The DOI model has its roots in sociology, and has four main elements: time, the innovation, channels of communication and the social system (Rogers & Shoemaker, 1983). The model considers the transfer of innovations within social systems via appropriate channels of communication over time to predict users’ adoption patterns (Rogers & Shoemaker, 1983; Taherdoost, 2018).

In addition to the main elements of innovation diffusion, the DOI model has three main characteristics: innovation characteristics, adopter characteristics and the innovation-decision process (Taherdoost, 2018). The innovation characteristics are trialability, relative advantage and complexity (Rogers & Shoemaker, 1983). Adopters are characterised as early adopters, innovators, laggards, early majority and late majority (Rogers & Shoemaker, 1983; Taherdoost, 2018). The innovation-decision process outlines five phases of an information-seeking process in which the potential adopters are encouraged to reduce their perceived uncertainty about a technological innovation and thus increase adoption (Miranda, Farias, Schwartz, Pascualote, & De Almeida, 2016).

The TOE framework was developed as a way to unveil the dimensions of an organisation’s context that affect the rate at which innovations are adopted (Aboelmaged, 2014). The framework consists of three contextual dimensions: technological, organisational and environmental (Aboelmaged, 2014; Al-jabri & Alabdulhadi, 2016). The technological context describes adopters’ perceptions of the technology’s attributes, for example, perceived benefits and ease of use; the organisational context examines the organisational characteristics such as firm size and management structure; and the environmental context describes the firm’s interaction with other firms, regulatory bodies and consumers (Amini & Bakri, 2015; Awa, Ojiabo, & Orokor, 2017; Tornatzky & Fleischer in Lin, 2014). Unlike the DOI model, the TOE framework focuses solely on firm-level technology adoption whilst also incorporating environmental aspects (Amini & Bakri, 2015).

Scholars (Aboelmaged, 2014; Lin, 2014; Lin & Lin, 2008) have examined the TOE framework to validate its theoretical depth and effectiveness in examining readiness for various innovations. Aboelmaged (2014) found that the TOE determinants – such as anticipated benefits and firm size, amongst others – influenced elements of e-maintenance technology readiness. Competitive pressure was found to influence electronic business (e-business) diffusion (Lin & Lin, 2008), whilst the readiness of e-maintenance in manufacturing firms was not influenced by this determinant (Aboelmaged, 2014).

‘Technology readiness’ describes an individual’s propensity to embrace new technology to accomplish goals in his or her work lives or homes (Parasuraman & Colby (2001) in Lai, 2017). The TRI model measures the extent to which an individual is ready to accept and use new technologies, and has four dimensions: optimism, innovativeness, discomfort and insecurity (Nugroho & Fajar, 2017; Nugroho, Susilo, Fajar, & Rahmawati, 2018). High levels of optimism and innovativeness promote technology adoption, whilst an individual’s level of discomfort and insecurity inhibit adoption (Aboelmaged, 2014; Awa et al., 2017). The technology readiness level categorises individuals into laggards, paranoids, sceptics, pioneers or explorers.

Regarding 4IR technologies in the financial sector, laggards may prefer banking via traditional physical banking channels (such as visiting the bank branches) over modern digital banking platforms. These users are generally the last group to adopt emerging technology (Nugroho et al., 2018). Innovations excite paranoids; however, their perception of technology is also driven by some elements of fear, and so these users will consider risk factors such as the regulatory implications of P2P lending before completely adopting the technology. Sceptics have a low motivation to accept innovations. These users must, therefore, be exposed to the potential benefits of incurring lower banking fees before they can accept 4IR technologies. Pioneers show significant optimism about innovations; however, these users will also stop using the technology if they encounter any significant challenges (Nugroho et al., 2018). Lastly, explorers are typically the first group of users to try out new technology, and will eagerly commit to sharing data about their driving patterns with insurance firms to try out innovative ‘pay-as-you-drive’ capabilities (Tselentis, Yannis, & Vlahogianni, 2016).

Knowledge of innovation: Awareness, exposure and a high-level understanding of 4IR technologies were found to influence technology adoption. An acute exposure to ML and AI was found to influence short-term usage intent, whereas long-term use behaviour was significantly influenced by the need for the technologies in the lives of consumers. This was prevalent amongst respondents in the banking sector.

Exposure was influenced by individual adopters’ personal experiences with the technology. For example, individuals who had been exposed to the negative aspects of emerging technologies were less likely to adopt these technologies. It was, thus, crucial to have credible sources introducing concepts of 4IR technologies to potential adopters in order to improve adoption.

Adequate knowledge of emerging technologies, particularly AI and DLT, had a role to play in reducing the fears arising from a lack of understanding of these technologies. Respondents in the insurance and financial regulatory sectors were concerned about potential job losses arising from the proliferation of 4IR technologies in the sector. These findings are supported by the following statements:

These findings support proposition P1-1, and suggest that knowledge of the innovation positively influences technology adoption.

Technical skill: Technical training and sufficient work experience were beneficial for the adoption of 4IR technologies amongst working professionals. However, this was not as relevant to end-users. Instead, end-users were more concerned about the value that the technologies would add to their lives than with the technical components comprising these technologies.

Sufficient technical understanding of emerging technologies by professionals enabled them to develop technologies that were beneficial to their customers or the financial institutions that employed them. For example, a thorough understanding of DLT and ML enabled investment banks to research and develop technology that would reduce interbank settlement periods from days to minutes. Retail banks were also able to leverage AI to improve profitability whilst reducing operating costs.

These findings support proposition P1-2, and suggest that technical skill positively influences technology adoption.

Personal innovativeness: Personal innovativeness was characterised by basic constructs such as the willingness to innovate, a curiosity about exploration and the ability to take risks and try out new technologies as they become available. These basic constructs had a stronger influence on individuals’ (both professionals and customers) adoption intent and use behaviour than those that characterised the technical skill organising construct.

Individuals who were willing to try out ML and AI were more likely to adopt these technologies than individuals who had the technical skill but lacked the willingness. This was prevalent amongst respondents from the banking and insurance sectors. Whenever individuals had the curiosity to explore the benefits that the bank could harness from these technologies, they were able to motivate the acquisition of resources (budget allocation, hiring of skilled professionals, etc.) for technology adoption. These findings are supported by the following statement:

These findings support proposition P1-3, and suggest that personal innovativeness has a strong positive influence on technology adoption. It can, thus, be inferred from the findings that adopter traits are an influential determinant of the adoption intent and use behaviour with emerging 4IR technologies in the SA financial sector.

Trialability: The ability of data scientists to experiment with AI prior to large-scale development reduced the risk of failure. This culture encouraged technologists to ‘fail fast and recover quickly’, thus contributing to innovation. Deploying ML algorithms in a phased approach increased the chances of obtaining successful outcomes. It also enabled data scientists to demonstrate the benefits of AI to the rest of the business, and garner support from executives. The following statement supports this finding:

Customers also preferred to test new technology before fully adopting and using it in long-term. This also minimised the potential risks associated with the technology, creating a sense of comfort for the customer. The following statement supports this finding:

These findings support proposition P2-1, and suggest that trialability positively influences technology adoption.

Complexity: Customers were very quick to abandon the use of a new technology if they experienced it as complex. This increased firms’ technology marketing efforts, and subsequently the costs associated with the technology. At the firm level, cloud computing platforms increasingly became the preferred data storage mechanism owing to the reduced maintenance costs. The complexity associated with DLT inhibited firms from trying out blockchain-related technologies. The following statement supports these findings:

Delays caused by the complexity of deploying AI models rendered the models irrelevant to solving the business problems for which they were specifically designed. These models then had to be retrained with new data, which further reduced confidence in the adoption of AI. This was prevalent amongst fintech firms.

These findings support proposition P2-2, and suggest that complexity has a negative influence on technology adoption.

User insecurity: Professionals in the banking and insurance sectors were concerned that cloud computing and AI would make their current technical expertise irrelevant, potentially leading to job losses. This discomfort was exacerbated by the high unemployment levels in SA, and generated scepticism about the adoption of these technologies. Cyber security concerns also inhibited adoption, particularly in banking institutions. Professionals were also concerned about the misrepresentation of AI and ML and the types of problems these technologies could solve. It was feared that ongoing misrepresentation would generate a negative sentiment about the technology, and contribute to reduced adoption. The following statement supports these findings:

These findings support proposition P2-3, and suggest that user insecurity negatively influences technology adoption. It can, thus, be inferred that technology usability is an influential determinant of the adoption of emerging 4IR technologies in the SA financial sector.

Competitive pressures: Banks relied significantly on technology as a source of competitive advantage, and this generated interest in AI adoption in the sector. The agility of fintech firms exerted pressure on incumbent financial institutions to accelerate innovation.

Unlike AI, cloud computing technology did not generate a clear-cut competitive advantage for adopting firms. This was attributed to the costs, which often reversed the benefits of the technology. Firms were motivated by the ability to leverage cloud computing platforms to service clients more seamlessly, thus gaining a competitive edge over rivals. The following statement supports these findings:

These findings support proposition P3-1, and suggest that competitive pressure positively influences technology adoption.

Regulatory su pport: Financial regulators in SA were reported to have consulted widely about the adoption of 4IR technologies. However, the reactive nature of the regulators inhibited adoption, as innovating firms would be expected to explain their innovations to the regulators. This was attributed to a lack of understanding of the technology.

A proof-of-concept called ‘Project Khokha’, facilitated by the SARB, was used as one example of the consultative approach taken by regulators to promote the adoption of DLT and improve the settlements infrastructure in the banking sector. Amendments to the existing policy framework were recommended to improve the industry’s adoption of emerging technologies whilst maintaining regulatory oversight on the use of these technologies.

The regulation of fintech start-ups was minimal to non-existent. This was attributed to the nature of the data used by fintech firms, as it was deemed to be less sensitive than that used by regulated financial institutions. A lack of understanding of the risks associated with emerging innovations also contributed to the lack of regulatory oversight.

These findings support proposition P3-2, and indicate that regulatory support positively influences technology adoption.

Customer demands: Clients with a significant stake in a firm had a strong influence on the firm’s technology strategy, including its adoption of emerging technologies. For fintech firms, shareholders could significantly influence senior management to incorporate AI in their innovation strategies. Investment banks also yielded to the demands of clients with a significant financial stake in the bank. However, individual retail banking clients and insurance policy holders had a weaker influence on the technology adopted by their service providers. These clients were increasingly using digital channels as a way to engage. The wide adoption of smart devices by individual and retail clients was a strong motivating factor for fintech start-ups as they built innovations that leveraged smart devices. Regulators were also prompted to establish working groups to respond to the rate at which regulated institutions were adopting ML, AI and DLT. The following statement supports these findings:

These findings support proposition P3-3, and suggest that customer demands positively influence technology adoption. It can, thus, further be suggested from the findings that industry characteristics (proposition P3) are an influential determinant towards the adoption use behaviour of AI, cloud computing and DLT in the SA financial sector.

Opinion leadership: Opinion leaders played a significant role in persuading the organisation to accept 4IR technologies. In the banking environment, these leaders had great foresight into market trends, and were able to influence senior management to support the development of AI, DLT and ML platforms. The following statement supports these findings:

Leaders who inspired employees to take part-ownership of the innovation strategy were more effective in influencing banking institutions to adopt 4IR technologies. This created a community of leaders in the bank, instead of a single focal point of leadership.

In fintech firms, leaders were also founders and or shareholders in the firms, and had a significant influence over the firms’ technology strategies. This generated a lack of diversity in the leadership opinion governing the firm, and negatively impacted the chances of successfully adopting innovations. Diversity in the leadership structures was necessary to interrogate the business need and clearly articulate the capabilities of AI for the adopting firm.

These findings provide support for proposition P4-1, and suggest that opinion leadership positively influences technology adoption. In addition to opinion leadership, leadership diversity was also found to influence technology adoption positively.

Decision-maker risk orientation: Emerging 4IR technologies represented uncharted territories in which the potential benefits of technologies such as AI were not yet fully understood. Therefore, it was crucial for leaders in a firm to acquire resources in the awareness that the expected benefits might not be realised. The following statement supports these findings:

These findings suggest that the decision-makers’ risk orientation positively influences technology adoption, in line with proposition P4-2. It can be inferred from these findings that organisational leadership is an influential determinant of the adoption of emerging 4IR technologies in the SA financial sector.

Firm size: Fintech start-up firms were more agile than larger incumbent financial institutions, and had insignificant legacy constraints (people, processes and technologies) to overcome when adopting innovations. In the context of cloud computing technologies, fintech firms were swift to replace on-site data centres with cloud infrastructure. Fintech firms implementing conversational AI technology were able to react swiftly to their customers’ demands, constantly improving their product offerings to suit their market.

Larger incumbent financial institutions, particularly regulated banks and insurance firms, were more risk-averse, and had significant protocols in place that hindered the proliferation of emerging 4IR technologies in these organisations. However, where technology adoption was successful, these institutions were able to reach larger adoption scales owing to the significant number of resources at their disposal. The following statement supports these findings:

These findings, therefore, do not support proposition P5-1, and indicate that neither increasing nor decreasing the firm’s size influences technology adoption.

Top management support: Senior executives in the banking sector have a key role to play in articulating the necessity of AI technology adoption by a bank. These leaders could incentivise ideation processes in the organisation, motivating employees to solve business problems using ML. Senior managers were also tasked with outlining the bank’s strategic direction, and so were not expected to take a conservative approach to innovation. Having innovation champions at the senior executive level was also essential to acquiring financial resources and communicating the need for innovation amongst other senior business executives. The following statement supports these findings:

Departments in the banks whose CEOs were cognisant of cloud computing technology encountered fewer challenges during the deployment of this technology. Amongst insurers, CIOs who managed to articulate an innovation strategy clearly, to facilitate transparency and to celebrate successes were more effective in encouraging cloud computing technology adoption in their departments.

These findings support the proposition P5-2, and suggest that top management support positively influences technology adoption.

Communication mechanisms: Flatter organisational structures were more effective in disseminating information across the organisation, and thus were more conducive to innovation. Effective communication mechanisms enabled senior bank managers to diffuse strategy across the organisation and enable employees to communicate their innovative technological ideas to senior management. The following statements support these findings:

These findings support proposition P5-3, and suggest that effective communication structures positively influence technology adoption. It can, thus, further be suggested from the findings that effective communication structure (proposition P5) is an influential determinant towards the adoption use behaviour of AI, cloud computing and DLT in the SA financial sector.