Page 704 - ISC PROCEEDINGS 21.4

Using an integrated TAM-TPB perspective, Nguyen and Ha [7] show that perceived
behavioral control, computer self-efficacy, and attitude toward AI positively influence
behavioral intention, while perceived usefulness and perceived ease of use positively
affect attitude. Linh [8] similarly reports that perceived usefulness, mobility, and
convenience positively influence intention to use ChatGPT, although the role of perceived
ease of use is less stable across contexts. In more specific learning settings, Tran [9] finds
that performance expectancy, effort expectancy, social influence, and facilitating
conditions all significantly influence acceptance of ChatGPT in academic writing among
English-majored students, with performance expectancy exerting the strongest effect.
More recent work has broadened the focus from ChatGPT to large language models
and AI-supported learning more generally. Ngo et al. [10] show that perceived ease of use,
perceived usefulness, and trust positively influence attitudes, while attitude, subjective
norms, and perceived behavioral control positively influence intention to use large
language models for learning. Thai et al. [11] further demonstrate that digital competence
enhances perceived ease of use and self-efficacy, while perceived usefulness remains a
strong driver of AI adoption and perceived risks inhibit acceptance.
Taken together, several consistent patterns emerge. First, perceived usefulness
almost always appears as a strong determinant of intention to use AI for learning. Second,
perceived ease of use also tends to play a positive role, but its magnitude varies across
contexts. Third, learner-related constructs such as self-efficacy and trust have become
increasingly important. Fourth, the social and academic environment remains relevant,
even when AI adoption is often treated as an individual technology decision. Despite this
progress, the literature still leaves room for further empirical evidence on AI tools for
learning beyond ChatGPT alone, especially in Vietnam and from integrated models that
combine technological, social, and learner capability factors.
2.2. Theoretical background
This study draws on TAM, UTAUT, and TPB. TAM posits that perceived usefulness
and perceived ease of use are two core determinants of technology acceptance. UTAUT
highlights the role of social influence and reminds us that students do not adopt
technologies in isolation; their intentions are also shaped by peers, instructors, and
academic norms. TPB, meanwhile, helps explain the role of personal beliefs and perceived
control, offering an appropriate conceptual foundation for AI self-efficacy.
From this perspective, students’ intention to use AI tools for learning can be
understood as the combined result of technology-related perceptions, social influences,
and individual capability- and trust-related factors. Because the refined SPSS results show
that a direct regression model with five independent variables provides strong
explanatory power, the present study adopts a direct-effects model linking PEOU, PU, SI,
AISE, and TR to IU.
2.3. Perceived ease of use and intention to use AI tools for learning
Perceived ease of use reflects the extent to which students believe that AI tools are
understandable, easy to operate, and do not require excessive effort. In the learning
context, the more convenient a tool is, the more easily it can be incorporated into
activities such as information searching, idea generation, concept explanation, and
writing support. As use barriers decrease, students are more likely to develop a stronger
intention to use AI tools for learning.
H1: Perceived ease of use (PEOU) positively affects intention to use AI tools for
learning (IU).

703