Navigating the Path to AI and Virtual Immersion: An Exploratory Study of Educational Escape Rooms with the ED-SCALE Model

Abstract

The growing integration of immersive technologies into education is opening new possibilities for teaching and learning, while also raising concerns about the reliability and potential distortion of knowledge in artificial intelligence-mediated environments. Understanding how users perceive and accept artificial intelligence-generated content in immersive learning systems is therefore essential. This study explores the factors that influence user acceptance of artificial intelligence-driven virtual reality educational applications and explains it through a multidimensional framework that extends the Technology Acceptance Model, the Theory of Reasoned Action, and the Theory of Planned Behavior—a new ED-SCALE model. We innovated the previous models by adding an ergonomic dimension, often overlooked in virtual reality-based education. To test the model, we developed an artificial intelligence-driven virtual reality educational escape room designed to simulate adaptive and interactive learning experiences. Data were collected from 213 participants using a questionnaire measuring subjective norms, perceived behavioral control, attitudes toward artificial intelligence-mediated instruction, perceived informational efficacy, and ergonomic quality. The findings show that ergonomic quality, intuitive interfaces, physical comfort, and social influence play an important role in shaping user trust and long-term adoption intentions. The results suggest that the success of artificial intelligence-driven immersive learning systems depends not only on technological performance but also on user experience and social context, confirming our first hypothesis regarding new variables that are conditional for virtual technology acceptance.

1. Introduction

The rapid evolution of digital technologies is increasingly shaping educational environments, with artificial intelligence (AI) and extended reality (XR) emerging as key drivers of this transformation. The significant acceleration of research activity in AI-enhanced education, particularly following the emergence of generative AI systems, has introduced new paradigms for personalized, adaptive, and scalable learning experiences [1]. These technologies enable intelligent content generation, real-time feedback, and multimodal interaction, thereby extending traditional pedagogical boundaries and fostering more dynamic, learner-centered educational ecosystems [2]. For example, XR technologies, including virtual (VR), augmented (AR), and mixed reality (MR), enable learning environments in which users can interact more directly with content, supporting experiential learning and improved engagement compared to traditional digital tools [3].

From a pedagogical perspective, immersive technologies may influence learning processes through features specific to virtual environments, particularly the learner’s sense of presence and perceived control during interaction. The Cognitive Affective Model of Immersive Learning (CAMIL) conceptualizes how these characteristics interact with instructional design to shape motivation, engagement, and learning outcomes in immersive settings [4]. Empirical evidence further suggests that XR-based environments can support both conceptual understanding and practical skill development, especially in domains that rely on experiential learning, such as medical education [5].

Despite their transformative potential, the integration of AI and XR into education poses complex pedagogical, ethical, and technological challenges. While AI-driven systems support personalized learning, improve student motivation, and optimize instructional processes, they also raise critical concerns related to academic integrity, data privacy, algorithmic bias, and overreliance on automated systems [4]. Similarly, although XR environments have demonstrated significant benefits in enhancing cognitive outcomes, engagement, and practical skill acquisition, their widespread adoption remains constrained by high implementation costs, accessibility limitations, and the need for specialized pedagogical design [5]. Furthermore, the effectiveness of immersive learning environments depends on both technological capabilities and their alignment with instructional design, particularly in relation to immersive features that influence cognitive load, motivation, and learning transfer.

The convergence of artificial intelligence and immersive technologies is reflected in the broader concept of the Metaverse, which integrates persistent virtual environments with intelligent and interactive capabilities, offering new opportunities for both businesses and individuals [6]. Driven by advancements in blockchain, 5G, and edge computing [7], the Metaverse represents a disruptive technological framework still in its early stages [8,9]. Within this digital space, users (represented by avatars) can overcome physical limitations to engage in work, play, and social interactions [10]. Immersive technologies enable highly realistic online interactions, fostering creative and personalized experiences [11]. In educational settings, it enables the integration of physical and virtual environments, supporting more immersive and adaptive learning opportunities [12,13].

Despite its potential, adoption and user acceptance remain below expectations, highlighting a critical need to understand the factors influencing engagement [14]. AI integration in education offers benefits such as personalization, accessibility, and efficiency, but also introduces challenges, including data privacy concerns, algorithmic bias, and potential inaccuracies in AI-generated content [15]. Research shows that user acceptance is a primary determinant of successful integration, influenced by factors like perceived usefulness, ease of use, social presence, and technological readiness [16,17,18].

The convergence of generative AI and XR represents a promising yet insufficiently explored research direction, particularly in relation to immersive educational environments. Although existing studies have investigated AI or XR independently, there remains a lack of integrated frameworks capable of leveraging both technologies to support real-time personalization and adaptive learning processes [3]. However, there is a limited understanding of how immersive, AI-driven environments influence user acceptance and educational outcomes [1,4].

Building on foundational models such as the Technology Acceptance Model (TAM) [19], Theory of Reasoned Action (TRA) [20,21], and Theory of Planned Behavior (TPB) [22,23], this study introduces ED-SCALE, an extended framework incorporating ergonomics as a critical variable affecting user experience in immersive learning. The current implementation is a VR-based educational environment enhanced with AI-driven feedback mechanisms. Which should not be directly classified as a Metaverse system, but Metaverse integration as a possible future extension of the VR+AI framework. Ergonomics encompasses device comfort, ease of use, and physical effort, ensuring a sustainable and engaging interaction.

This study aims to investigate user interactions and learning outcomes in a 3D AI-driven virtual reality educational escape room, with a particular focus on the factors influencing technology acceptance and perceived educational effectiveness. By analyzing quantitative and observational data, the study provides insights into how AI-generated content, perceived control, ergonomic design, and social context shape user attitudes, engagement, and intention to use immersive learning environments.

The objectives of the study are as follows:

✓

Exploring how AI-generated content in VR influences user perceptions.

✓

Evaluating attitudes towards AI-driven VR and the perceived level of control in VR.

✓

Analyzing the perceived educational effectiveness introduced by AI-generated learning content.

✓

Assessing the impact of VR ergonomics on technology adoption.

This study makes two main contributions: (1) it proposes ED-SCALE as an extended framework for analyzing technology acceptance in AI-driven VR learning environments, and (2) it empirically explores this framework through the design and testing of a 3D AI-driven educational escape room.

The hypothesis serving as the foundational premise of our pilot study:

H1: 

Exposure to AI-driven VR learning environments through friends or colleagues positively influences users’ intention to adopt this technology.

H2: 

The use of AI-generated content in VR increases engagement and motivation for learning.

H3: 

The use of AI-driven virtual reality in education improves engagement and the perception of learning effectiveness.

H4: 

A higher perception of control over the AI-personalized VR interactions increases the likelihood that users will want to use the technology frequently.

H5: 

Users with a positive attitude toward new technologies will perceive interaction with AI-generated VR interactions as more intuitive and easier to use.

H6: 

The level of physical comfort experienced during VR use positively influences users’ intention to adopt the technology in the long term.

By combining quantitative and observational data, the study provides empirical evidence on how technological, cognitive, and physical factors influence user acceptance in AI-driven immersive learning environments.

2. Theoretical Background

2.1. Educational Potential of AI-Driven Immersive Technologies

The convergence of AI and virtual reality offers a completely new vision for education. This technological integration with elements from online gaming has the potential to revolutionize learning by stimulating engagement, enhancing accessibility, and promoting personalized experiences [24]. These advancements offer significant potential for developing innovative methodologies to improve educational quality [25]. Research by [26,27,28,29] indicates that AI and VR technologies can significantly enhance student engagement, improve knowledge retention, and facilitate skill development, underlying the transformative impact of the Metaverse on educational practices, including its ability to create social connections and persistent virtual spaces. Virtual reality and gamified learning environments have been shown to positively impact emotional engagement while reducing anxiety and enhancing learning outcomes [30]. Previous research on the pedagogical potential of immersive technologies has highlighted several critical factors influencing user adoption and learning outcomes, including interaction design, system usability, privacy and security concerns, accessibility, inclusivity, and the social dynamics embedded within virtual spaces [31,32,33,34]. While some authors have focused on practical aspects such as the impact of immersive experiences on user engagement and motivation [35,36,37,38], others have examined theoretical frameworks or other methodological approaches to better understand user acceptance, human-technologies interaction, and to provide robust support and guidance for practical applications and decision-making processes [39,40,41].

AI technologies, particularly those based on generative models, enable real—time personalization, adaptive feedback, and autonomous generation of learning content. Furthermore, such systems promote intercultural teamwork [42] and personal learning through context-specific environments and risk-free skill simulations [43], paving the way toward a truly engaging and empowering future of education [44].

Unlike traditional digital-based education, the Metaverse offers an immersive, continuous learning environment across various fields. Recent studies highlight that extended reality (XR) technologies enable the integration of heterogeneous digital representations and interactive environments, supporting immersive and multidisciplinary learning experiences across domains such as education, healthcare, creative industries, and engineering [45,46]. For example, using VR, the student is engaged in medical or nursing skills training [5,47], participating in real-time surgical procedures [48], military training [49], industrial manufacturing training and maintenance [50]. A significant use case is language learning in the Metaverse, which, for example, allows EFL (English as a Foreign Language) learners to immerse themselves in an English-speaking virtual environment, integrating learning with daily activities [51]. Other notable applications include visualizing historical events and figures, enabling students to witness virtual recreations of historical events and interact with historical characters [52]. In medical education, students can explore the human body in 3D, performing experiments on digital objects to understand complex anatomical structures [53,54]. Interactive science experiments using AR/VR allow students to simulate physical phenomena and observe chemical reactions, fostering hands-on learning [55,56]. Virtual trips to global locations, including inaccessible sites like the Great Barrier Reef, provide instant access to detailed information [57]. Virtual laboratories offer a safe, controlled environment for conducting experiments, complementing physical labs [58].

The integration of AI and immersive technologies in the educational process enhances users’ engagement and learning outcomes [59]. Research indicates that these technologies increase motivation and improve academic performance [60]. The integration of AI in education opens new possibilities, such as personalized learning experiences, real-time feedback, automation of routine tasks, and in-depth analysis of learner needs. AI-driven adaptive learning tools can dynamically adjust content based on individual progress, providing instant feedback and customized instructional materials.

Despite their advantages, these new technologies encounter several challenges that delay effective implementation in education. A primary concern is the cost, as the necessary hardware and software can be prohibitively expensive for many academic settings [53]. Accessibility and cybersecurity are critical concerns: unequal access to VR/AR technology may worsen educational inequality [61,62], while data privacy and security, especially for children, require careful attention [63,64,65]. Besides the cost, AI raises many ethical challenges and risks, such as integrity, transparency, and reliability of AI-generated content. In the context of generative models embedded in educational XR experiences, an algorithm may curate or synthesize information, which may lack clear provenance or factual validation, or introduce cognitive biases, which may introduce a vulnerability to information disorder and undermine the epistemic truth. Current studies have advanced our understanding of AI-generated content in immersive technologies’ educational context, but a more nuanced view of how adoption factors affect this process, especially from the user acceptance, is needed. Additionally, research has largely focused on factors that facilitate adoption, often neglecting user resistance and barriers to non-adoption.

2.2. Need for a New Model

Ref. [66] found that students expressed greater enjoyment and a stronger intention to use VR for learning, noting that VR’s realistic simulations and interactive features enhance engagement. Refs. [64,67] observed that user-friendly VR experiences lead to more favorable attitudes and higher adoption intentions, underscoring the importance of perceived ease of use in technology acceptance.

In the process of testing the developed application, a broader TAM, TRA, TPB perspective on the evaluation of educational applications based on virtual reality was adopted. The Technology Acceptance Model is a widely used framework for predicting technology adoption and user behavior [68]. It has been extensively applied in educational technology research [69,70], though studies suggest that additional context-specific factors may be needed for a more complete understanding [13]. Research indicates that students generally favor technology in education, with VR being particularly engaging and effective for immersive learning [71]. The TAM states that perceived usefulness and ease of use are key determinants of users’ attitudes and intentions toward a system [19]. Usefulness is shaped by ease of use, and also by external factors such as user characteristics, training, system development, and organizational support. It reflects the extent to which a person believes that a system will enhance performance, while ease of use refers to the expected level of effort required to operate the technology efficiently. This framework enables researchers to anticipate adoption behaviors after users have had the opportunity to interact with a system.

In turn, the TRA is used by social sciences to predict human behaviors based on their previous intentions and experiences, and it was introduced by [20] and extended later by [21,72]. TRA adds subjective norms (the social pressure to perform a behavior or not) to the attitude toward behavior (favorable or unfavorable) and considers these two variables to be the main predictors of behavior. An important extension of TRA is the TPB [22], which introduces another important variable—perceived behavioral control, the perception of the difficulty of performing an action. Individuals’ experiences and the anticipated obstacles condition this perception.

Existing technology adoption models, such as TAM, TRA, and TPB, have been widely used to study digital learning environments. However, these do not completely capture the multisensory and immersive nature of immersive technologies, where both cognitive and physical experiences influence user engagement. While previous models emphasize usability and intention, they do not address the critical role of ergonomics in VR-based education. This study proposes a new model, ED-SCALE, which integrates key adoption factors from prior models while introducing ergonomics as a core determinant of acceptance in VR learning environments.

Usability frameworks such as Nielsen’s Heuristics for User Interface Design [73] and ISO 9241-11 [74] usability principles have been widely applied in evaluating digital interfaces. These models focus on core factors like learnability, efficiency, error prevention, and satisfaction, which are critical to general software usability. Traditional usability models do not account for physical strain factors, since they were designed for 2D screen-based interactions rather than fully immersive environments. Although these usability principles were considered for measuring our participants’ satisfaction, AI&VR-based learning environments introduce new dimensions of user experience that go beyond traditional usability metrics.

2.3. Brief Comparison Between ED-SCALE and Existing Models

Technology acceptance in AI&VR-based educational environments has traditionally been analyzed using well-established models, including TAM, with its extensions TAM2 [75] and TAM3 [76], TRA, and TPB. Each of these models provides a framework for understanding user intentions, motivations, and barriers in adopting new technologies. However, these models were primarily designed for generic digital technologies and do not fully account for the complex, immersive, and multisensory nature of merging technologies-based education.

To address these context-specific limitations, the ED-SCALE model expands on them by integrating an additional key dimension—Ergonomics (E)—that is critical for user acceptance in VR environments but remains underdeveloped in prior models. Below, we outline how ED-SCALE improves upon TAM, TRA, and TPB in the context of VR-based learning:

Comparison with TAM. While TAM effectively predicts the acceptance of general digital tools, it does not fully account for the unique physical, cognitive, and sensory challenges posed by immersive VR environments. VR-based learning applications require not only intuitive navigation but also physical comfort, reduced motion sickness, and interaction with 3D spaces, aspects that are largely absent from the original TAM framework. Moreover, TAM assumes that ease of use is primarily a cognitive factor (e.g., interface simplicity, learnability), whereas in VR, ergonomics influences whether users engage with or abandon an application. In ED-SCALE, ergonomics (E) expands TAM’s focus on usability by integrating physical comfort, immersion fatigue, and device adjustability into the technology acceptance equation.

Comparison with TRA. It suggests that behavioral intention is the primary predictor of actual behavior, meaning that if an individual has a favorable attitude toward using a technology and perceives social approval, they are more likely to adopt it. While TRA is effective for understanding general decision-making processes, it does not account for external barriers such as usability constraints, technological complexity, or physical challenges, which are particularly relevant in the context of VR-based learning environments. ED-SCALE provides a more holistic framework for VR-based education, considering both psychological and technological factors that affect user acceptance and learning outcomes.

Comparison with TPB. The Theory of Planned Behavior states that user behavior is influenced by their belief in their ability to perform an action. However, TPB lacks technology-specific determinants, making it less effective for assessing VR-based learning environments. While TPB considers perceived ease of performing an action, ED-SCALE breaks this down into two complementary dimensions:

-

Perceived Behavioral Control (C) → ease of interaction with the VR environment (software side);

-

Ergonomics (E) → ease of physical immersion and device adaptability (hardware side).

This dual approach offers a more precise explanation for why some users disengage from VR applications due to physical discomfort, even when they recognize its educational benefits.

Compared with existing models, the proposed ED-SCALE model integrates a broader range of variables, capturing the complex interplay of factors that influence technology acceptance in learning processes as outlined in

Table 1. Comparison of ED-SCALE with foundational models.

Dimensions	TAM	TRA	TPB	ED-SCALE
User Attitude (A)	X	X	X	X
Social Influence (S)	-	X	X	X
Perceived Behavioral Control (C)	-	-	X	X
Immersive Device Ergonomics (E)	-	-	-	X
Engagement and Learning Effectiveness (L)	-	-	-	X

:

By integrating ergonomic considerations (E) alongside perceived usefulness, behavioral control, and social influence, ED-SCALE provides a more adapted model for evaluating VR-based educational adoption, ensuring both usability and physical comfort are considered in long-term adoption.

3. Methodology for Evaluating Interactions in AI-Driven Immersive Reality

3.1. The New ED-SCALE Research Model

Based on previous studies [77,78] and existing approaches in the literature, in this study, we propose a model that builds upon established technology acceptance frameworks (TAM, TRA, TPB) to understand the way ergonomics of virtual reality devices and subjective contexts of individuals may influence their behavior.

Upon the variables of previous theories, we expanded by introducing an additional one that we consider at least as important for the acceptance of immersive technologies, which we named Ergonomics. This category includes those characteristics of immersive devices that offer a comfortable and tireless experience and are easy to use without requiring considerable effort on the part of the individual.

Factors influencing user acceptance include attitudes, preferences, behaviors, needs, and experiences as individuals interact with virtual environments. With recent advancements in virtual and augmented reality, understanding these factors leads to more intuitive and engaging Metaverse experiences [79,80]. Studies indicate that 80% of users are more likely to engage with platforms that prioritize customer-centric experiences [81]. Tailoring virtual spaces and interactions to individual preferences is emerging as a highly effective strategy for boosting user engagement and loyalty.

Virtual experiences are shaped by various factors that collectively influence users’ acceptance of technology. To effectively evaluate this impact, researchers require appropriate tools that offer a comprehensive perspective on assessing virtual reality-based educational applications. The present paper is based on an exploratory study, utilizing questionnaires as the assessment tool for data collection.

The primary aim of this study is to explore user interactions and expectations within a 3D educational escape room in the AI-driven immersive reality. We propose a new research model (ED-SCALE) that combines variables proposed by TAM, TRA, and TPB, to which we added and emphasized the ergonomics of virtual reality instruments as a new variable that we consider very important for user acceptance. By employing the proposed research model, this work evaluates user acceptance and learning outcomes in virtual reality environments. The study focuses on understanding how user-centered design can enhance the educational effectiveness of VR applications. Through quantitative data collection, including surveys and observational data, the research identifies key factors that influence acceptance and the overall success of immersive educational tools. Analyzing user experience and educational impact, this study contributes to the advancement of AI-driven immersive technologies in educational settings, ensuring they are both effective and engaging for diverse learner needs.

Based on this model and the analysis of other virtual reality application evaluation tools, a VR-based educational application evaluation framework was developed to assess user experience, engagement, and learning outcomes. This framework integrates multiple criteria, including usability, perceived usefulness, ease of use, and satisfaction, to provide a robust mechanism for evaluating the effectiveness of VR educational tools. Data collected from the questionnaires were rigorously analyzed using statistical methods to validate the reliability and validity of the evaluation framework.

Our conceptual model has 6 dimensions influencing behavioral intention: subjective norms, perceived behavioral control (perceived difficulty), attitude toward behavior, perceived effectiveness (learning effectiveness), ergonomic quality, and adding social context (Figure 1).

Subjective norms (S) are the individual’s perceptions of whether the people they consider important believe they should engage in a specific behavior or not, and also one’s personal evaluation to comply or not with these expectations [21] (p. 302). In traditional e-learning models, social influence is mainly limited to instructor and peer recommendations. Prior studies have shown that perceived social influence significantly affects students’ willingness to adopt emerging learning technologies [82]. In collaborative virtual spaces, avatars and AI-driven learning experiences shape new social norms.

Control (C) is a model dimension based on perceived behavioral control (TPB) or perceived ease of use (TAM), referring to the degree to which a person believes that using a particular system would be effortless. In the case of virtual reality-based e-learning applications, C is related to the ease of understanding and use of the main functions of the system. Students must feel confident when operating, customizing, and troubleshooting VR devices [83] while they acquire new navigation, gesture control, and spatial interaction skills.

Attitude toward behavior (A) refers to a favorable or unfavorable assessment of the behavior, including an evaluation of the possible outcomes of either performing a behavior or not. Research indicates that affective engagement in VR-based learning leads to higher motivation and long-term retention [4,84,85,86]. VR environments trigger stronger emotional responses than traditional learning platforms, and positive emotions can increase willingness to adopt VR, while negative emotions may create barriers to acceptance.

The perceived learning effectiveness (L) of educational applications based on virtual reality is measured by the rate of achievement of the educational objectives proposed by the learners. In this research, through learning effectiveness, we try to measure the perception of educational utility of applications based on virtual reality. Immersive learning environments can increase knowledge retention compared with traditional learning [85].

The ergonomics of VR devices (E) is an important component of usability. In this context, the usability of virtual reality applications must be as high as possible, especially since these applications are very complex and need robust devices to run smoothly. Unlike standard e-learning, VR requires users to wear headsets, use controllers, and move physically, which introduces new usability challenges. Poor ergonomics can stimulate cyber sickness [86], reduce session duration, increase fatigue, and negatively impact adoption rates. Nevertheless, usability is a necessary but not sufficient condition for virtual reality applications to be accepted by learners.

The intention to use (INT) of VR applications in education is the next step towards their effective use (ED). The actual use of a VR system is determined by the behavioral intention to use, which is determined by the joint action of two factors: perceived usefulness and attitude.

These variables have been complemented by the ergonomics variable of virtual technology, resulting in a model that explains the perception and acceptance of technology, which we have named ED-SCALE (Figure 1).

3.2. The Procedure Methodology

To empirically assess technology acceptance in AI-driven -based education, we conducted a pilot study using ICI EDscape Room, an AI-driven virtual reality educational application that we developed in Unity to simulate interactive learning experiences (Figure 2). This application was designed as an immersive escape room where participants engage in problem-solving tasks related to ICT knowledge and digital competencies. The development was based on the framework design for reinforcing the potential of XR technologies in transforming inclusive education. The ICI EDscape Room features interactive challenges and real-time decision-making, aiming to enhance user engagement and retention of educational content.

The app integrates curated AI elements, ensuring that the AI-driven components of the application provide accurate, unbiased, and pedagogically sound learning experiences. The AI elements included in the application were created using the Llama-2-7B-Chat model, which was deployed and executed locally. The Llama-2-7B-Chat model was a transformer-based large language model with 7 billion parameters, fine-tuned using supervised instruction tuning and reinforcement learning from human feedback (RLHF) to generate conversational responses.

Running the model on local infrastructure removed the need to send user inputs, learning progress, or contextual educational information to external servers, thereby improving data integrity and confidentiality while supporting institutional privacy requirements. This approach ensured that prompts, intermediate processing steps, and generated responses remain within the controlled environment of the application, reducing risks associated with third-party data handling. In addition, the local deployment made it possible to configure model parameters in a stable and predictable way while using curated prompt templates and domain-specific safeguards aligned with the pedagogical objectives of ICI EDscape Room. The model operated within a constrained knowledge context and relied on carefully designed prompts to reduce hallucinations and maintain consistency with validated educational content. Furthermore, running the system in an offline or restricted-network mode prevented unintended updates, limited exposure to external content changes, and avoided dependence on evolving cloud-based AI services. This setup helped preserve consistency and reproducibility across learning sessions while maintaining control over both the generated content and the data used during interaction.

Our study employed this application as a controlled test environment to evaluate user acceptance, perceived usability, and learning effectiveness within AI-driven VR education. By applying the ED-SCALE model, we analyzed participants’ interactions and behavioral responses to assess the critical factors influencing the adoption of virtual learning technologies (Figure 3).

The research aimed to evaluate the usability and learning outcomes of participants through an exploratory pilot study. The exploratory design included a single group of participants who used the ICI EDscape room and then completed a questionnaire. Before data collection, participants were familiarized with virtual reality technology, focusing on the use of the educational application developed for Meta Quest 2 devices.

Then, each participant tested the application based on the tasks received. App usage time was quantified for each individual participant. The subjects were asked to explore the features offered by the app for each task.

After testing, participants were asked to respond to the evaluation questionnaire by rating the items on a 5-point Likert scale (1—strongly disagree, 2—agree, 3—neutral, 4—agree, and 5—strongly agree).

3.3. The Data Collection Tool

To empirically test the ED-SCALE model, we developed an evaluation instrument which includes specific questions for each dimension of the proposed model [78] and also descriptive items related to age, gender, education level, and previous experience with VR applications.

Table 2. Model dimensions and variables.

Dimensions/Constructs	Items	Variables
Subjective Norms (S)	S1	I have friends who have used virtual reality already.
	S2	Generally speaking, I care a lot about what my friends are expecting for me to do.
Perceived Behavioral Control (C)	C1	The AI-driven virtual reality application is intuitive.
	C2	It is easy to interact with AI-personalized virtual reality content.
	C3	In the future, it will be easy to remember how to use AI-driven virtual learning environments.
	C4	Identifying and understanding the cues in AI-generated learning scenarios is easy.
	C5	Reading AI-generated educational content is easy.
	C6	Selecting a menu item is easy.
Attitude toward Behavior (A)	A1	The virtual learning environment is comfortable.
	A2	I like testing AI-enhanced educational applications.
	A3	I would like to frequently use AI-generated virtual learning environments.
Perceived Learning Effectiveness (L)	L1	Using the AI-driven VR application helps in quicker knowledge assimilation.
	L2	Learning can be more engaging using the AI-personalized VR content.
	L3	Overall, I believe the AI-driven VR system will be useful for learning.
	L4	The feeling of immersion in the virtual world was intense.
Ergonomics (E)	E1	Adjusting virtual reality (VR) devices is easy.
	E2	Using the controllers in the VR application is easy.
	E3	Adjusting the sound is easy.
	E4	Observing virtual elements is clear.
	E5	I felt comfortable with the headset and controllers.
	E6	I experienced dizziness during use.
Intention to Use in Education (ED)	ED1	I intend to use AI-driven VR applications for learning.
	ED2	I will recommend other colleagues to use AI-generated learning content in VR applications.

presents the operationalization of constructs.

In addition, each participant was asked to self-rate the time spent using the VR app.

To identify the latent structure of the proposed ED-SCALE framework, an Exploratory Factor Analysis (EFA) was conducted on responses from participants using Principal Axis Factoring with Varimax rotation. This procedure was selected to identify the underlying dimensions of the instrument and to obtain a simplified and interpretable factor structure. For interpretation purposes, factor loadings of 0.40 or higher were considered meaningful.

3.4. Participants

All participants were volunteers, recruited at different technical events between April 2024 and November 2025. The decision to select participants from tech events is because immersive technologies are still in an early stage of adoption in Romania. Considering the curve of innovation adoption [87], we targeted this category of participants based on the premise that they represent the pioneers in using AI and implementing immersive technologies in education. Evaluating their perceptions through this exploratory study allows us to better understand the pace at which this innovation can be integrated into the Romanian education system. The sample consists of 213 people—94 women and 119 men—and the participants’ ages ranged from 20 to 60 years, with an average age of 36.5. Among the participants, 113 were familiar with AI and VR technology, while 100 reported having never used AI-driven VR applications.

4. Results

4.1. Exploratory Factor Analysis (EFA)

The adequacy of the dataset for factor analysis was first assessed using the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity. The KMO value reached 0.862, indicating very good suitability of the data for factor extraction. In addition, Bartlett’s test of sphericity was statistically significant (χ² = 2237.870, df = 253, p < 0.001), confirming that the correlation matrix contained sufficient inter-item correlations to justify the application of exploratory factor analysis.

An exploratory factor analysis was conducted to examine the underlying structure of the ED-SCALE instrument. This solution is consistent with the theoretical conceptualization of the ED-SCALE model, which includes Subjective Norms (S), Perceived Behavioral Control (C), Attitude toward Behavior (A), Perceived Learning Effectiveness (L), Ergonomics (E), and Intention to Use in Education (ED). The retained factor solution accounted for 47.15% of the total variance, as in

Table 3. Cumulative variance for the six-factor proposed model.

Factor	Variance Explained (%)	Cumulative Variance (%)
S	29.75	29.75
C	6.12	35.87
A	4.41	40.27
L	2.68	42.96
E	2.58	45.54
ED	1.61	47.15

. Although moderate, this level of explained variance is considered acceptable in exploratory research, particularly given the complexity and multidimensionality of user perceptions in immersive learning environments, and suggests a reasonably latent structure underlying the measured constructs.

While the EFA extracted six factors, the empirical structure exhibited a degree of conceptual overlap rather than perfectly discrete dimensions. Specifically, items related to perceived usefulness, attitudes, and behavioral intentions loaded onto shared factors. This alignment reflects the theoretical interdependence inherent in behavioral models (e.g., TAM), where cognitive evaluations and intentionality are naturally conflated. Such findings are consistent with the exploratory nature of this study and suggest that, in this specific context, participants may perceive these constructs as a single, integrated evaluative dimension.

4.2. Factorial Structure and Construct Validity

Although the Exploratory Factor Analysis (EFA) extracted six factors, according to

Table 4. Rotated factor matrix.

	Factor
	1	2	3	4	5	6
L2	0.915
L1	0.895
L4	0.879
L3	0.855
ED2	0.782
A3	0.760
ED1	0.688
A2	0.677
C6	0.433
A1		0.753
E5	0.435	0.478
C4			0.658
E2			0.400
E3				0.816

, the empirical structure did not replicate the theoretical dimensions in a discrete manner. Instead, several items from conceptually related constructs loaded together, indicating a partial overlap between perceived usefulness, favorable attitudes, and future behavioral intention. Such patterns are common in exploratory research involving newly developed behavioral models, where constructs are often theoretically interconnected and experienced holistically by the participants.

Factor 1: Educational Value and Adoption Intention is the most substantial, primarily defined by items L1, L2, L3, L4, ED1, ED2, A2, A3, C6, and E5. This factor integrates educational effectiveness, motivational engagement, and future adoption intention. The results suggest that respondents tend to perceive educational value and their readiness to use the technology as a single, integrated evaluative dimension rather than separate cognitive processes.

Factor 2: Comfort and Affective Experience integrates comfort-related evaluations (E5) with affective attitudinal components (A1). Notably, item E5 exhibited cross-loading between Factor 1 and Factor 2, suggesting that physical comfort in an immersive environment serves both as a functional requirement and an affective driver of the overall experience. This indicates that physical ergonomics are intrinsically linked to the user’s emotional state during the simulation.

Factor 3: Operational Ergonomics and Interaction Control, defined by items C4 and E2, focuses on the practical interaction with VR hardware, including device adjustment and controller use. This dimension confirms the presence of a distinct operational component related to the ease of hardware interaction.

Factor 4: Navigation Clarity and Spatial Confidence (represented by item E3) is linked to interaction cues and navigation clarity. It reflects the user’s ability to identify signals and understand the virtual environment, representing a specific component of interaction self-efficacy.

Factors 5 and 6 are residual latent components, show limited item saturation, and have lower eigenvalues, preventing robust standalone interpretation. However, their emergence suggests the existence of minor latent nuances within the ED-SCALE framework that may require further refinement or a larger sample size to reach statistical maturity.

4.3. Discussion on Model Fit

While our data shows that users link attitude and intention, this overlap does not undermine the original ED-SCALE model. Instead, it shows how people experience AI-based immersive environments as a whole. We will continue to use the six theoretical dimensions for the rest of our analysis, as they allow us to look at the details of the user experience. Adopting the aggregate factors suggested by the data would mask important nuances and make it harder to compare our findings with future studies. Furthermore, maintaining the original structure allows for identifying exactly where the scale can be further examined in future confirmatory studies using CFA or SEM techniques.

4.4. Reliability Analysis

To assess the internal consistency of the questionnaire, Cronbach’s Alpha coefficient was calculated. The scale yielded a coefficient of 0.803, indicating good internal reliability. This value suggests that the instrument items are highly intercorrelated and measure the intended construct consistently.

4.5. Preliminary Assessment of Discriminant Validity

The EFA results provide preliminary evidence for the discriminant validity of the ED-SCALE framework, as participants clearly distinguished between educational value and technical usability. Specifically, Perceived Learning Effectiveness emerged as the dominant factor, while ergonomic indicators and interaction control cues, such as hardware handling and navigation clarity, demonstrated separate empirical identities. This separation confirms that users perceive the system’s operational comfort as a distinct component from its pedagogical impact.

In contrast, a partial overlap was observed among Attitude toward Behavior, Perceived Learning Effectiveness, and Intention to Use in Education, which clustered into a broader acceptance dimension. This suggests that these constructs operate as interconnected facets of a holistic user experience rather than entirely independent processes. While the weaker saturation of Subjective Norms and Perceived Behavioral Control suggests a need for refinement, the overall structure supports the model’s multidimensional premise.

4.6. Descriptive Statistics

Table 5. Descriptive statistics for the ED-SCALE model.

	N	Minimum	Maximum	Mean	Std. Deviation
S1	213	1	5	2.1362	0.80988
S2	213	1	5	3.2535	0.77188
C1	213	2	5	4.3192	0.65980
C2	213	2	5	4.3850	0.77834
C3	213	3	5	4.3756	0.64411
C4	213	1	5	4.2311	0,89144
C5	213	2	5	4.3005	0.70299
C6	212	2	5	4.5283	0.64884
A1	213	2	5	4.2441	0.79889
A2	213	2	5	3.6485	0.64853
A3	213	1	5	3.9718	0.90552
L1	213	1	5	3.8779	0,73596
L2	213	1	5	3.6714	0.87144
L3	213	1	5	4.4460	0.68208
L4	213	3	5	4.4413	0.65343
E1	213	2	5	4.5305	0.64080
E2	213	2	5	4.2770	0.84858
E3	213	2	5	3.9155	0.70201
E4	213	2	5	4.4883	0.58769
E5	213	1	5	4.3568	0.74252
E6	212	2	5	4.1132	0.69964
ED1	213	1	5	3.5540	1.02015
ED2	213	1	5	4.0939	0.92686

shows the descriptive statistics indicators calculated based on the answers given by the participants. As can be seen, the mean values for the ergonomics items (E1–E6) are relatively high (between 3.98 and 4.42), indicating a general positive perception of the ergonomics of virtual reality devices. Item E4 (“Observation of virtual items is clear”) has the highest mean (4.4231), suggesting that participants consider the observation of virtual items to be clear. The lowest mean is recorded for item E3 (Sound adjustment is easy), at 3.9808, and suggests that participants find sound adjustment relatively easy, but not as easy as other aspects of ergonomics.

Subjective Norms Cluster (S1–S2): The values for these two questions show how social perceptions and group pressures can influence the use of VR technology (H1). The low mean of responses to the question S1 “I have friends who have used virtual reality already” indicates that exposure to virtual reality is limited among respondents in this sample. At the same time, respondents acknowledge being influenced by their friends’ expectations, which shape their behaviors (S2).

Perceived Behavioral Control (C1–C6): The high means for items in this cluster (around 4.23 to 4.52) indicate that participants perceive the AI-driven VR application as easy to use and intuitive, and they have a strong perception of control during application use.

Attitude Toward Behavior Cluster (A1–A3): The mean scores for questions in this cluster also reflect a generally positive attitude toward using VR applications, which may result from the participant’s tech events profile (maybe more prone to technology). A high-to-medium score on A2 suggests openness to experimenting with new technologies, while a high score on A3 indicates a strong intention to use AI and VR frequently in the future; however, the results also imply the presence of a certain level of apprehension that tempers responses, preventing them from being overwhelmingly positive. The intention to frequently use AI&VR in the future strongly correlates with the Perceived Behavioral Control (C) cluster.

Learning Perceived Effectiveness Cluster (L1–L4): High scores in this cluster reflect perceptions of the AI-based VR application’s pedagogical effectiveness, particularly in enhancing engagement and motivation for learning. Additionally, high scores in this area suggest that participants recognize VR’s positive impact on the learning process, with L3 and L4 likely yielding the highest average scores—highlighting an appreciation for the technology’s utility and immersive nature. A significant correlation underlines the connection between the perception that AI-driven VR systems are useful for learning and the perceived ease of interaction with AI-personalized VR content (with C1 (r = 0.269, p ≤ 0.000) and with C6 (r = 0.426, p ≤ 0.000)).

Ergonomics Cluster (E1–E6): The average mean values for this cluster indicate a generally positive perception regarding the ergonomics of our AI-driven virtual reality devices (mean values ranging from 3.91 to 4.53). The highest means were recorded for E1 (4.5305) and E4 (4.4231), which suggests that the clarity of virtual elements is good and easy to adjust, and the lowest average score was for E3 (3.9155), indicating that sound adjustment should be more attentively constructed. Significantly variable results were recorded for E6, which reflects discomfort (dizziness), suggesting that some participants experienced physical difficulties related to the use of VR devices.

Furthermore, a considerable number of items were found to correlate strongly with the six variables included in the ergonomics cluster (E1–6), providing robust evidence of the central role of this construct, as in

Table 6. Correlations of Ergonomy cluster with intention to use VR in education (** correlation is significant at the 0.01 level (2-tailed); * correlation is significant at the 0.05 level (2-tailed)).

	E1	E2	E3	E5	ED 1	ED 2
E1 Pearson Correlation	1	0.440 **	0.118	0.383 **	0.294 **	0.310 **
Sig. (2-tailed)	–	0	0.086	0	0	0
N	213	213	213	213	213	213
E2 Pearson Correlation	0.440 **	1	0.142 *	0.464 **	0.255 **	0.275 **
Sig. (2-tailed)	0	–	0.038	0	0	0
N	213	213	213	213	213	213
E3 Pearson Correlation	0.118	0.142 *	1	0.121	0.149 *	0.168 *
Sig. (2-tailed)	0.086	0.038	–	0.079	0.029	0.014
N	213	213	213	213	213	213
E4 Pearson Correlation	0.211 **	0.125	0.054	0.247 **	0.071	0.078
Sig. (2-tailed)	0.002	0.069	0.437	0	0.303	0.257
N	213	213	213	213	213	213
E5 Pearson Correlation	0.383 **	0.464 **	0.121	1	0.313 **	0.371 **
Sig. (2-tailed)	0	0	0.079	–	0	0
N	213	213	213	213	213	213
E6 Pearson Correlation	0.098	0.004	−0.03	0.097	0.035	0.048
Sig. (2-tailed)	0.157	0.955	0.661	0.161	0.615	0.484
N	212	212	212	212	212	212
ED 1 Pearson Correlation	0.294 **	0.255 **	0.149 *	0.313 **	1	0.623 **
Sig. (2-tailed)	0	0	0.029	0	–	0
N	213	213	213	213	213	213
ED 2 Pearson Correlation	0.310 **	0.275 **	0.168 *	0.371 **	0.623 **	1
Sig. (2-tailed)	0	0	0.014	0	0	–
N	213	213	213	213	213	213

. This convergence suggests that ergonomics is not a peripheral factor, but rather a key dimension influencing users’ perceptions and interactions with immersive technologies. Among these correlated items, particular attention is given to the intention to use in education such technologies, which emerges as a highly relevant indicator. This reinforces the argument that ergonomics should be systematically incorporated into any theoretical model aiming to explain and understand the acceptance of immersive technologies, as it captures essential aspects that directly influence users’ willingness to adopt and continue using these systems.

Intention to Use in Education Cluster (ED1-ED2): This cluster measures the participant’s intention to continue using generative AI and VR technology and to recommend it to others. High scores indicate a strong tendency toward the adoption and promotion of these technologies in the future.

A significant positive correlation was found between feeling comfortable with the headset and controllers (E5) and: 1. the intention to use AI-driven VR for learning (ED1) (r = 0.313, p ≤ 0.000) and 2. recommendation to use this technology to other colleagues (ED2) (r = 0.371, p ≤ 0.000).

Age:

An inversely proportional correlation was identified between age and several key perceptions related to AI-driven VR. Specifically, as age decreases, respondents report a higher perceived ease of use of VR devices, including adjusting the equipment and using controllers (C1: r = −0.194, p ≤ 0.005; C4: r = −0.189, p ≤ 0.006). Younger participants also perceive VR as more comfortable to use, indicating greater agreement with feeling at ease with the headset and controllers (A1: r = −0.555, p ≤ 0.000; A2: r = −0.314, p ≤ 0.000) and intention to use them frequently (A3: r = −0.456, p ≤ 0.000). In addition, age is negatively correlated with the perceived effectiveness of VR for learning, as younger individuals are more likely to agree that AI-personalized VR content enhances engagement (L1: r = −0.432, p ≤ 0.000; L2: r = −0.399, p ≤ 0.000; L3: r = −0.446, p ≤ 0.000; L4: r = −0.501, p ≤ 0.000). This can be explained by the fact that older participants developed their learning habits during periods when such technologies were not available, relying instead on more traditional study methods. As a result, they may find it more difficult to adapt to and fully appreciate newer, immersive tools, whereas younger individuals, who are more accustomed to digital environments, perceive them as more engaging and effective. This trend also extends to younger participants’ greater openness toward further experimentation with such technologies and their intention to use VR in educational contexts (ED1: r = −0.277, p ≤ 0.000; ED2: r = −0.350, p ≤ 0.000). They are more likely to express a clear willingness to adopt this technology in the future and to promote it among peers and acquaintances, reinforcing their role as early adopters. Overall, this suggests that younger users not only find VR easier and more comfortable to use but also perceive it as a more valuable and desirable learning tool.

Studies level:

The Pearson coefficient reveals a significant inverse correlation between education level and the perception that using virtual reality devices is easy (E2: r = −0.279, p ≤ 0.000) and comfortable (E5: r = −0.292, p ≤ 0.000). While this finding may seem counterintuitive, it can be understood in the context of digital natives’ greater familiarity with AI and VR technology, suggesting that the ease of adjusting devices is less related to formal education and more to practical experience with ICT devices. In fact, VR technologies appear particularly valuable for individuals with lower formal education, as they can provide an alternative to traditional learning methods and help bridge gaps for those who did not fully engage with conventional study approaches. These results highlight the critical importance of ergonomics, confirming that ease of use and comfort are essential factors in making VR an effective and accessible learning tool for all users, regardless of educational background.

Perception of time in the AI-driven immersive reality:

Table 7. Perception of time in the AI-based VR application.

Time Perception
	N	Minimum	Maximum	Mean	Std. Deviation
Real Time	213	2.00	25.00	11.1737	5.04909
Perceived Time	213	2.00	30.00	9.5587	4.73073

shows the average time (expressed in minutes) spent by study participants to test the AI-based VR application. Actual time is the time measured for each test participant by the researchers. Perceived time is the time each participant estimated they spent using the app, thus reflecting each participant’s subjective perception.

The real time ranged from 2 to 25 min testing the app, while participants estimated that they spent between 2 and 30 min. On average, subjects spent approximately 11 min using the VR app and estimated, on average, that they spent approximately 9.5 min testing the VR app. Interpretation of these values indicates that the perceived time is generally less than the actual time spent using the app. It is possible that immersion in the VR space and active engagement in the app’s activities created the feeling that they spent less time than the actual duration. However, the variation is quite large, suggesting that some subjects may perceive the time spent differently than others.

5. Discussions

This study aimed to validate the proposed ED-SCALE model by examining the factors influencing user acceptance of AI-driven virtual reality in educational contexts. The findings provide empirical support for all six hypotheses, highlighting the combined role of social, cognitive, and ergonomic factors in shaping technology adoption.

H1 assumes that exposure to AI-driven VR learning environments through friends or colleagues positively influences users’ intention to adopt this technology. The results for the subjective norms cluster support this assumption. Although the mean value for S1 was low (2.1362), indicating limited prior exposure to VR among participants’ peers, the mean for S2 (3.2535) indicates that respondents remain sensitive to the expectations of their social circle. This pattern suggests that the relatively limited diffusion of VR in the Romanian context may constrain broader adoption, while also indicating that greater peer exposure could increase users’ willingness to adopt such technologies. Thus, H1 is supported, as the findings point to social visibility and peer familiarity as relevant drivers of intention to adopt AI-driven VR.

H2 stated that the use of AI-generated content in VR increases engagement and motivation for learning. Participants reported positive evaluations for items related to quicker knowledge assimilation (L1 = 3.8779) and enhanced engagement through AI-personalized VR content (L2 = 3.6714), indicating that AI-supported immersive content was perceived as motivating and conducive to active learning. These findings suggest that AI-generated educational content contributes to a more stimulating learning experience, thereby supporting H2.

H3 posited that the use of AI-driven virtual reality in education improves engagement and the perception of learning effectiveness. The highest mean values within the learning cluster were recorded for L3 (4.4460), reflecting the perceived usefulness of the AI-driven VR system for learning, and L4 (4.4413), reflecting the intensity of immersion. Together, these results indicate that participants perceived the educational escape room not only as engaging but also as pedagogically valuable. Therefore, beyond motivation alone, AI-driven VR appears to strengthen the perceived effectiveness of learning, validating H3.

H4 assumed that a higher perception of control over AI-personalized VR interactions increases the likelihood that users will want to use the technology frequently. The results clearly support this hypothesis. All items in the perceived behavioral control cluster showed high mean scores, indicating that the system was considered intuitive and easy to use. More importantly, the intention to frequently use AI- and VR-based applications (A3) was significantly correlated with several control-related variables, including intuitiveness (C1: r = 0.285, p ≤ 0.000), ease of understanding cues (C4: r = 0.220, p ≤ 0.001), and ease of menu use (C6: r = 0.385, p ≤ 0.000). These findings confirm that when users perceive a strong sense of control over interaction, they are more likely to express a desire for frequent future use.

H5 assumes that users with a positive attitude toward new technologies will perceive interaction with AI-generated VR environments as more intuitive and easier to use. The attitude cluster reflected generally favorable views toward AI-enhanced educational applications, particularly in relation to openness to experimentation and willingness to use such environments in the future. In addition, the Results section reports significant correlations between the perception that the AI-driven VR system is useful for learning (L3) and ease-of-use indicators such as C1 (r = 0.269, p ≤ 0.000) and C6 (r = 0.426, p ≤ 0.000). Within the logic of the model, these associations indicate that a favorable disposition toward emerging technologies is accompanied by stronger perceptions of intuitiveness and usability. Therefore, H5 is supported.

H6 stated that the level of physical comfort experienced during VR use positively influences users’ intention to adopt the technology in the long term. Among all hypotheses, this one receives particularly strong support. The ergonomics cluster recorded consistently high mean values, showing that participants generally evaluated the devices as easy to adjust, clear in visual display, and comfortable to use. Most importantly, significant positive correlations were found between comfort with the headset and controllers (E5) and both intention to use AI-driven VR for learning (ED1: r = 0.313, p ≤ 0.000) and intention to recommend the technology to colleagues (ED2: r = 0.371, p ≤ 0.000). These results confirm that physical comfort is not a secondary usability issue, but a central predictor of long-term adoption in immersive educational environments.

6. Limitations and Future Work

A primary limitation of this research is the absence of confirmatory structural modeling; however, given the early stage of the scale’s development, an exploratory approach was deemed more appropriate to identify initial construct interactions before imposing the rigid constraints of SEM.

Another limitation is the lack of a control group for comparison with non-AI&VR learning methods. However, the study specifically aimed to evaluate user perceptions after firsthand experience with VR headsets in an AI-driven e-learning application. Unlike comparative learning studies, which assess knowledge retention or performance differences between AI&VR and traditional methods, this research focuses on user acceptance, usability, and perceived effectiveness within a fully immersive learning environment. Future studies could explore comparative experimental designs, but such an approach would require a different research focus, shifting from perception-based evaluation to an outcome-based assessment of learning effectiveness. In the context of this study, the absence of a control group does not diminish the validity of the findings, as the primary objective was to assess factors influencing user adoption within a VR-enhanced educational setting through the lens of the ED-SCALE model.

Additionally, recruitment from technology-focused events may have introduced a selection bias, as participants possessed higher-than-average technical proficiency and a positive predisposition toward new technologies. Although the findings offer meaningful insights into AI-driven VR acceptance, they may not be fully representative of a broader educational audience.

Future research will include a more diverse participant sample, incorporating individuals from various educational backgrounds, age groups, and levels of prior VR experience.

Testing the model through confirmatory approaches (CFA/SEM) in diverse educational domains and learning scenarios will refine its predictive accuracy and ensure its practical relevance.

7. Conclusions

The present study reveals valuable insights into user interactions in AI-driven immersive reality. By specifically focusing on users’ interactions and experiences within the AI-driven educational escape room, we worked on systematically evaluating the usability, engagement, and learning outcomes of educational applications based on new technologies. An expected outcome of our research is the critical role of user-centered design for an engaging VR experience. Participants in our study consistently pointed out the ease of navigation and intuitive interfaces as key factors that maintained their engagement and contributed to the overall experience. This aspect reinforces the necessity for VR developers to prioritize intuitive design elements that satisfy user preferences and behaviors.

Our empirical study was based on the educational potential of escape rooms in serious games. The escape room game, blending challenge and interaction, proved to be an effective educational tool that can transform traditional learning paradigms. The immersive nature of these environments not only attracts participants but also significantly improves knowledge retention and engagement. These results are consistent with prior research that highlights the virtues of immersive learning environments for promoting active and experiential learning.

The ED-SCALE research model has proven effective but can benefit from further refinement and expansion. The ERGONOMICS of the devices used emerged as a main element in user satisfaction. Participants reported high levels of comfort and ease of use, suggesting that current technologies are well-received due to their simplicity. We also examined the impact of sociodemographic variables and found that age significantly influences technology acceptance, a finding that can be explained by previous frequent interactions with technology, as well as a highly favorable attitude towards its use among digital natives. They also believe that learning is much more intense and engaging when using immersive tools like those tested in this study. Those who found the experience educationally effective tended to spend more time in the application—not approaching it as mere entertainment but with the seriousness of someone deeply committed to learning.

Future research should explore how a multi-criteria analysis or structural equations modeling can be customized to assess different types of educational content and delivery methods within the AI-/VR-based learning environments. Also, longitudinal studies are needed to assess the long-term impact of such applications on learning outcomes and user engagement. Understanding how sustained usage of such apps influences knowledge retention and skill development over time will provide deeper insights into the effectiveness of emerging technologies in educational settings.

Understanding societal impacts of emerging technologies can guide the development of policies and regulations that support their responsible and equitable growth, ensuring that their benefits are widely distributed and their challenges are effectively managed. We consider that such investigations should be continued, and addressing this gap is the primary aim of our forthcoming study, which seeks to complement existing research in this area.