Davis et al. (1989) suggested that TAM determines the antecedents of technology and system acceptability. The antecedents, namely PU, PEoU and ATT from the mentioned model, are adapted for this study to examine their influence on consumers' intention to adopt AI-based financial services. Venkatesh and Davis (2000) also pointed out that these antecedents affect each other. For example, PEoU will affect PU of using a new technology. When the particular system is easy to use, it is more valuable to the users to improve their work execution. Besides, PU will also affect ATT and individual's intention to accept the system's technology.

Behavioural intention defines the degree to which a person performs a particular behaviour after conscious planning or self-directed motivation ( Mark & Paul, 2005). According to Chuang, Liu and Kao (2016) and Cham et al. (2018), TAM is used to investigate a person's behaviour and the results show that PU, PEoU and ATT may be used to predict a person's readiness to adopt new technology.

PU is a crucial factor to determine the acceptance of using AI-based financial services. PU means the extent to which an individual perceived a system that improves their job efficiency ( Davis et al., 1989). Pai and Huang (2011) found that users' willingness to accept AI is related to the system's PU. Another study also demonstrates that a more positive attitude is related to a higher willingness to use AI products ( Lunney et al., 2016). Likewise, Pillai and Sivathanu (2020) studied the impact of using AI-powered customer service technology in the tourism industry. The result shows that users are willing to use AI devices to plan for their trip due to the easy and interactive booking process and service.

To master or use a system or technology without much effort is identified as perceived ease of use ( Davis et al., 1989). Hamid et al. (2016) found that PEoU is positively related to the adoption of e-government system. It indicates that the respondents are keen on using the new system if the technology seemed easily understood. Furthermore, Patil and Mugdha (2019) found a significant positive relationship between PeoU and adoption of chatbot services in financial services. This is because the system is a simple drag-and-drop interface and hence respondents are more willing to use it.

Azjen (1991) defined attitude as the degree of a user's assessment and expression of the behaviour's execution. Favourable or unfavourable attitude emerges from the personnel's beliefs on behavioural outcomes ( Cordano & Frieze, 2000). The user's attitude toward AI will affect their willingness to accept AI in financial services ( Schepman & Rodway, 2020). Previous studies discovered a positive relationship between attitudes and intentions to use the system ( Chang, Vitell, & Lu, 2019; Le & Kieu, 2019; Prendergast & Tsang, 2019; Tweneboah-Koduah, Adams, & Nyarku, 2019).

Research framework ( Figure 1) is shown in the proposed research framework.

The Cronbach's alpha test is used to determine the questionnaire's reliability. Additionally, the SPSS software is used to evaluate the reliability analysis with 30 respondents. Results shown in Table 1 indicates that Cronbach's alpha's value is greater than 0.7 for all the analysed variables, and thus, the questionnaire is reliable.

Table 2 tabulates the descriptive analysis. Among 150 respondents, 44% are male, while 56% are female respondents. According to Dolot (2018), individuals born after 1995 are referred to as Gen Z. As a result, respondents are between the ages of 20 and 22 (55.3%), between 23 and 25 (22.0%), under 20 (14.0%), and above 25 (8.7%) years.

This study performs a measuring model to evaluate the elements of a path model, which contains indicators and their relationships with the constructs. The minimum threshold of 0.708, 0.70, and 0.50 are assessed on factor loadings, composite reliability and average variance, respectively. They are evaluated on the validity and reliability of the constructs ( Hair et al., 2019). Hence, PU1, PU2, PEOU1, ATT1, ATT3, INT1 and INT5 were removed because their respective loadings were less than 0.708. The factor loadings, composite reliability, and average variance extracted for the final measurement model met the minimum threshold, which the values are tabulated in Table 3. The results also suggested that the measurement model was reliable and had adequate convergent validity. Subsequently, a discriminant validity assessment was performed because the content and substance of the constructs had to be identified. Therefore, this study employed the Heterotrait-Monotrait Ratio (HTMT) ( Henseler et al., 2015). Firstly, the construct - the intention had failed to meet the HTMT 0.9 criteria. Upon checking with the cross-loading, INT3 was removed. Table 4 shows that all the values are lower than 0.90, which is the required threshold suggested by Gold et al. (2001). Hence, the findings shows that the proposed hypotheses are accepted and verified by the value of the discriminant validity.

Table 5 presents the hypotheses statements in this study. The structural model is presented in Figure 2. These findings revealed that the Coefficient of Determination (R ²) value of the model is 0.467, which suggested a moderate predictive model ( Chin, 1998). It also means that 46.70% of the INT can be explained by PU, PEoU and ATT.

Bootstrapping was applied to determine the significant level of the constructs. The path coefficient and corresponding t-values were gathered with 5,000 resamplings. As a result, PEoU positively influenced the INT, and ATT positively influenced the INT. Hence, H ₂ (t-value = 2.683) and H ₃ (t-value = 5.408) were supported. Nonetheless, PU and INT were insignificant in this study. Thus, H ₁ (t-value = 0.552) was not supported. Figure 2 presented the structural model, and Table 6 summarises the hypotheses testing.

Data archiving and networked services (DANS): Survay of Malaysian view on the roles of artificial intelegence for the future changes in the financial services industries

DOI: https://doi.org/10.17026/dans-2xr-x6wm (Chan et al., 2021).

This project contains the following underlying data:

File 1: The blank questionnaire used in this study.

File 2: Responses to the survey of the Malaysians’ view on the roles of artificial intelligence for the future changes in the financial services industry.

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver ( CC0 1.0 Public domain dedication).