	Neutral	Agree/Strongly Agree	Mean	SD
EF1	Count	30	31	89	3.59	1.118
	Row n%	20.0	20.7	59.3
EF2	Count	27	34	88	3.56	1.147
	Row n%	18.1	22.8	59.1
EF3	Count	22	29	98	3.72	1.083
	Row n%	14.7	19.5	65.8
EF4	Count	46	30	72	3.26	1.269
	Row n%	31.0	20.3	48.7
EF5	Count	31	40	78	3.44	1.135
	Row n%	20.8	26.8	52.4
EF6	Count	40	37	72	3.22	1.104
	Row n%	26.8	24.8	48.4
EF7	Count	32	27	90	3.58	1.158
	Row n%	21.5	18.1	60.4
EF8	Count	60	47	40	2.73	1.247
	Row n%	40.8	32.0	27.2
EF9	Count	59	48	40	2.83	1.230
	Row n%	40.1	32.7	27.2

Note: EF refers to environmental factors—see Table A3 of Appendix A for EF descriptions.

Table 4 above tested the statement relating to the environmental factors as attributes for the adoption of AI in financial reporting. In this regard, EF1 represents the environmental factor [1] which tested the statement that the perceived lagging from competitor organisations to integrate AI technologies in the financial reporting process is pushing the finance department (has already pushed the finance department) to acquire AI technologies for the financial reporting process. A total of 65.8% of surveyed accountants strongly agreed/agreed with this statement, whereas 19.5% of respondents indicated that they were neutral and 20.0% indicated that they strongly disagreed/disagreed with the statement. The mean for EF1 was 3.59 and the standard deviation was 1.118. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

EF2 sought to gauge the respondents’ views on the statement which sought to test companies that are subsidiaries or holding companies as to whether the perceived lagging from the group (group/subsidiary organisations) to integrate AI technologies in the financial reporting process is also pushing the finance department (has already pushed the finance department) to acquire AI technologies for the financial reporting process. The findings are that 59.1% of surveyed accountants strongly agreed/agreed with this statement, whereas 22.8% of respondents indicated that they were neutral and 18.1% indicated that they strongly disagreed/disagreed with the statement. The mean for EF2 was 3.56 and the standard deviation was 1.147. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

EF3 sought to gauge the respondents’ views on the statement which sought to test whether the perceived lagging from the industry/international standards to integrate AI technologies in the financial reporting process is also pushing the finance department (has already pushed the finance department) to acquire AI technologies for the financial reporting process. The findings point to the fact that 65.8% of surveyed accountants strongly agreed/agreed with this statement, whereas 19.5% of respondents indicated that they were neutral and 14.7% indicated that they strongly disagreed/disagreed with the statement. The mean for EF3 was 3.72 and the standard deviation was 1.083. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

EF4 sought to gauge the respondents’ views on the statement which sought to test whether the pressure from regulators is pushing the organisation (has already pushed the organisation/finance) to acquire AI technologies in the finance function and for financial reporting. The findings of this statement indicate that 48.7% of surveyed accountants strongly agreed/agreed with this statement, whereas 20.3% of respondents indicated that they were neutral and 31.0% indicated that they strongly disagreed/disagreed with the statement. The mean for EF4 was 3.26 and the standard deviation was 1.269. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

EF5 sought to gauge the respondents’ views on the statement which sought to test whether the pressure from investors is pushing the organisation (has already pushed the organisation/finance) to acquire AI technologies in the finance function and for financial reporting. The study found that 52.4% of surveyed accountants strongly agreed/agreed with this statement, whereas 26.8% of respondents indicated that they were neutral and 20.8% indicated that they strongly disagreed/disagreed with the statement. The mean for EF5 was 3.44 and the standard deviation was 1.135. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

EF6 sought to gauge the respondents’ views on the statement which sought to test whether the pressure from suppliers is pushing the organisation (has already pushed the organisation/finance) to acquire AI technologies in the finance function and for financial reporting. The study found that 48.4% of surveyed accountants strongly agreed/agreed with this statement, whereas 24.8% of respondents indicated that they were neutral and 26.8% indicated that they strongly disagreed/disagreed with the statement. The mean for EF6 was 3.22 and the standard deviation was 1.104. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

EF7 sought to gauge the respondents’ views on the statement which sought to test whether the pressure from customers is pushing the organisation (has already pushed the organisation/finance) to acquire AI technologies in the finance function and for financial reporting. The study found that 60.4% of surveyed accountants strongly agreed/agreed with this statement, whereas 18.1% of respondents indicated that they were neutral and 21.5% indicated that they strongly disagreed/disagreed with the statement. The mean for EF7 was 3.58 and the standard deviation was 1.158. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

EF8 sought to gauge the respondents’ views on the statement which sought to test whether the pressure from trade unions is pushing the organisation (has already pushed the organisation/finance) to acquire AI technologies in the finance function and for financial reporting. The study found that 27.2% of surveyed accountants strongly agreed/agreed with this statement, whereas 32.0% of respondents indicated that they were neutral and 40.8% indicated that they strongly disagreed/disagreed with the statement. The mean for EF8 was 2.73 and the standard deviation was 1.247. Given that the questionnaire used a five-point Likert scale, a mean below 3 means that most respondents disagreed or strongly disagreed with this organisational factor.

Finally, EF9 sought to gauge the respondents’ views on the statement which sought to test whether the pressure from lobby groups is pushing the organisation (has already pushed the organisation/finance) to acquire AI technologies in the finance function and for the purposes of financial reporting. The study found that 27.2% of surveyed accountants strongly agreed/agreed with this statement, whereas 32.7% of respondents indicated that they were neutral and 40.1% indicated that they strongly disagreed/disagreed with the statement. The mean for EF9 was 2.83 and the standard deviation was 1.230. Given that the questionnaire used a five-point Likert scale, a mean below 3 means that most respondents disagreed or strongly disagreed with this organisational factor. Having analysed the information relating to the organisational factors, Table 5 below depicts the constructed custom table for AI adoption factors using the aggregated scale of disagree/strongly disagree, neutral (no aggregation here), and agree/strongly disagree.

Table 5. AI adoption factors.

Adoption Factors		Disagree/Strongly Disagree	Neutral	Agree/Strongly Agree	Mean	SD
AF1	Count	19	18	112	3.92	1.062
	Row n%	12.8	12.0	75.2
AF2	Count	20	14	115	3.95	1.058
	Row n%	13.4	9.4	77.2
AF3	Count	20	16	111	3.95	1.109
	Row n%	13.6	10.9	75.5
AF4	Count	16	16	115	4.04	1.039
	Row n%	10.9	10.9	78.3
AF5	Count	23	20	106	3.84	1.139
	Row n%	15.4	13.4	71.2
AF6	Count	21	19	109	3.91	1.156
	Row n%	14.1	12.8	73.2
AF7	Count	27	20	99	3.74	1.163
	Row n%	18.5	13.7	67.7
AF8	Count	36	24	89	3.57	1.233
	Row n%	24.1	16.1	59.7

Note: AF refers to adoption factors—see Table A4 of Appendix A for AF descriptions.

AF1 represents the AI adoption factor which gauged the respondents’ views that tested the statement that AI tools are used to harness financial and nonfinancial data as part of the financial reporting process. A total of 75.2% of surveyed accountants strongly agreed/agreed with this statement, whereas 12% of respondents indicated that they were neutral and 12.8% indicated that they strongly disagreed/disagreed with the statement. The mean for AF1 was 3.92 and the standard deviation was 1.062. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

AF2 gauged the respondents’ views that tested the statement that AI tools are used to store financial and nonfinancial data as part of the financial reporting process. A total of 77.2% of surveyed accountants strongly agreed/agreed with this statement, whereas 9.4% of respondents indicated that they were neutral and 13.4% indicated that they strongly disagreed/disagreed with the statement. The mean for AF2 was 3.95 and the standard deviation was 1.058. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

AF3 gauged the respondents’ views that tested the statement that AI tools are used to integrate financial and nonfinancial data as part of the financial reporting process. A total of 75.5% of surveyed accountants strongly agreed/agreed with this statement, whereas 10.9% of respondents indicated that they were neutral and 13.6% indicated that they strongly disagreed/disagreed with the statement. The mean for AF3 was 3.95 and the standard deviation was 1.109. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

AF4 gauged the respondents’ views that tested the statement that AI tools are used to analyse financial and nonfinancial data as part of financial reporting process. A total of 78.2% of surveyed accountants strongly agreed/agreed with this statement, whereas 10.9% of respondents indicated that they were neutral and 10.9% indicated that they strongly disagreed/disagreed with the statement. The mean for AF4 was 4.04 and the standard deviation was 1.039. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

AF5 gauged the respondents’ views that tested the statement that AI tools are used to identify errors in financial and nonfinancial data as part of financial reporting process. A total of 71.2% of surveyed accountants strongly agreed/agreed with this statement, whereas 13.4% of respondents indicated that they were neutral and 15.4% indicated that they strongly disagreed/disagreed with the statement. The mean for AF5 was 3.84 and the standard deviation was 1.139. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

AF6 gauged the respondents’ views that tested the statement that AI tools are used to conduct period-to-period comparisons in financial and nonfinancial data as part of the financial reporting process. A total of 73.2% of surveyed accountants strongly agreed/agreed with this statement, whereas 12.8% of respondents indicated that they were neutral and 14% indicated that they strongly disagreed/disagreed with the statement. The mean for AF6 was 3.91 and the standard deviation was 1.156. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

AF7 gauged the respondents’ views that tested the statement that AI tools are used to interpret financial and nonfinancial data as part of the financial reporting process. A total of 67.7% of surveyed accountants strongly agreed/agreed with this statement, whereas 13.7% of respondents indicated that they were neutral and 18.6% indicated that they strongly disagreed/disagreed with the statement. The mean for AF7 was 3.74 and the standard deviation was 1.163. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

AF8 gauged the respondents’ views that tested the statement that AI tools are used in the presentation of a financial report to stakeholders. A total of 59.7% of surveyed accountants strongly agreed/agreed with this statement, whereas 16.1% of respondents indicated that they were neutral and 14.2% indicated that they strongly disagreed/disagreed with the statement. The mean for AF8 was 3.57 and the standard deviation was 1.233. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with this organisational factor.

5. Conclusions and Discussion

At the onset, this study sought to explore and analyse the views of South African accountants regarding the factors that affect the adoption of AI in financial reporting. In other words, the study aimed to understand what accountants in South Africa think about the use of AI in their field, especially concerning its integration into financial reporting practices. The following is the summary of findings on the views of South African registered accountants:

• For the organisational factors, the means of all organisational factors measured were above 3. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with the gauged organisational factors.
• For the technological factors, the mean of six technological factors measured was above 4 and only one factor was close to 4. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with the gauged technological factors.
• For the environmental factors, of the nine factors that were gauged from the respondents, the mean for seven environmental factors measured was above 3. There were, however, two factors that had a mean of less than 3. Given that the questionnaire used a five-point Likert scale, a mean above 3 means that most respondents agreed or strongly agreed with the gauged environmental factors. A mean below 3 means that most respondents to these factors disagreed or strongly disagreed with the gauged environmental factors.

This study concludes, then, that the perceptions of South African accountants, overall, are that organizational, technological, and environmental factors play a role in adopting AI in financial reporting. Having said this, there were two factors that South African accountants had a perception did not play a role in the decision to adopt artificial intelligence in financial reporting, for instance, South African accountants did not see the trade union pressure and the lobby groups as having a say in the decision to adopt artificial intelligence in financial reporting.

5.1. Limitations

This study only considered the views of South African accountants. Further, only those accountants who were members of the IFAC-recognized bodies were considered. Another limitation of the paper is that it did not examine the relationships among the chosen factors towards AI adoption.

5.2. Areas of Future Research

The current study sought to explore and analyse the views of South African accountants regarding the factors that affect the adoption of AI in financial reporting. This means that this study aimed to understand what accountants in South Africa think about using AI in the field, especially concerning its integration into financial reporting practices. This research was focused on the South African context and future research could be driven from the International Federation of Accountants (IFAC) level. A survey issued by IFAC could undertake a cross-country comparison to determine areas of convergence and areas of divergence among accountants in different countries. A further area of future research could look into the use of AI in planning, performance, and reporting. A future study could also examine the relationships among the chosen factors towards AI adoption. This could be examined using statistical analysis and verification analysis.