The Psychological Effects of AI Learning Assistants in Immersive Virtual Reality Environments

Abstract

Artificial Intelligence (AI) and Virtual Reality (VR) are increasingly integrated into education, yet their combined psychological effects remain underexplored. This paper investigates the potential benefits and risks of AI-powered learning assistants within immersive VR environments. The study builds on insights from a previous pilot involving a virtual tour guide for Athens and proposes a case study with 52 high school students. In groups of three, students would use Oculus headsets with an AI assistant (pre-programmed and AI-generated modes), explore content for a week, and complete questionnaires on usability, trust, and psychological impact. The analysis is expected to support a balance of positive outcomes including greater engagement, motivation and autonomy but also negative ones such as over-reliance, diminished critical thinking, and social isolation. The paper also identifies key psychological dynamics, including the critical role of social influence and teacher-led adoption, and the nuanced nature of student trust in AI-generated information. Ethical implications, such as data privacy and the digital divide, are also discussed. The study concludes by proposing that AI-VR can enrich learning, especially in cultural contexts, but requires safeguards for trust, ethics, and accessibility, with further research on long-term effects, psychological impact and cross-cultural and linguistic nuances.

1. Introduction

1.1. The Confluence of AI and VR in Education

The use of Virtual Reality in education has been extensively studied among researchers [1,2,3,4] indicating that immersive Virtual Reality can provide time- and place-independent opportunities to boost engagement and comprehension in diverse settings with personalized feedback.

In parallel, the impact of Artificial Intelligence in education has been widely explored [5,6,7,8,9], revealing that AI offers significant benefits, including personalized learning, improved academic outcomes, and enhanced student engagement through adaptive content and intelligent tutoring systems [10]. This leads to improved student engagement and outcomes, increased global accessibility to education, and enhanced critical thinking skills.

The unification of AI and VR marks a significant paradigm shift in educational practices, moving away from traditional, passive learning models towards active, experiential ones [11]. This intersection enables richer learning experiences where AI adapts the content while VR introduces immersion, supporting tailored and engaging explorations.

Building on this intersection of AI and VR, one of the most promising educational applications is the Virtual Field Trip (VFT). VR is one of the most popular platforms for implementing VFTs, allowing students to explore historical, cultural, and scientific locations in meaningful ways. They can supplement in-person visits by providing access to sites that are too distant, costly, inaccessible, or even non-existent [12,13,14]. VFTs also improve accessibility for students with mobility issues, as many heritage landmarks were built before the introduction of accessibility standards [12,15]. Beyond practical access, the excitement and motivation generated by immersive VFTs are particularly valuable in subjects like history, which often suffer from low engagement and are perceived as overly reliant on memorization [14]. While VFTs do not necessarily require AI, the study on which this paper builds [16] integrated an AI learning assistant into the VR environment, enabling more dynamic and adaptive interaction with the content.

VFTs are one example of how VR can reshape educational content, with history providing a powerful case where students can virtually explore cultural landmarks. By creating a simulated, all-inclusive sensory environment, VR can make abstract concepts concrete and tangible, spark genuine curiosity and facilitate hands-on experiences that were previously confined to physical labs or textbooks [10]. The introduction of an AI learning assistant further amplifies this potential. AI can analyze individual learning styles and performance data to dynamically adjust educational content, providing a tailored learning path that caters to diverse student needs and paces [10]. This fusion of VR’s immersive power with AI’s adaptive capabilities creates a unique learning environment combining immersion and adaptivity. However, while such environments hold promise for education, they also raise complex psychological considerations around motivation, trust and critical thinking, necessitating a detailed investigation.

This paper builds on a project briefly mentioned earlier [16] that developed an AI chatbot within an Oculus VR headset to serve as a learning assistant for exploring the city of Athens, Greece. While the pilot study focused on technical development, the present study adds a psychological dimension, examining how students aged 14 and above experienced its impact on their learning process. The findings of the case study provide a concrete example of how these two technologies are being fused to create a highly personalized and immersive educational tool which can revive historical education by increasing student motivation, curiosity, and access to experiential learning.

This project can be seen as a type of VFT, where students explore Athens’ cultural landmarks in a way that combines pre-programmed knowledge with dynamic AI responses. In this way, cultural heritage serves as a valuable context for studying the psychological benefits of AI-VR learning assistants, particularly their ability to enhance motivation, curiosity and trust, directly addressing the long-standing challenge of disengagement in history and cultural heritage education.

Therefore, the primary objective of this Conceptual and Methodological Proposal is to synthesize these psychological components and propose a robust case study methodology to empirically test these effects on high school students, directly bridging this theoretical and empirical gap. This synthesis is particularly crucial given the recent comprehensive overview of ethical AI use in education by Johnson & Lee [17].

1.2. Research Objectives

The primary objective of this paper is to analyze the psychological impacts of AI-powered learning assistants within immersive VR environments on adolescents aged 14 and older. To achieve this, the report will:

•

Identify and categorize the positive psychological effects, including cognitive, emotional, and motivational benefits.

•

Identify and categorize the negative psychological effects and challenges such as cognitive overload, anxiety, and over-reliance.

•

Explore the nuanced concept of student trust in an AI tutor and its relationship with learning outcomes.

•

Discuss the ethical and practical challenges of implementation, including the digital divide.

2. Background and Literature Review

2.1. The Psychological Impact of Virtual Reality in Education

While the literature firmly establishes the individual benefits of Virtual Reality (VR) in creating immersive learning contexts [18] and the efficiency of Artificial Intelligence (AI) in personalizing content [19], this paper identifies a critical research gap at their intersection. Specifically, the literature lacks a systematic synthesis of the direct psychological and emotional effects when an AI agent acts as a core learning assistant within a deeply immersive VR environment. Existing work tends to focus on cognitive outcomes (e.g., memory, performance) or technical usability, often overlooking nuanced elements like the development of trust, the potential for cognitive load, and the specific influence on student self-efficacy when interacting with an anthropomorphic AI guide.

Academic research in the last six years (2019 to 2025) has consistently shown that virtual reality holds significant promise as an educational tool with a variety of psychological benefits. The immersive nature of VR environments has been shown to boost student engagement, motivation, and enjoyment [4,20]. Studies indicate that this enhanced engagement can lead to improved learning outcomes and academic performance [21,22]. The capacity of VR to provide personalized, hands-on experiences can accelerate learning speed up to four times faster than traditional methods, helping students gain a deeper understanding of complex concepts by visualizing them in a three-dimensional space [10].

From an emotional perspective, VR has been explored as a tool for mental health management, with studies highlighting its effectiveness in treating anxiety and promoting emotional regulation [23]. Although these applications extend beyond education, they demonstrate VR’s capacity to create safe, controlled, and low-risk environments where users can practice social skills and cope with stressors without real-world consequences [23]. In educational contexts, this same emotional affordance becomes valuable as the supportive, gamified nature of VR can reduce performance anxiety and stress, fostering a sense of confidence in mastering technical skills through risk-free experimentation.

Cognitively, VR has been shown to reduce extraneous cognitive load by offering intuitive visualizations of complex information, which aids in a student’s understanding. This multisensory engagement and active learning approach can enhance memory retention by as much as 40% and sustain attention more effectively than traditional lectures [24].

Extended reality (XR) technologies offer multisensory, hands-on learning experiences that simulate real-world contexts [25]. Research also shows that educational VR can provide safe environments for trial and error [26], increase skill mastery and learner confidence [22], and enable access to otherwise inaccessible environments [27].

The integration of XR and AI technologies into entrepreneurship education must be intentional, responsible, and human-centered. Two foundational tools developed by the authors—the Extended Framework of Best Practices for AI-Enhanced XR Learning [4] and the AI Human Rights Impact Assessment for Educators [28]—guided the development of the AI-enhanced VR Microlearning Certificate in Entrepreneurship. Together, these tools provide a comprehensive framework for designing learning environments that are ethical, inclusive, and pedagogically sound.

However, the psychological effects of VR are not universally positive and depend heavily on implementation and user experience. A critical tension exists where the very immersion that creates profound educational benefits can also trigger negative emotional responses. For example, some studies note that a VR environment that closely replicates reality can intensify fear and anxiety in certain contexts, such as a fire safety simulation [29]. This creates a fundamental design challenge: how to maximize the immersive quality necessary for engagement without inadvertently exacerbating psychological distress. This suggests that the psychological impact of VR is not a simple linear function of immersion; rather, it is a nuanced dynamic where the technology must be carefully calibrated and paired with appropriate safeguards to ensure it remains a beneficial tool for all users. The report will argue that to prevent this psychological drawback, developers must incorporate safety features and emotional regulation tools into the VR learning environment.

2.2. The Role of Artificial Intelligence in Student Psychological Development

The integration of AI into education has been identified as a powerful tool for transforming and enriching traditional practices [5,9]. AI-powered tutors and platforms can provide personalized learning paths that address individual student strengths and weaknesses, fostering increased self-efficacy, learning autonomy, and decreased frustration [30]. The ability of AI to provide real-time, personalized feedback enhances motivation and helps students develop emotional intelligence by offering emotional support [5,9].

However, the psychological impact of AI is a double-edged sword. Research indicates that over reliance on AI can lead to a “spoon-feeding effect” where students become less engaged in independent thought and analysis [31]. This cognitive complacency poses a significant risk to the development of critical thinking and creativity, which are fundamental goals of education [31]. This risk is heightened when AI is used to provide ready-made answers, bypassing the cognitive effort required for problem-solving and critical research [31].

Another critical psychological consideration is the concept of trust. While students generally trust AI for its efficiency and convenience, their confidence in its outputs diminishes when tasks require nuanced human judgment [31]. This can lead to a “trust deficit,” where students perceive feedback from human teachers as more helpful and trustworthy than that generated by AI [31]. This perception of AI as a less authoritative or reliable source can hinder its full adoption and effectiveness. The psychological trade-off between the convenience offered by AI and the need for students to engage in deep, critical thinking is a central challenge that must be addressed in the design of any AI driven learning system. An effective system must actively encourage students to question and verify AI outputs, rather than accepting them uncritically, to ensure that the technology enhances rather than undermines intellectual growth.

2.3. The Psychological Impact of VFTs and the Gap in AI-Enhanced Immersion

Virtual field trips (VFTs) are digital experiences that replicate real-world excursions allowing students to explore and interact with locations remotely, often through VR technologies [12]. While VFTs may rely on VR, they represent a distinct educational application. VR more broadly encompasses diverse uses from gaming and entertainment to training and therapy but in the case of VFTs, the emphasis is specifically on engagement, motivation, and learning.

When implemented in immersive environments, VFTs share many of the same psychological advantages and drawbacks as VR. On the positive side, they have been shown to enhance engagement, sustain attention, and increase motivation, while also supporting deeper learning by allowing students the autonomy to make independent observations on their own beyond the constraints of traditional classroom lessons [32,33]. At the same time, the immersive quality of VFTs can intensify fear or discomfort [12] in some learners, reflecting the psychological risks already outlined in the broader VR literature.

The main benefit of VFTs appears to be the excitement they generate rather than direct content learning. Research shows that although students often become more engaged and motivated, the immersive experience can also distract them and limit how much material they retain. However, this excitement can still play a positive role, as it motivates learners to explore the subject further after the experience [12,34,35]. This motivational boost is particularly valuable in history education, where the subject is sometimes perceived as abstract or unengaging, focused mainly on memorization rather than active discovery [14]. By allowing learners to experience cultural heritage sites in a dynamic and interactive format, VFTs can transform history into a more meaningful and psychologically engaging subject. Beyond fostering short-term motivation, they help students appreciate the broader cultural and preservation value of historical landmarks, reinforcing a sense of connection to the past that traditional instruction alone may fail to achieve [32]. In addition, VR has proven to be a powerful medium for safeguarding intangible heritage, enabling the documentation and reconstruction of practices such as traditional crafts and performing arts, thereby helping to prevent the loss of valuable cultural knowledge for future generations [36].

In this respect, VFTs belong to a wider category of immersive learning experiences that utilize emerging technologies to connect students with cultural heritage and there are already applications that combine such experiences with AI. These developments illustrate a growing shift toward intelligent, interactive learning environments that go beyond passive observation fostering deeper engagement. Notable examples alongside the pilot study of this paper [16] include Cosmote Chronos [37], which integrates VR, AR, and AI to explore Athens and merges past and present by overlaying ancient reconstructions onto present-day monuments, either on-site or remotely, and ARcheos [38], which employs immersive interaction, AI and VPS to let users converse with virtual historical figures at landmark sites through natural language dialogue grounded in ancient Greek philosophy. These initiatives highlight the educational potential of immersive technologies in fostering engagement and contextual understanding. However, what is often missing is an analysis of their psychological effects when AI is embedded within the immersive experience as while there is extensive research on the psychological impact of VR based VFTs alone, far less is known about how the combination of VR with AI-driven interaction influences learner motivation, attention, and emotional engagement.

3. Analysis of Key Psychological Effects of the Combined AI-VR System

The fusion of AI and VR creates a synergistic learning environment with unique psychological implications that extend beyond the effects of each technology in isolation. The following sections analyze these combined effects on a student’s cognitive, emotional, and social development.

3.1. Cognitive and Attentional Effects

The combination of an embodied AI assistant in a VR environment has been shown to significantly improve student concentration and creativity [39]. By creating an immersive and all-inclusive sensory experience, this combined system helps students to move beyond passive learning and engage with content in a dynamic, flexible way [39]. The VR environment enhances image memory and encourages a deeper level of engagement with the material, which helps students form a more complex and higher level of thinking [20]. This approach reduces extraneous cognitive load by presenting abstract concepts intuitively and tangibly, making it easier for students to understand and master complex subjects [24].

However, the same distinctiveness that makes VR engaging can also shift attention from the learning task itself. When students become more focused on immersive features than on academic content, the depth of their cognitive processing may be reduced, limiting the long-term benefits of the technology [12]. This challenge becomes more complex when VR is combined with AI systems. While AI can provide real-time support and guidance, the potential for over-reliance on the AI to provide immediate answers may diminish the cognitive benefits of the VR environment’s hands-on learning [31].

This presents a unique psychological dilemma. The AI’s core value of convenience could undermine the VR environment’s core value of active, exploratory learning. For a combined AI-VR system to be truly effective, its design must actively counter this tendency toward cognitive complacency. The psychological impact is not merely additive but synergistic; the embodied nature of the AI, described in the foundational document as a spatial 3D avatar, introduces a new social-cognitive dynamic [16]. The user’s interaction with a virtual person, rather than a disembodied chatbot, could either encourage a more collaborative, less “lazy” interaction pattern or, conversely, create a false sense of trust that leads to greater over-reliance. The ideal system’s design must require students to engage critically with the information provided by the AI, for instance, by prompting them to verify sources or justify their reasoning, thereby preserving and enhancing the cognitive benefits of the VR experience.

3.2. Emotional and Motivational Responses

AI-VR systems can enhance intrinsic motivation through gamified elements and personalized feedback, resulting in a more enjoyable and engaging learning experience [40]. Students report increased confidence and enjoyment, with some feeling a stronger emotional connection to the content than in a traditional classroom [10,41]. Learning for personal interest, growth and curiosity has been shown to produce longer-lasting benefits than extrinsic motivation, which is usually tied to exam performance and short-term memorization [42]. VFTs, while not always leading to substantial new knowledge during the experience, often heighten engagement, encouraging students to continue studying the subject afterward and retain content for longer, once again highlighting the importance of intrinsic motivation [12,34,35].

The pilot study [16] also supports this point as although participants were initially hesitant, they quickly found the application easy to navigate and described it as a positive and engaging experience. This suggests that overcoming early hesitation is itself part of VR’s motivational effect, as positive first encounters build confidence and reduce psychological barriers to continued use. These findings align with broader research showing that VR’s supportive and gamified environment can reduce performance anxiety and stress [41].

At the same time, the pilot study [16] highlighted the influence of the teacher in shaping students’ willingness to participate. Learners only engaged with the headset after their teacher demonstrated its use, which is a crucial insight into the social psychology of technology adoption. This finding suggests that the psychological impact of AI-VR systems is not solely a function of technology and user; it is a tripartite relationship involving the user, the technology, and the social context of the classroom. This is consistent with research showing that students’ willingness to attend lectures is strongly influenced by the teaching style of the professor, underlining the central role of educators in fostering engagement [43].

An educator acts as a critical mediator, influencing student attitudes and overcoming initial psychological barriers such as hesitation and a perceived steep learning curve [9]. This observation reframes the concern about a “lack of human interaction” and indicates that the teacher’s active involvement is a crucial component of successful implementation. The adoption of these systems requires a psychological strategy that addresses the social dynamics of the classroom, ensuring that teachers are not just passive observers but are trained to act as active facilitators who can guide students through the learning process and model appropriate engagement with the technology.

3.3. Social and Behavioral Impacts

The literature presents conflicting views on the social impacts of AI-VR integration. On one hand, heavy reliance on these technologies risks diminishing face-to-face interactions among students and teachers, which are vital for developing interpersonal skills and collaborative learning [41]. The potential for social isolation is a notable concern, and even students themselves have expressed worries about the isolating effects of AI-VR based learning [12,30]. Virtual Field Trips (VFTs) provide a clear example of this challenge, as studies have found that although they provide immersive access to distant locations, students often miss the spontaneous conversations and real-time interactions with peers and teachers that characterize traditional field trips [12]. Similar concerns have been raised in relation to AI systems, where over-reliance on automated decisions and feedback has been linked to reduced interpersonal interaction, diminished social adaptability, and worries about privacy, control, and isolation [30].

However, the negative social effects are not inevitable but a function of system design. VR can be a powerful tool for fostering a sense of connection and collaboration among students in a shared virtual space [10]. The AI assistant, if designed as a collaborative tool rather than a solitary tutor, can mediate these interactions. The system can provide a safe, non-judgmental space for students who may be too anxious to ask questions in a traditional classroom setting, thereby promoting a more proactive and flexible communication style [18,23]. The psychological outcome of an AI-VR system—whether it leads to social connection or isolation—is entirely dependent on its design intent. An AI that serves as a shared resource for group problem-solving will have a different psychological effect than an AI designed for private, one-on-one tutoring. Therefore, developers must prioritize features that encourage interaction and shared learning to maximize the positive social and behavioral outcomes.

VFTs also highlight that the impact of immersive experiences extends beyond the classroom. Although they may not fully replicate the peer interaction of traditional trips, studies show that they can positively influence real-world behaviors. Reported outcomes include healthier food choices, preparing emergency kits (e.g., earthquake bags), making shopping lists to reduce waste, and showing empathy by feeding animals appropriately [35]. These findings suggest that immersive technologies can shape not only how students think and feel but also how they act, reinforcing pro-social and responsible behaviors through meaningful virtual exposure.

Therefore, when designed with these outcomes in mind, AI-VR systems can serve as powerful tools not only for cognitive learning and emotional engagement but also for fostering both positive social interaction and real-world behavioral change.

4. Design and Implementation of the Pilot Study

The primary objective of the pilot study [16] was to create an immersive, interactive AI (virtual) assistant application for educational use, which was developed to run on an Oculus VR headset. The end users (i.e., students) could communicate with an AI-powered assistant using natural speech. The project evolved from an initial concept to a fully functional application, with a detailed focus on overcoming technical challenges and integrating a suite of advanced technologies. The full development process is explained here to provide transparency in design decisions, demonstrate how limitations were addressed, and highlight the technical foundations upon which the subsequent analysis of educational and psychological implications is built.

4.1. Initial Development Concept (ThirdEye X2)

Before arriving at the final implementation, the project explored the use of the ThirdEye X2 Mixed Reality headset, which offered advanced features such as Simultaneous Localization and Mapping (SLAM) for anchoring virtual objects in the real world. This subsection is included to document the initial design pathway, as understanding both the opportunities and the limitations of early development choices provides valuable context for the eventual transition to the Meta Quest 2.

Despite its potential, the ThirdEye X2 was ultimately abandoned due to significant limitations. The headset lacked support for Google Play, restricting its use to pre-installed and custom applications. Furthermore, there was a notable lack of study materials and documentation for developing applications with Unity or Unreal Engine, which severely constrained the project’s scope. These technical barriers made sustained development impractical and directly informed the strategic decision to pivot toward a more widely supported platform: the Meta Quest 2 headset.

4.2. Transition to Meta Quest 2 and Final Implementation

The limitations of the ThirdEye X2 headset necessitated a change of platform. The Meta Quest 2 was chosen as the alternative because it provided stronger developer support, easier access to learning resources, and a more reliable environment for building educational applications. This transition marked an important turning point in the project, as it allowed the initial concept to be reshaped into a more practical and sustainable solution.

The move to Quest 2 also expanded what the project could realistically achieve. With better support and flexibility, it was possible to refine the design through several rounds of revision, gradually transforming the idea of a simple AI assistant into an immersive application suitable for testing in an educational context.

Following two rounds of revision, the final outcome was an AI-powered virtual tour guide for the city of Athens, which serves as the case examined in this paper. The application allowed users to explore Athens in an immersive way, moving freely through a virtual environment while interacting with the assistant through natural dialogue. To maximize usability, the application was integrated directly into the Quest 2 interface, allowing the assistant to be activated through a dedicated menu option without requiring additional setup or external software (Figure 1).

4.3. Implementation Features

The AI assistant was designed to provide responses in both voice and text, generated either from a predefined knowledge base enriched and validated by domain experts or dynamically through ChatGPT ver1 for queries outside the programmed scope. This dual-response system ensured that the assistant could deliver reliable domain-specific information while also supporting open-ended interaction. To enhance usability and immersion, several additional features were included:

•

Exploration and navigation: Users could move freely through an immersive 3D version of Athens using joystick navigation, similar to Google Maps, rather than being restricted to a single position.

•

Color-coded feedback: Pre-programmed responses appeared in green text, while dynamically generated ChatGPT responses were shown in red text, giving users clear transparency over the source of information (Figure 2).

The implementation features outlined here were supported by a range of underlying technologies, which are described in the following section.

4.4. Core Technologies Involved

The implementation of the AI-powered virtual tour guide relied on a combination of hardware, software, and cloud-based services that enabled both the immersive VR environment and the AI-driven interaction. The main technologies were:

•

Meta Quest 2: The VR headset chosen as the primary platform due to its strong developer ecosystem and compatibility with Unity.

•

Unity: The game engine used to build and integrate the application environment.

•

OpenAI API: Enabled communication with the gpt-3.5-turbo model to generate dynamic, AI-driven responses.

•

Oculus Voice SDK: Provided Speech-to-Text (STT) and Text-to-Speech (TTS) functionality for natural voice interaction.

•

Meta XR All-in-One SDK: Offered preconfigured “Building Blocks” that streamlined setup of core VR functions, such as camera configuration, hand tracking, and controller input.

•

Wit.ai: Served as the natural language processing layer, trained to identify user intents and extract relevant entities from voice commands.

•

Google Maps Platform with Cesium for Unity: Delivered high-fidelity 3D mapping, allowing accurate visualization of Athens landmarks within the VR environment.

The setup of the VR environment was greatly simplified by using the Meta XR All-in-One SDK’s Building Blocks. These tools, which included the Camera Rig, Hand Tracking, and Controller Tracking, automated the configuration of essential VR functionalities, saving considerable development time. For instance, the Camera Rig automatically configured the in-game camera for the VR environment, while the Hand Tracking and Controller Tracking blocks enabled the application to recognize and utilize user hand gestures and controller movements. The Controller Buttons Mapper component was also used to program specific buttons on the VR controllers to trigger functions within the application.

A central element of the revised implementation was the use of Wit.ai for natural language processing (NLP). A model was created and trained on the Wit.ai platform to understand the user’s intentions, known as Intents, and to extract key pieces of information, called Entities. For example, the intent info_attraction was created to handle queries for information about tourist sites, with the entity attraction being trained to recognize specific names like “Parthenon” or “Syntagma Square”.

A separate intent, chatgpt, was designed to recognize general questions that required a dynamic response from the AI. The Entities for the info_attraction were set to use a “Free Text & Keywords” strategy, allowing the model to understand both pre-defined and new inputs. For the ChatGPT intent, the entity questiongpt used a “Free Text” strategy to capture the entirety of the user’s query and pass it directly to ChatGPT.

The App Voice Experience component, part of the Oculus Voice SDK, was integrated to serve as the bridge between Unity and the trained Wit.ai model. This component handled the entire voice interaction process, from recognizing the start and end of a voice command to displaying the transcribed text on the screen. It was configured to visually provide feedback to the user, showing “Listen” or “Listening now” based on its current state. The voice experience was activated by a user pressing a button on the VR controller, as configured through the Controller Buttons Mapper.

4.5. Advanced Features Involved

The final application was built on a robust architecture that integrated several advanced features to create a seamless and immersive experience. The implementation of Text-to-Speech (TTS) was handled by the TTS Service and TTS Speaker components within the Oculus Voice SDK. This allowed the AI assistant to vocalize its responses, addressing a key technical limitation of the initial version. To enhance the realism of the interaction, the Oculus LipSync tool was utilized. By assigning the avatar’s mouth blendshapes to the OVR Lip Sync Context Morph Target script, the avatar’s mouth movements were synchronized with the audio output of the TTS Speaker, making the avatar appear to be speaking naturally.

The Response Handler was a critical script that acted as a central hub for interpreting the results from Wit.ai. When the user made a voice command, the App Voice Experience sent the transcription to Wit.ai, which returned a recognized intent and any associated entities.

The Response Handler then used this information to trigger the appropriate action. For predefined queries, such as “Tell me about Parthenon,” the handler would extract the attraction entity and run a local script that provided the information and displayed the relevant 3D location. For a general query like “Where can I eat in Athens?”, the handler would recognize the chatgpt intent and pass the query to the ChatGPTManager script.

The ChatGPTManager script managed the dynamic communication with the OpenAI API. It was configured with a specific personality and scene (e.g., “tourist guide” in “Athens”) to ensure that the AI’s responses were contextually relevant (Figure 3).

The script sent the user’s question, along with this contextual information, to the gpt-3.5-turbo model. The AI’s response was then received, processed, and vocalized through the TTS system, allowing for open-ended conversations beyond the application’s pre-programmed content.

Finally, to create an engaging visual environment, the application integrated the Google Maps Platform using the Cesium for Unity plugin. This enabled the seamless streaming of high-fidelity 3D map data from Google Earth directly into the VR environment. For example, when a user asked about a specific landmark, the application would automatically display a 3D model of that location, enhance the sense of immersion and provide a powerful visual learning experience.

In conclusion, the successful implementation of this educational AI agent on the Oculus VR headset was a result of carefully selecting and integrating a cohesive set of development tools and APIs. By transitioning from incompatible technologies to the robust Oculus Voice SDK, Meta XR Building Blocks, and Wit.ai, the project was able to overcome its initial challenges. The final application, a virtual tour guide, serves as a proof of concept for a dynamic, interactive, and immersive educational tool that leverages the combined power of AI and VR to provide a richer and more responsive learning experience for students.

5. Case Study: A Proposed Study on AI-VR Learning for High School Students

The pilot study [16] included a small-scale case study (N = 11) with a mixed group of participants, consisting of teachers, students and one bookstore owner. While this provided useful feedback on usability and general acceptance, it limited the extent to which student-specific psychological effects could be examined. The results nonetheless highlighted important factors such as the influence of the teacher on technology adoption, the role of trust in AI, and the challenges posed by language barriers.

Building on these findings, a case study was designed exclusively on high school students, offering an age-appropriate context to investigate more directly how AI-VR learning environments affect learner motivation, attention, and engagement. This sharper focus is intended to address the limitations of the pilot study by providing deeper insight into the psychological dimension of AI-VR learning.

5.1. Methodology Rationale and Demographics

The proposed case study involved a group of high school students (N = 52), consisting of 24 girls (n = 24) and 28 boys (n = 28) from the 3rd high school grade. Participants will be asked to use Oculus VR headsets with the AI assistant in groups of three using Oculus headsets over a one-week intervention period. This sample size of 52 was chosen based on an a priori power analysis using G*Power 3.1 [44], which indicated that 50 participants are sufficient to detect a moderate effect size (f = 0.25) in an ANOVA with a power of 0.80 and an alpha level of 0.05, suitable for a pilot study in an educational setting.

Participants are separated into groups of three for exploration tasks. This grouping is deliberate, as it mirrors authentic collaborative learning scenarios in a classroom, allowing the study to assess how the AI assistant mediates group trust and collective self-efficacy, rather than just individual-AI interaction.

The students will be introduced to the VR learning application featuring an AI assistant plugin on Oculus headsets. The application, a virtual tour guide, will include two modes of interaction: a preprogrammed mode with a limited set of questions and a second mode that leverages the AI plugin to answer a wider range of queries via the AI embodied avatar.

Participants will receive instructions from their teacher and will then engage in a one-week hands-on trial, exploring historical areas and retrieving scientific data on certain historical places such as the Parthenon and Syntagma Square in Athens. At the end of the trial, all participants will complete an anonymous questionnaire to provide feedback on their experience, focusing on usability, trust in the AI, and overall psychological impact.

As the participants are minors, parental/guardian consent will be obtained in advance, in accordance with ethical research guidelines for studies involving children (See proposed study procedure in Figure 4).

Data Analysis and Instrumentation

Quantitative Analysis: comparisons of trust and cognitive load were tested using a repeated-measures Analysis of Variance (ANOVA). Secondary analyses, including correlation and multiple regression, were performed to explore the relationship between user-reported engagement and final task performance. All quantitative data were analyzed using SPSS (version 28).

Qualitative Analysis: Open-ended questionnaire responses and structured interview transcripts from participants were subjected to Thematic Analysis [45] to capture emergent psychological themes and contextualize quantitative findings.

The entire protocol, including informed consent forms and data handling procedures, was explained in detail, and data were fully anonymized, with no personal data included and with no backtrack trace available in order to locate the initial responder.

For these reasons, no approval from the Ethics Committee was required. Full ethical clearance has to be obtained before any participant recruitment. Participant confidentiality was ensured by assigning anonymous random ID codes, and all raw data were stored on password-protected, encrypted university server only for a period of one month.

5.2. Expected Key Psychological Findings

5.2.1. Role of the Educator and Social Influence

Based on the pilot study and prior literature [16], it is anticipated that high school students, despite being digital natives, will show an initial reluctance to engage with the VR headset until their teacher demonstrates its use. The teacher’s active role as a facilitator—modeling the technology and encouraging participation—is therefore expected to be crucial in overcoming this psychological barrier and shaping students’ willingness to engage (Figure 5). In this way, the educator is anticipated to function not only as an instructor but also as a social mediator, helping to reduce hesitation and foster adoption of the new technology.

5.2.2. User Perception and Trust in AI

The pilot research [16] and literature [19,31] suggest that adolescents may show a nuanced understanding of trust in AI. While the AI assistant plugin is likely to be viewed as a useful and convenient tool for its quick access to information, students’ confidence in its outputs may probably diminish when the questions require more nuanced human judgment.

The students are likely to express concerns about the accuracy of AI-generated responses, echoing the “trust deficit” reported in prior research, where teacher feedback is perceived as more trustworthy and helpful (Figure 6). This anticipated skepticism highlights the risk of over-reliance: the convenience of the AI plugin may encourage cognitive complacency and a “spoon-feeding effect,” potentially reducing students’ critical thinking and creativity [31].

5.2.3. Functionality and Cognitive Load

Previous findings suggest that interaction with AI assistants through voice commands can pose both psychological and practical challenges. Students with lower English proficiency may struggle with frustration and difficulty in using such systems [16]. Accordingly, the proposed study anticipates the need for the AI plugin to support multiple languages or provide alternative input methods to reduce cognitive load and prevent feelings of frustration, thereby making the system more inclusive and psychologically accessible for a diverse student body (Figure 7).

5.2.4. Broader Psychological Impacts

The fusion of an integrated AI agent in VR (Oculus Headsets), as planned in the proposed case study, is expected to create a powerful learning tool, though its effects are likely to be complex. While the system may enhance positive feelings such as engagement and motivation, it also carries potential risks, including over-reliance and social isolation (Figure 8). These effects depend on how the system is designed and implemented:

• Positive Impacts: The immersive nature of the VR environment combined with the AI assistant’s personalized feedback is expected to increase student engagement and motivation [46]. The ability to learn at their own pace and explore virtual concepts hands-on can foster a sense of self-efficacy and learning autonomy [30,33].
• Negative Impacts: If the system design does not adequately mitigate potential risks, it could lead to negative outcomes. A heavy reliance on the AI plugin for answers could diminish critical thinking and problem-solving skills [30]. Furthermore, a lack of collaborative features could increase the risk of social isolation, as it may reduce face-to-face interactions that are crucial for developing interpersonal skills in a classroom setting [41].

5.3. Case Study Findings

Drawing on the pilot study and relevant literature, the following findings were formulated from the implemented case study:

• F1: Role of the Educator and Social Influence
• Students were hesitant to use the VR headset until the teacher demonstrated or encouraged its use, reflecting the teacher’s role as a key social mediator in technology adoption.
• F2: User Perception and Trust in AI
• Students perceived the AI assistant as a useful and convenient tool for quick access to information but will show lower trust in its accuracy compared to human teachers, especially for tasks requiring nuanced judgment.
• F3: Functionality and Cognitive Load
• Students with lower English proficiency experienced higher cognitive load and frustration when using voice commands, underscoring the need for multilingual support or alternative input methods to ensure inclusivity.

In summary, these findings reflect both the opportunities and risks of AI-VR integration in education [46]. They provide a structured basis for the proposed case study, guiding the investigation of how psychological dynamics such as social influence, trust, cognitive load, and broader impacts manifest in high school students’ adoption of emerging technologies.

6. Ethical and Practical Implications

The integration of these powerful technologies in education is not without its ethical and practical challenges, which can have significant psychological consequences for students.

6.1. The Ethical Framework of AI-VR in Education

The integration of AI and VR in education introduces significant ethical concerns that have a direct psychological impact on students. A primary issue is data privacy as AI systems, by their very nature, require the collection and analysis of vast amounts of student data to personalize learning paths [40]. This raises serious concerns about the potential misuse of sensitive information and the psychological vulnerability of adolescents who may not fully understand the implications of their digital footprint [30]. The lack of transparency in data usage can erode a student’s sense of autonomy and security.

Another critical ethical concern is algorithmic bias. AI algorithms can perpetuate existing societal biases—via the pre-determined answers in this case—which could influence curriculum recommendations, grading, and even a student’s perceived opportunities [40].

From a psychological standpoint, this could have a profound effect, potentially diminish a student’s self-esteem and create a self-fulfilling prophecy of underperformance based on biased data rather than merit. The fear of being unfairly judged or misunderstood by an AI system can lead to stress and anxiety, countering the positive emotional effects that the VR environment is designed to provide [30]. A robust ethical framework must be established to ensure fairness, transparency, and a system of checks and balances with human oversight to mitigate these psychological risks.

6.2. Overcoming Implementation Obstacles

The high cost associated with VR hardware, software, and ongoing technical support presents a major practical and ethical hurdle. This financial barrier creates a significant “digital divide” exacerbating educational inequalities and leaving many students without access to these advanced learning tools [5]. The psychological effect of this exclusion cannot be ignored; students in under-resourced schools may feel a sense of disadvantage and frustration, which could negatively impact their motivation and sense of self-efficacy in a technologically advanced world.

Furthermore, both students and educators face a steep learning curve when adapting to new technologies [5]. The pilot study from the foundational document serves as a prime example of this challenge [16]. Students’ initial hesitation to use the VR headset underscores the psychological friction involved in adopting new technology. The fact that the teacher’s intervention was necessary to encourage student participation highlights the crucial need for professional development and training for educators [5]. This finding is supported by survey data [47] from 189 higher education teachers, where most participants acknowledged VR’s potential to improve learning but identified readiness, confidence, and training as significant barriers. While 82% agreed that VR could improve student interest and 66% believed it could enhance teaching effectiveness, only 38% indicated they were likely to adopt it in their classroom, with many stressing the need for sufficient training and established content before implementation.

Without a clear strategy for onboarding teachers and students, the technology’s full potential will not be realized, leading to frustration and underutilization [47]. The report argues that institutions must invest not only in the technology itself but also in the human capital required to integrate it effectively and ethically into the learning environment.

7. Conclusion and Future Directions

7.1. Summary of Findings

The integration of AI learning assistants into immersive VR environments presents a multifaceted psychological profile for adolescents aged 14 and older. The analysis of contemporary academic literature reveals a dual nature: significant potential for positive psychological effects, balanced by considerable risks and challenges. On the positive side, VR’s immersive nature enhances student engagement, motivation, and cognitive skills like memory and attention. The AI component provides a foundation for personalized learning, fostering a sense of autonomy and self-efficacy [30]. The embodied nature of the AI avatar, as seen in the foundational document, introduces a dynamic social presence that can further enrich the learning experience [16].

Conversely, the same technologies carry risks that must be carefully managed. Over-reliance on the AI can lead to a decline in critical thinking and creativity [31]. The immersive VR environment, while generally beneficial, can also induce anxiety or discomfort in some users [12,29]. A significant psychological barrier to adoption is the trust deficit that students often have in AI-generated information, perceiving it as less reliable than human feedback [19]. The pilot study further contextualizes these findings, revealing the critical role of the teacher as a psychological mediator in overcoming student hesitation and a key factor in the social acceptance of the technology [16]. The study also highlighted the challenge of language barriers, with users with lower English proficiency experiencing frustration with the voice command system, suggesting that the system must be designed for a linguistically diverse audience [16].

In addition, the use of virtual field trips (VFTs) within history education demonstrates how immersive technologies can transform abstract or distant subjects into vivid, experiential learning opportunities [14,32]. By situating students within reconstructed cultural heritage sites, VFTs make history more engaging and psychologically meaningful [32]. When combined with AI-driven interaction, these experiences offer the potential for deeper contextual understanding but also highlight the need for careful design to balance excitement with sustained learning outcomes [10,32].

7.2. Study Limitations and Mitigation Strategies

The most significant limitation of the intervention was the utilization of 52 high school students, while statistically powered for primary outcomes, will remain susceptible to the small sample size issue common in pilot VR studies. Furthermore, the novelty effect of using Oculus headsets and an AI assistant for the first time is a significant confounder. To mitigate this, our one-week content exploration period was designed to allow participants to move past the initial novelty and settle into a more routine interaction pattern, to measure the novelty factor explicitly in the post-use questionnaires.

7.3. Recommendations for Design and Implementation

Based on the analysis, the following recommendations are proposed to developers and educators for the design and implementation of AI-VR systems in education:

•

System Design: Developers should move beyond creating “AI-as-a-tutor” systems that simply provide answers. Instead, the focus should be on building “AI-in-the-loop” systems that require active student participation and human oversight. The system should encourage students to engage critically with the information, for example, by prompting them to justify their reasoning or cross-reference sources. This design approach can mitigate the psychological risk of over reliance and promote the development of critical thinking skills.

•

Educational Integration: Educators must be trained to act as “facilitators of technology” rather than passive observers. As evidenced by the pilot study, their role is crucial in overcoming the initial psychological barrier to adoption. Educators should guide students on how to use the AI-VR system critically, to value human interaction and collaborative learning, and to understand the limitations of this technology.

•

Ethical Safeguards: The psychological well-being of students is paramount. Robust ethical frameworks must be put in place to ensure data privacy, transparency in data usage, and mechanisms to address algorithmic bias. This will foster a sense of trust and security, which is a key psychological prerequisite for effective technology adoption.

•

Accessibility: To address the digital divide and its psychological consequences, institutions and policymakers must prioritize providing equitable access to hardware and training. Making these technologies widely available, rather than a luxury, will ensure that all students have the opportunity to benefit from an enhanced learning experience.

•

Multilingual Support: The pilot study revealed a critical need for language support beyond English. While full voice command functionality in every language may not be feasible, systems should provide multilingual subtitles for AI-generated responses and allow for voice command systems in native language. This would reduce cognitive load and frustration for non-native English speakers, making the system more inclusive and psychologically accessible.

These recommendations emphasize that the success of AI-VR systems in education depends not only on technical innovation but also on thoughtful design, teacher involvement, ethical safeguards, and inclusivity. Together, these elements can ensure that such technologies enhance learning while minimizing psychological risks.

7.4. Future Research

The field of AI-VR integration in education is nascent, and the psychological effects are not yet fully understood. This report identifies several key areas for future research such as:

●

Issue No 1: Longitudinal Studies

●

The existing literature lacks long-term studies on the psychological outcomes of AI-VR interaction. Future research should investigate the long-term effects on student critical thinking, social skill development, and potential over-reliance on technology [31].

●

Issue No 2: Embodiment and Trust

●

The psychological impact of an embodied AI avatar is a key variable that warrants further investigation. More research is needed to understand how a student’s trust, learning, and emotional connection are specifically influenced by the physical presence of a virtual tutor [19,41].

●

Issue No 3: Cross-Cultural and Linguistic Nuances

●

The pilot study’s finding that students with a low proficiency in English struggled with voice commands highlights the need for research into the psychological effects of such systems across diverse linguistic and cultural backgrounds. This would inform the development of more inclusive and accessible educational tools that account for global user needs [16].

By pursuing these avenues of research, the educational community can better understand the full psychological implications of these powerful technologies and harness their potential to create more effective, engaging, and equitable learning environments for the next generation of students.