Integrating Large Language Models into Accessible and Inclusive Education: Access Democratization and Individualized Learning Enhancement Supported by Generative Artificial Intelligence

Abstract

This study explores the integration of large language models (LLMs) into educational environments, emphasizing enhanced accessibility, inclusivity, and individualized learning experiences. The study evaluates trends in the transformative potential of artificial intelligence (AI) technologies in their capacity to significantly mitigate traditional barriers related to language diversity, learning disabilities, cultural differences, and socioeconomic inequalities. The result of the analysis highlights how LLMs personalize instructional content and dynamically respond to each learner’s educational and emotional needs. The work also advocates for an instructor-guided deployment of LLMs as pedagogical catalysts rather than replacements, emphasizing educators’ role in ethical oversight, cultural sensitivity, and emotional support within AI-enhanced classrooms. Finally, while recognizing concerns regarding data privacy, potential biases, and ethical implications, the study argues that the proactive and responsible integration of LLMs by educators is necessary for democratizing access to education and to foster inclusive learning practices, thereby advancing the effectiveness and equity of contemporary educational frameworks.

1. Introduction

The rapid evolution of large language models [1] (LLMs), exemplified by accessible systems such as OpenAI’s ChatGPT [2], represents a transformative milestone in educational methodologies. Based on advanced natural language processing (NLP) algorithms, LLMs exhibit remarkable capabilities in generating coherent, contextually sensitive interactions, understanding complex inputs, and facilitating meaningful educational engagements [3]. This technological breakthrough offers promising solutions to long-standing educational challenges, including those addressing accessibility and inclusivity [4], enabling new opportunities that significantly benefit learners with diverse educational requirements [5].

Educational practices have relied predominantly on rigid, one-size-fits-all methodologies grounded in memorization and uniform instruction [6]. Such traditional approaches often overlook the diverse spectrum of student abilities, socioeconomic backgrounds, and different learning styles, inadvertently marginalizing or excluding a significant number of learners. Although the integration of digital technologies in the late 20th century marked an important step forward by introducing digital content and self-directed learning, these advances still fell short of fully personalized educational experiences due to their limited adaptability [7]. It was not until the emergence and rapid proliferation of artificial intelligence (AI)-driven environments, particularly sophisticated tutoring systems [8] and, more recently, LLMs, that true adaptability and individualized learning has started to become a tangible reality [3].

The introduction of AI, especially through LLMs, has catalyzed a fundamental shift from standardized instructional methods to highly personalized educational experiences. As described in [3], these intelligent models provide tailored learning paths, offering personalized feedback and dynamically adapting to each student’s unique pace, cognitive style, and educational preferences. Thus, learners gain access to continuous and responsive support that extends well beyond traditional classroom hours, bridging the gap often created by restrictive instructional periods. This flexibility and sustained accessibility prove especially critical for students requiring extended time, repeated exposure to educational material, or alternative learning modalities, further amplifying the inclusivity and reach of educational resources [9,10,11].

Recent developments in the LLM domain have further deepened their educational impact, notably with the emergence of localized on-device models such as Meta’s Llama, Microsoft’s Phi, or Mistral models. These localized AI solutions enhance user privacy, substantially reduce latency, and enable offline operation, making them exceptionally valuable resources in under-resourced regions or in contexts with limited internet connectivity. By promoting culturally relevant, locally adapted learning experiences, such innovations hold profound implications for educational equity, offering underserved communities unprecedented opportunities for inclusive and personalized education.

Yet, despite these clear advantages, considerable apprehension remains prevalent within the educational community [12]. Concerns persist about potential threats to academic integrity, the misuse of AI tools, and possible displacement of educators [13]. These anxieties underscore the critical need for an approach that reframes the role of AI not as a replacement but rather as a collaborative partner. Recognizing and leveraging AI’s strengths in creating personalized and adaptive learning environments can inspire educators to redefine their roles, shifting the focus towards more strategic, creative, and student-centered pedagogical practices [14].

Emerging evidence demonstrates the efficacy of AI-driven personalized educational systems, suggesting measurable improvements in learning outcomes compared to traditional classroom methodologies [3]. Intelligent tutoring systems have already shown significant improvements in student performance by continuously adapting instructional strategies to the learner’s progress [15], interests, and capabilities. For instance, such personalized tutoring environments, powered by AI, have already demonstrated effectiveness surpassing conventional teaching methods, supporting the argument that adopting AI tools is not merely beneficial but increasingly essential for future educational practices [16].

The trajectory of AI integration within educational contexts points decisively towards an increasingly central role for technologies such as intelligent tutoring systems [15], automated grading mechanisms [17], and dynamic content generation tools [18]. Collectively, these advanced personalized AI learning systems represent the front line of a broader educational transformation, emphasizing an approach centered on the individual learner. While the proliferation of AI tools continues to reshape the educational landscape, it simultaneously generates new ethical, practical, and pedagogical challenges that must be proactively addressed [19].

In this study, we analyze the progressive incorporation of LLMs and AI technologies into education and how it signals a profound transformation into pedagogical practices, with a focus on accessibility and inclusivity. Rather than resisting or fearing such advancements, educators must proactively integrate technological advancements to harness AI’s potential responsibly. By thoughtfully addressing ethical considerations, promoting equitable access, and embracing AI as an educational collaborator, lectures can achieve a more inclusive, adaptable, and effective educational paradigm that is tuned to the needs of learners [20]. This work also conducts an exploratory investigation into the methodology of the integration of LLMs within educational settings. Its primary aim is to synthesize current perspectives and highlight how LLMs can function as pedagogical catalysts tools that, under the guidance of human educators, dynamically adapt to individual learners’ cognitive, linguistic, emotional, and cultural needs. The main contribution of the work is to offer a structured conceptual framework informed by recent developments in AI and educational theory. Specifically, it emphasizes the potential of LLMs to enhance accessibility, inclusivity, and personalized learning, with practical attention to areas such as English as a second language, as well as broader educational contexts where learners face traditional barriers including language diversity, learning disabilities, cultural mismatches, and socioeconomic challenges. We summarize in Figure 1 the key points of the integration of LLMs into inclusive education, depicted as a conceptual schema of key educational barriers.

This study is structured as follows: Section 1 introduces the rationale and thematic scope of the investigation. Section 2 addresses the theoretical underpinnings of accessibility and inclusion, while Section 3 explores LLMs’ technical and pedagogical applications in supporting equitable learning. Section 4 extends this analysis by focusing on individualized learning strategies, and Section 5 and Section 6 provide a critical discussion and conclusions, respectively. As a result, the present study offers educators and researchers an integrative overview of how LLMs might be ethically and effectively integrated to support inclusive education. The manuscript contributes to an emerging conversation about the future of AI enhanced pedagogy, grounded in a human-centered, educator-led approach.

2. Accessibility and Inclusion Within Education and the Role of Generative AI

Accessibility within education focuses on creating instructional materials and learning environments that foster full participation for students, irrespective of disabilities or diverse abilities [4]. Central to achieving comprehensive educational accessibility is the broader concept of inclusion, which extends beyond merely accommodating students’ varying needs to actively integrating all learners into mainstream educational contexts [9]. In doing so, inclusive practices aim not only to ensure equitable academic opportunities but also to enrich social and emotional experiences, addressing effective learner development [21,22,23,24].

An influential approach underpinning the successful achievement of accessibility and inclusion is the universal design for learning (UDL) paradigm [25]. UDL prioritizes flexibility in educational practices through diverse modes of content representation, varied means of expression, and engaging adaptable interaction. This framework actively addresses the broad spectrum of cognitive, sensory, emotional, and physical needs of students. In alignment with the UDL principles, LLMs provide robust support mechanisms by enabling personalized educational experiences. Through multimodal content presentation [26], interactive dialogues [27], and adaptable responses [28], LLMs serve as dynamic tools to meet diverse student needs, effectively improving both educational equity and engagement [9].

Beyond the structured curricular elements typically addressed by traditional pedagogies, educational systems have historically placed a limited emphasis on students’ emotional, social, and cultural development [3]. However, effective education must integrate support for these dimensions, as they are deeply interconnected with academic performance and overall student well-being [29]. In this context, LLMs have emerged as promising technological interventions capable of addressing these often neglected domains [30]. With their apparent capacity for empathetic interaction [31], LLMs seem to engage students in supportive, non judgmental conversations, facilitating emotional expression and mitigating feelings of stress, anxiety, or isolation. Such emotional support, provided in real time and personalized to the learner’s emotional state, could significantly enhance students’ resilience, promoting an emotionally conducive learning atmosphere [32].

In parallel with emotional well-being, the integration of LLMs into educational environments can substantially enhance students’ social development, as shown in [33]. Using collaborative learning scenarios and interactive virtual environments, LLMs promote enhanced peer interactions, teamwork, and effective communication. Such digital platforms and interactive tools have been shown to facilitate improvements in peer engagement, reinforcing the communicative and collaborative skills essential for students’ long-term personal and professional growth [3,14]. Although promising, more research is needed to thoroughly assess the durability and real-world applicability of these social skills provided by LLM interactions [34].

In addition, educational accessibility also includes cross-cultural competence and understanding [35], areas in which traditional educational paradigms may fall short due to limited cultural sensitivity or linguistic diversity. LLMs, with their inherent multilingual capabilities [2] and sophisticated cultural adaptability [36], can facilitate deeper multicultural understanding and inclusivity. By simulating cross-cultural dialogues [35], providing culturally sensitive guidance [37], and offering personalized multilingual support [27], these advanced technologies enable students to navigate diverse global contexts effectively, fostering empathy and respect for cultural diversity. Nevertheless, the extent of LLMs’ cultural sensitivity and the depth of their effectiveness in genuinely enhancing intercultural competence is in need of further empirical investigation [38].

The personalized learning approaches inherent to LLM technology also hold significant potential in the realm of emotional adaptability [28]. These models are capable of dynamically adjusting their educational delivery based on a student’s academic performance and emotional signals, such as frustration or disengagement. Through proactive intervention, such as suggesting breaks, alternative activities, or tailored motivational feedback, LLMs can help foster students’ emotional self-awareness, adaptability, and overall resilience [3]. Personalized educational experiences enabled by these technologies can effectively tackle growth mindsets, facilitating persistent engagement and confidence in students’ own abilities [7]. Yet, further research is necessary to assert the long-term impacts of these emotionally adaptive educational strategies, as many studies fall short in duration due to the rapid adoption of the technology [3].

Furthermore, the foundation for the advancement of inclusive education is based on the social model of disability [39], a framework that emphasizes that disability predominantly arises from societal barriers and exclusionary practices rather than individual impairments alone. In educational settings, societal and institutional barriers, such as rigid curricula, a lack of emotional responsiveness, and limited adaptive technologies, contribute significantly to the exclusion of learners with diverse abilities [3]. LLMs actively challenge and mitigate these structural barriers by offering personalized, empathetic, and adaptive learning environments [3]. Through their tailored support capabilities, LLMs help dismantle exclusionary educational practices, providing more inclusive experiences that acknowledge and adapt to the unique abilities and circumstances of each learner.

3. Enhancing Accessibility Through LLMs

LLMs significantly improve educational accessibility through their versatile functionalities, especially by generating multimodal educational content tailored to diverse learning needs [26]. This includes simplified texts, comprehensive summaries, and real-time language translations across various languages, effectively catering to multilingual learners and students with specific learning disabilities [40]. Such inclusive capabilities of LLMs notably contribute to bridging linguistic and cultural gaps in educational settings, as emphasized in [36]. A similar study [41] underlines the transformative impact of real-time language translation features, highlighting their potential to create inclusive and equitable learning environments that accommodate the diverse linguistic backgrounds of students.

Beyond language-specific barriers, LLMs offer meaningful support to students experiencing communication challenges. By accurately interpreting limited or fragmented student inputs, these models generate comprehensible outputs that foster active participation within educational dialogues. This capability is particularly beneficial for students who face difficulties in verbal or written expression, enabling them to engage confidently and meaningfully in classroom interactions. Additionally, AI-driven assistive technologies, guided by LLM capabilities, play an essential role in supporting learners with disabilities. Ref. [40] advocates for the integration of such inclusive technological models, underscoring their effectiveness in personalizing educational experiences and addressing the unique needs of each student.

LLMs also critically contribute to enhancing distance education, effectively mitigating geographical barriers that traditionally limit access to quality learning [9]. Ref. [42] reveals that generative AI tools like ChatGPT significantly improve the delivery of educational content remotely, enabling students, irrespective of their location, to access robust educational resources and participate actively in learning experiences. The continuous availability and instantaneous responsiveness of these AI tools ensure that educational assistance remains accessible at any time, particularly supporting students requiring additional time or repeated exposure to instructional materials [4]. Consequently, learners benefit immensely from the autonomy afforded by such systems, fostering independent and reflective learning practices facilitated through personalized and immediate feedback mechanisms provided by LLMs.

However, in addition to these substantial benefits, the implementation of LLMs into educational settings raises critical considerations, particularly related to data privacy, bias, and ethical implementation. Ref. [43] warns educators and institutions about potential data privacy risks inherent to employing generative AI, stressing the imperative for robust privacy protections and adherence to established accessibility standards. Ref. [44] further highlights the risk that unchecked biases in generative AI could inadvertently amplify existing educational inequalities rather than reduce them, calling for a thoughtful, ethical application of these technologies to ensure fairness and equity in educational outcomes.

Addressing these concerns needs active and informed engagement from educators, who play a pivotal role in integrating AI into educational contexts ethically and effectively. According to [45], educators must adopt AI-driven tools as complementary to traditional instructional methods, carefully considering their potential impacts on learning quality and student equity. Educators’ proactive participation ensures these technologies enhance rather than detract from learning experiences, fostering inclusive, accessible, and ethically sound educational environments.

Table 1. Accessibility barriers and mitigation strategies involving the usage of LLMs.

Challenge	Description	Traditional Solutions	Impact and Role of LLMs
Language Barriers	Non-native speakers face difficulties understanding and communicating in the language of instruction.	ESL programs, bilingual tutors, translation tools.	LLMs provide real-time translation, multilingual tutoring, and simplified content explanations, supporting personalized language development. High capacity for assistance but may struggle with nuanced academic terminology or dialects.
Previous Educational Quality	Learners with inconsistent or poor prior education may lack foundational skills.	Diagnostic tests, tutoring, remedial programs.	LLMs offer adaptive explanations, identify knowledge gaps through dialogue, and provide targeted practice. Strong support potential, though oversight by educators is needed to ensure alignment with curriculum.
Learning Disabilities and Special Needs	Students require specialized strategies that account for cognitive or sensory challenges.	IEPs, assistive tech, differentiated instruction.	LLMs can deliver content in multiple formats (text-to-speech, summaries, simplified text), adapt tone and pacing, and support executive function. Moderate-to-high capacity but must be used with accessibility-compliant interfaces.
Cultural Differences	Diverse cultural backgrounds can create disconnects with standardized curricula.	Culturally responsive pedagogy, inclusive materials.	LLMs can localize examples, generate content with cultural sensitivity, and incorporate diverse narratives. Good potential if trained on inclusive datasets; requires guidance to avoid cultural bias.
Socioeconomic Disparities	Unequal access to devices, internet, and materials restricts participation.	Tech grants, free materials, community centers.	LLMs offer low-cost, on-demand educational support accessible via mobile apps or offline versions. High capacity if infrastructure is present; limited impact without basic digital access.

summarizes the key barriers to educational accessibility, paired with specific mitigation strategies to be applied by LLMs.

As summarized in Table 1, LLMs offer promising solutions to some of the most persistent challenges in making education more accessible and equitable. These advanced AI systems are capable of understanding, generating, and adapting language in highly flexible ways, making them valuable allies in the effort to support diverse learners. One of the most immediate applications of LLMs is in addressing language barriers. For students learning in a non-native language, LLMs can provide real-time translation [46], explain complex terms in simpler language [47], and even simulate conversational practice [48]. While traditional ESL (English as a second language) programs remain critical, LLMs supplement these efforts by offering individualized and continuous assistance, especially valuable in asynchronous or online learning environments. In extending the discussion on LLMs’ role to support ESL learners, it is important to emphasize how these technologies can personalize language development pathways [48]. LLMs facilitate real-time translation, allowing students to access instructional materials in their native language while gradually building English proficiency [46]. For example, in a classroom where instruction is delivered in English, a student can use an LLM-powered tool to translate complex terms or entire passages instantly, reducing cognitive overload. Moreover, bilingual tutoring through chat interfaces enables students to ask questions and receive explanations in both English and their first language, fostering deeper understanding [47]. LLMs can also simplify content by rephrasing academic texts into more accessible versions tailored to the learner’s proficiency level [48]. In a practical classroom scenario, an ESL student struggling with a science textbook might engage with an LLM to generate simplified summaries and bilingual glossaries aligned with their reading level. These tools complement traditional ESL programs and bilingual tutoring by offering continuous, adaptive support, even outside class hours. In doing so, LLMs not only reinforce language acquisition [49] but also help maintain learner confidence and engagement, bridging gaps that often hinder second language learners in mainstream educational environments.

In terms of prior educational quality, LLMs help level the gap by identifying misconceptions through interactive questioning and adaptive feedback. These models can offer supplementary instruction tailored to a student’s pace and comprehension level. Unlike static diagnostic tools, LLMs engage learners in dynamic exchanges that can pinpoint misunderstandings and reinforce core concepts. However, human oversight remains essential to ensure alignment with educational standards and avoid reinforcing misconceptions [20].

While much of the current discussion emphasizes LLMs’ role in supporting language learning, their potential in enhancing accessibility and personalization within STEM education warrants greater attention [50]. STEM subjects such as mathematics, computer science, and the natural sciences often present steep learning curves for students from under-resourced backgrounds or those with limited prior exposure. In these contexts, LLMs can serve as powerful assistant tools, offering real-time adaptive support that complements traditional instruction. For instance, in mathematics, an LLM can walk students through problem-solving steps using simple language, visual aids, or analogies tailored to their comprehension level [51]. In computer science education, the integration of LLMs has shown promise in supporting code literacy. As exemplified in [52], aligning human feedback with generated code snippets can aid in reinforcing correct patterns and guiding novice programmers through iterative improvement. In the classroom, this could translate into personalized code suggestions, syntax correction, or real-time debugging assistance, significantly reducing frustration and fostering confidence.

Moreover, for students learning STEM content, LLMs can provide English language scaffolding that contextualizes technical terminology and complex concepts. A student struggling with a physics explanation might receive a simpler version in English along with multilingual support, bridging both conceptual and linguistic gaps. LLMs can also generate problem-solving prompts in conversational English, helping learners articulate reasoning in both spoken and written formats. Such features are not only crucial for comprehension but also for participation in assessments and collaborative STEM activities.

Educators play a pivotal role in ensuring that these AI-generated outputs align with curricular standards and ethical classroom practices [20]. By supervising LLM use, teachers can prevent misuse while tailoring support to meet local linguistic and cultural needs. Integrating LLMs into STEM education, particularly when guided by thoughtful pedagogical frameworks, broadens their utility across disciplines. This expansion not only enhances the inclusivity and adaptability of learning experiences but also underscores LLMs’ capacity to democratize education, especially for learners who face compounded barriers due to linguistic, socioeconomic, or disciplinary challenges.

For students with learning disabilities and special needs, LLMs bring a degree of personalization that few conventional tools can match [40]. They can adjust their tone, structure, and delivery based on the learner’s preferences, supporting different learning styles and cognitive abilities. Text can be converted to simpler formats or even integrated with screen readers and other assistive technologies. This flexibility is particularly helpful for students with dyslexia, ADHD, or processing disorders. Yet, accessibility must be built into the platforms through which LLMs are delivered, as these students may still face difficulties in interacting with standard interfaces [53].

Cultural differences also play a significant role in educational engagement and success. LLMs can be trained or fine-tuned to reflect culturally diverse perspectives [54], offer content relevant to various contexts [55], and generate examples that resonate with specific student backgrounds [56]. This function promotes inclusive curricula and helps students feel seen and respected in their learning environments. Also, to ensure an effective and ethical integration of LLMs in language education, educators must maintain oversight over how these tools handle terminology, regional dialects, and contextual subtleties [57]. By customizing prompts and filtering outputs, instructors can align LLM responses with specific cultural and linguistic norms. This not only enhances relevance and learner engagement but also helps prevent misuse or cheating by embedding LLM use within guided, curriculum-aligned frameworks tailored to classroom objectives and assessment integrity. Nevertheless, a careful curation of training data and monitoring for bias are necessary to ensure cultural sensitivity [37].

Finally, socioeconomic disparities remain a substantial obstacle. LLMs can reduce educational costs by offering free or low-cost tutoring, homework help, and content generation, provided students have access to digital devices and internet connectivity. While their reach is significant, the digital divide must be addressed in tandem to ensure that the benefits of AI-driven learning are equitably distributed. Overall, LLMs hold transformative potential in expanding access and personalization in education, provided they are thoughtfully integrated into broader equity strategies by an expert lecturer.

4. Promoting Inclusion via Individualized Learning

Inclusive education requires a deep recognition and proactive accommodation of individual differences [40], ensuring that all learners can fully participate and thrive in their educational experiences. At the intersection of this goal and emerging technology, LLMs offer powerful capabilities to support personalized educational pathways [22]. Through a sophisticated analysis of student interactions, these systems can dynamically tailor educational materials to match each student’s unique learning trajectory, continuously calibrating the delicate balance between challenges and the support required for optimal learning outcomes [3]. In this way, LLMs provide educators with advanced tools capable of sustaining meaningful differentiation at scale, augmenting but never supplanting the educator’s critical role in guiding learning processes [20].

An essential aspect of LLM functionality in inclusive education is the system’s responsiveness to cultural and linguistic diversity [58]. By rapidly adapting educational content and methodologies to reflect diverse student backgrounds, experiences, and linguistic abilities, LLMs play a central role in cultivating an educational environment where all learners feel recognized, respected, and included [31]. Such culturally responsive adaptability goes beyond a mere translation of content into various languages; it encompasses a thoughtful alignment with students’ cultural contexts, interests, and lived experiences. Consequently, students from varied cultural and linguistic backgrounds experience a deeper sense of belonging and engagement [3], which is foundational to their overall academic success.

Moreover, the adaptability inherent in LLMs directly addresses diverse learning styles and preferences, ensuring educational engagement and comprehension across different students. These models can seamlessly provide instructional content in formats that effectively engage discussion with individuals, whether auditory explanations, visual representations, interactive experiences, or blended modalities [58]. By accommodating different cognitive and sensory preferences, LLMs effectively bridge gaps that traditional, uniform educational approaches often struggle with [3]. Such personalized and adaptive instruction helps ensure that each student has equitable access to educational opportunities and experiences tailored specifically to their strengths and areas of growth.

Beyond the capacity to generate personalized instructions, LLMs hold significant potential to enrich peer collaboration within an inclusive educational setting [59]. LLMs facilitate inclusive dialogue by mediating group interactions, clarifying complex or challenging concepts in real time, summarizing collective inputs, and guiding discussions along productive, inclusive conversational pathways [60]. Through these functions, LLMs actively encourage broader participation, validating diverse student perspectives and empowering quieter or historically marginalized voices. By fostering this kind of inclusive dialogue, LLMs enhance collaborative learning environments, promoting mutual understanding, empathy, and the development of critical interpersonal and cross-cultural competencies among students.

The transformative potential of integrating LLMs into education must always remain grounded in robust educational theory. Effective integration is not merely a technological sophistication but also involves embedding technology within strong pedagogical frameworks. Central to this approach is recognizing that teachers remain indispensable in interpreting the nuanced insights provided by LLM-generated data, setting clear educational goals, and ensuring the emotional, social, and cultural relevance of educational activities. Indeed, educators’ professional judgment, empathy, and contextual understanding are irreplaceable in translating algorithmic insights into genuine, meaningful, and inclusive educational experiences.

To ensure that the inclusive promise of LLMs is fully realized, educational theory must explicitly include principles such as constructivism, scaffolding, and differentiated instruction [61]. Constructivist theories encourage active, student-centered learning, in which knowledge is built through experience, collaboration, and inquiry [62]. Scaffolding ensures that instruction provided by both educators and LLM tools remains appropriately challenging and supportive, facilitating learners’ progression towards greater autonomy and mastery. Differentiated instruction aligns seamlessly with LLM capabilities by promoting personalized learning paths based on continuous assessments of student readiness, interests, and learning profiles.

Ethical considerations also remain critical in leveraging LLMs for inclusive education [12]. Educators must address potential biases inherent in large-scale language models, ensuring that content is equitable and culturally responsive and remains free from unintended biases or stereotypes [63]. Ethical inclusivity also entails the proactive mitigation of digital divides, ensuring equitable access to educational technologies across diverse socioeconomic and geographic contexts. By maintaining careful oversight, educators can ensure that generated content aligns with the educational standards, values, and principles of equity, thus safeguarding educational quality, inclusivity, and ethical integrity. Rather than viewing LLMs as replacements for educators, their most effective and responsible use lies in instructor-guided integration [20]. As proposed in [20], educators safeguard classroom dynamics, not only supervising the academic integrity of AI-enhanced instruction but also the emotional well-being and inclusivity of the learning environment. In

Table 2. The role of LLMs for individualized learning and the mitigation strategies supported by educator guidance.

Inclusive Education Element	LLM Contribution	Educator’s Role	Impact on Learners
Personalized Instruction	LLMs adapt content in real time based on student interactions and learning pace.	Curate and supervise AI-generated materials to ensure pedagogical soundness and emotional resonance.	Enhances learner autonomy, engagement, and mastery through tailored content.
Cultural and Linguistic Responsiveness	Content is customized to students’ cultural backgrounds, interests, and languages.	Validate cultural relevance, prevent bias, and ensure inclusivity in AI outputs.	Promotes a sense of belonging and increases participation for diverse learners.
Multimodal Learning Support	Delivers explanations visually, verbally, or interactively based on student preference.	Match modality to student learning profiles and integrate with curriculum goals.	Supports comprehension across various learning styles and abilities.
Inclusive Collaboration	Mediates peer interaction by clarifying ideas, encouraging quieter voices, and guiding discussions.	Facilitate group work, monitor tone and inclusivity, and intervene when necessary.	Encourages respectful discourse, empathy, and equitable participation.
Ethical and Equitable Use	Operates at scale with potential for wide-reaching impact.	Address algorithmic bias, monitor content accuracy, and bridge digital divides.	Ensures fair access, maintains trust, and upholds ethical standards.

, we summarize how LLMs can promote inclusive, individualized learning when guided by educators who ensure that AI-generated content remains pedagogically sound, culturally relevant, and ethically responsible.

As described in Table 2, inclusive education demands more than simply integrating diverse learners, as it requires actively recognizing and addressing individual differences in culture, language, learning style, and ability. In this regard, LLMs are emerging as transformative tools by enabling personalized learning paths, adapting content to suit individual needs, and supporting inclusive peer collaboration. LLMs can analyze student input to continuously tailor instructional materials, ensuring a balance between challenge and support. These technologies also offer culturally responsive content adaptation, not just in language but in aligning multimodal learning objectives with students’ experiences and identities.

5. Discussion and Limitations of LLMs in Education

The integration of LLMs within educational settings significantly enhances both accessibility and inclusion. By adopting UDL principles and the social model of disability, educational institutions utilizing LLM technology can effectively address the full spectrum of academic, emotional, social, and cultural learner needs. While LLMs demonstrate clear potential to enrich traditional educational frameworks, ensuring genuinely inclusive education needs continued exploration and rigorous research into their long term efficacy and responsible implementation. Through thoughtful deployment and ongoing evaluation, these advanced technologies promise to transform education into a truly inclusive practice, supporting diverse learners equitably.

LLMs present substantial opportunities for improving educational accessibility and inclusivity, whether by supporting multilingual learners, providing assistive technologies tailored to disabilities, or expanding educational access across geographical limitations. However, realizing the full potential of these technological advancements requires the conscientious management of data privacy, biases, and ethical considerations guided by active educator involvement and comprehensive adherence to accessibility standards. Leveraging LLMs thoughtfully and responsibly can substantially advance equitable education, enabling learners of diverse backgrounds and abilities to participate fully and meaningfully in their educational journeys.

Despite the many anticipated benefits of integrating LLMs into educational settings, several hazards remain that require further investigation [1,64,65,66], as they could potentially hinder their effective adoption. As determined in [63], LLMs suffer several hazards that underscore the importance of their evaluation beyond the inadequacy of traditional evaluation metrics. Ref. [63] highlights the growing complexity and societal integration of LLMs and organizes LLM evaluation into three critical dimensions, namely scope, extent, and procedure. While emphasizing the need for assessments that also account for ethics, sustainability, and human experience, central to the manuscript is the identification of a wide array of hazards that compromise both model performance and the integrity of the evaluation itself. These include logical errors such as the Reversal Curse, where models incorrectly infer that if “A implies B”, then “B implies A”, and a broader lack of common sense reasoning, where outputs may be absurd or inaccurate [67,68]. The issue of hallucination is plausible, but factually incorrect content is particularly troubling in domains requiring high factual fidelity [27,69,70]. Additional challenges include interpretability issues, which obscure the decision-making processes of increasingly complex models, and catastrophic forgetting, where new learning overwrites previously acquired knowledge [71,72,73]. Ethical risks are manifest in bias and stereotyping, where societal prejudices are amplified, and misinformation propagation, where LLMs spread inaccuracies absorbed during training [74]. Other technical and user experience hazards include overfitting and memorization [75], adversarial vulnerabilities [76,77], inconsistency in long-term interactions [78], and output toxicity [79]. Cultural and linguistic limitations also arise, as LLMs often underperform in low-resource languages or fail to grasp cultural nuance [65]. Moreover, the study flags concerns over misalignment with human values [80], poor handling of nuanced language [81], and significant environmental impact due to resource-intensive model training [82]. Finally, privacy and copyright risks, alongside benchmark over-reliance, call for more transparent and representative evaluation practices [83]. Altogether, these hazards further stress the urgent need for more integral and interdisciplinary evaluations that can guide the responsible development and deployment of LLMs in educational and broader societal contexts.

6. Conclusions

The integration of LLMs into education represents a critical turning point towards a more inclusive, personalized, and equitable learning environment. As analyzed throughout this work, the thoughtful application of LLMs within educational frameworks guided by educators broadens accessibility and enhances individual learner engagement. Far from replacing educators, these intelligent models best serve when strategically employed by instructors as catalytic tools, augmenting teaching effectiveness through tailored instructional approaches and sustained personalized support.

Educators are uniquely positioned to mediate and maximize the potential of LLMs, guiding their ethical deployment, ensuring culturally sensitive adaptation, and fostering emotionally supportive interactions. When instructors actively integrate these technologies, they facilitate individualized learning pathways, dynamically responsive educational dialogues, and inclusive collaborative environments. This proactive utilization of LLMs transforms traditional pedagogies, shifting from rigid standardized instruction towards learner-centered methodologies that empower all students to thrive academically, socially, and emotionally.

The effective integration of these sophisticated AI tools requires ongoing and continuous vigilance to address challenges such as data privacy, bias, and equitable digital access. As education embraces this innovative intersection of technology and human expertise, continuous research, reflective practice, and responsible oversight are paramount. Ultimately, by fully embracing the instructor-led integration of LLMs, educational institutions can substantially elevate the quality, accessibility, and inclusiveness of the teaching and learning process, laying a robust foundation for future generations to learn not just effectively but inclusively across diverse global contexts.

LLMs should be conceptualized as sophisticated tools serving to extend the reach and enhance the efficacy of learner-centered pedagogy, never as replacements for educators. In the ideal inclusive classroom powered by advanced technological capabilities, human educators retain primary responsibility for interpreting and contextualizing LLM-generated insights, adapting content to reflect classroom values and local contexts, and fostering emotionally supportive learning environments. The nuanced connection between human teachers and sophisticated LLM tools embodies a progressive educational model in which the human component remains central and technology acts as a complementary force, broadening access, fostering inclusion, and empowering students from all backgrounds to succeed.

By combining the technological strengths of large language models with a thoughtful grounding in educational theory, inclusive education can be significantly strengthened. LLMs provide unparalleled opportunities for personalization, cultural responsiveness, accessibility, and differentiated instruction, enhancing educational equity and engagement for diverse learners. However, it is only through deliberate human oversight, informed by robust pedagogical principles and ethical practices, that these technological advancements can be most effectively employed to foster inclusive, meaningful, and equitable learning experiences for all students.