Unequal Gains: The Divergent Impact of AI Literacy on Mental Health Across Socioeconomic Groups

Abstract

Artificial intelligence (AI) technologies are becoming increasingly integrated into the everyday lives of children, influencing how they learn, communicate, and develop emotionally. As AI assumes a more central role in children’s digital ecosystems, AI literacy—the ability to understand, engage with, and make informed decisions about AI systems—is no longer a luxury but a developmental necessity. This review explores how AI literacy intersects with children’s mental health, particularly through the lens of socioeconomic status. Drawing on Digital Capital Theory and Cumulative Advantage/Disadvantage Theory, the paper examines how inequalities in access to AI-related resources shape the emotional and psychological experiences of children. It argues that while AI literacy can enhance well-being across all social groups, its impact is especially transformative for children from low-income backgrounds. Children from middle- and high-income families often experience modest emotional gains from AI engagement, having already benefited from consistent digital exposure and support. In contrast, low-income children—who often begin with limited access and lower confidence—stand to gain disproportionately in terms of emotional resilience, self-esteem, and digital confidence when their AI literacy improves. The review concludes with policy and practice recommendations that prioritize equitable access and tailored interventions, especially for underserved populations who have the most to gain from both the cognitive and emotional benefits of AI literacy.

1. Introduction

In today’s rapidly evolving digital society, artificial intelligence (AI) has begun to influence nearly every domain of children’s lives. Whether through personalized learning platforms, algorithmic recommendations on social media, or virtual assistants embedded in household devices, AI is becoming a constant—if often invisible—presence [1,2,3]. As children interact with these technologies from a young age, there is growing recognition that digital fluency must include the ability to understand and critically navigate AI systems. We define AI literacy as the ability to recognize what AI is and is not, understand how AI systems are trained and make decisions, critically evaluate outputs for accuracy and fairness, interact with AI systems safely and responsibly, and reflect on the broader social and ethical implications of AI. This definition establishes AI literacy as distinct from broader digital literacy or general technology use. This competency encompasses both technical knowledge and emotional adaptability in digital environments [4,5,6].

At the same time, rising mental health concerns among children have sparked global discussions around psychological resilience and digital well-being. Anxiety, depression, and social isolation are increasingly prevalent among youth, and technology—while often implicated as a cause—can also serve as a critical resource for coping and connection [7,8,9,10]. The intersection of AI literacy and mental health thus offers a powerful lens for understanding how digital engagement can either exacerbate or alleviate emotional challenges. However, these opportunities are not distributed equally. Children from affluent families tend to have earlier and more consistent exposure to AI technologies. They are more likely to have digitally fluent parents, access to internet-enabled devices, and schools equipped with AI-integrated learning tools. In contrast, children from low-income households often face multiple, compounding barriers: lack of devices, unreliable internet access, limited adult support, and insufficient institutional investment [11,12,13]. These disparities do more than inhibit learning—they shape children’s sense of competence, belonging, and psychological well-being in increasingly digitized spaces [14,15]. Although children’s mental health is influenced by broad psychosocial factors such as family stress or baseline anxiety, AI literacy supports well-being through distinct mechanisms. Understanding how AI systems make decisions reduces digital uncertainty, enhances perceived control, and strengthens digital self-efficacy. These mechanisms directly support self-esteem and facilitate more adaptive emotional regulation in AI-mediated environments.

In recent years, these inequalities have become increasingly visible as AI applications are woven into formal and informal educational settings. Affluent schools may adopt adaptive learning platforms, integrate AI tutoring systems, and provide teachers with data-driven insights, while underfunded schools struggle with outdated devices, minimal training, and inconsistent access. At home, children in wealthier households may learn to question AI recommendations critically because parents explain how algorithms shape their digital environment. By contrast, low-income children often interact with the same systems passively, without guidance or context, reinforcing feelings of dependency or confusion. This divergence illustrates how early exposure to AI literacy not only affects immediate learning outcomes but also establishes long-term patterns of psychological security or insecurity in digital contexts.

The mental health consequences of these disparities can be profound. Children who lack confidence in navigating AI-driven technologies may feel left behind in social and academic settings, leading to stress, frustration, and diminished self-esteem. Conversely, children who are equipped with AI literacy skills often feel more capable, resilient, and adaptable, qualities that buffer against psychological distress. This growing divide raises important questions about equity: who is most likely to benefit from AI’s integration into everyday life, and who is at risk of being further marginalized?

This review aims to understand how AI literacy impacts children’s mental health and how that impact varies by socioeconomic background. More specifically, it explores whether improving AI literacy has the potential to enhance psychological well-being among disadvantaged children to a greater extent than among their more privileged peers. By drawing on interdisciplinary research, the paper seeks to inform practical strategies for intervention and investment that center equity and inclusion. Throughout this paper, we use the term “mental health” to cover two domains: (a) clinical indicators of psychological functioning, such as anxiety, depressive symptoms, and stress, and (b) psychosocial outcomes such as self-efficacy, resilience, belonging, and self-esteem. When studies focus on the latter, we identify them as psychosocial outcomes rather than clinical measures. This distinction ensures clarity in interpreting findings and avoids overstating effects.

2. Theoretical Framework

This review is guided by two complementary theoretical frameworks: Digital Capital Theory and the Cumulative Advantage/Disadvantage (CAD) Theory [16,17,18]. They offer a nuanced understanding of how access to digital technologies—and the skills to use them effectively—can influence children’s mental health outcomes across socioeconomic contexts.

Digital Capital Theory frames digital skills, technological access, and social supports as resources that operate similarly to economic and cultural capital [18]. In this framework, digital capital enables individuals to navigate the digital world effectively, increasing their chances of success in education, employment, and social life. For children, digital capital can mean having access to a tablet or smartphone, receiving parental guidance on how to use technology safely, and being enrolled in schools that teach digital literacy. These factors collectively influence not just what children can do online but how they feel while doing it—whether they approach technology with curiosity and confidence or with anxiety and avoidance [18].

Beyond individual outcomes, digital capital also shapes social identity and peer relationships. Children with strong digital skills may become leaders in group projects or social networks, while those without such skills risk exclusion, ridicule, or marginalization. For low-income children, the absence of digital capital can translate into compounded disadvantages: they are less prepared for future opportunities, and they may also internalize feelings of inferiority and isolation. Over time, these psychological burdens reinforce cycles of disadvantage, creating broader inequities that extend far beyond the digital realm.

To understand the long-term consequences of digital inequalities, this review also draws on the Cumulative Advantage/Disadvantage Theory. CAD theory posits that initial advantages (or disadvantages) tend to reinforce themselves over time [16,17]. In other words, children who start with better access to AI technologies and support are more likely to build on those experiences, gaining confidence, expanding their skills, and experiencing fewer frustrations. In contrast, children who face early barriers may fall further behind, not only in their technical competence but also in their psychological relationship with technology. They may experience digital environments as intimidating, alienating, or hostile—feelings that can compound into broader emotional and cognitive challenges [16,17].

This compounding process means that even relatively small gaps in early digital exposure can produce significant long-term disparities. A middle-income child who casually explores an AI-based educational game at home may later approach advanced digital tools in school with confidence, while a low-income child who missed that opportunity may experience anxiety or avoidance. Over time, these differences accumulate into diverging developmental trajectories, with privileged children advancing both academically and emotionally, while disadvantaged peers risk stagnation or decline. CAD theory thus provides a critical lens for understanding why interventions must occur early and be sustained—because once disadvantages accumulate, reversing them becomes far more difficult.

By integrating Digital Capital Theory and CAD theory, this review highlights the importance of early, context-sensitive interventions. AI literacy is not just a neutral skill set—it is a developmental resource that can either widen or narrow existing psychological and social disparities, depending on how it is distributed.

3. Methods

This review adopts a narrative approach to synthesize findings from multiple disciplines, including psychology, education, digital sociology, and public health. A narrative review is especially well suited to emerging and interdisciplinary topics, allowing for an integrative examination of both theoretical and empirical literature. Academic databases including PubMed, PsycINFO, Scopus, and Google Scholar were searched using terms such as “AI literacy,” “digital divide and children,” “mental health and technology use,” “low-income youth and digital access,” and “psychological outcomes of AI in education.” Inclusion criteria focused on peer-reviewed articles published between 2010 and 2025 that involved children or adolescents under 18 and addressed both digital technology use and mental health dimensions. Both quantitative and qualitative studies were considered. Study quality was assessed based on methodological clarity, peer-review status, and relevance to children or adolescents. Socioeconomic status was operationalized in varied ways across the literature, including household income, parental education, school resources, and neighborhood deprivation. The initial search identified approximately 350 records between 2010 and 2025. After removing duplicates, about 200 records were screened by title and abstract. Roughly 90 articles were reviewed in full text, with 25 ultimately included. Eligibility and exclusion criteria. Studies were eligible if they (a) involved participants under 18 or school contexts serving this age group; (b) examined AI literacy, adjacent computing literacy, or digital literacy with clear relevance to AI-mediated contexts; and (c) reported at least one psychosocial or clinical mental-health-related outcome or construct (e.g., self-efficacy, belonging, anxiety symptoms). Studies were excluded for “low methodological transparency” if key elements were missing (e.g., undefined sample, incomplete procedure or measures, or unreported analysis approach). Studies were excluded for “high risk of bias” when they relied solely on anecdotal evidence, had uncontrolled designs with strong causal claims, or presented outcomes without adequate description of measurement or timing. Two reviewers independently screened titles/abstracts for relevance, resolved disagreements by discussion, and prioritized studies that (a) clearly defined constructs, (b) used transparent methods, and (c) directly informed the thematic domains. Articles were grouped thematically into four domains: (1) AI literacy as a psychological resource, (2) barriers to AI engagement by income level, (3) comparative effects on mental health, and (4) effective intervention strategies. Particular attention was paid to studies incorporating child or caregiver perspectives to better capture the lived realities of digital inequality. Themes were developed using inductive analysis while being guided by the two theoretical frameworks.

4. Findings

4.1. AI Literacy as a Psychological Resource

Studies consistently suggest that children with higher digital and AI literacy report stronger self-efficacy, improved psychological well-being, and better problem-solving capacities and cognitive skills [19,20,21,22]. In most cases, these outcomes are described in terms of psychosocial indicators—such as digital confidence, sense of belonging, or self-esteem—rather than clinical measures of mental health. When children understand how AI works and how to interact with it confidently, they feel more in control of their digital environments, which reduces stress and fosters adaptive emotional responses. Exposure to AI tools, when paired with guided exploration, can also build resilience by helping children navigate ambiguity and develop critical thinking skills [22,23,24]. In particular, children who have higher digital literacy (for example, they better understand AI-driven recommendation systems, chatbot interfaces, or interactive educational tools) often report greater confidence and digital agency in online settings [25].

AI literacy also influences emotional growth. For instance, learning how to manage privacy settings or identify biased recommendations teaches children not only technical strategies but also fosters a sense of autonomy and control. This autonomy reduces feelings of helplessness that can arise in digital environments, especially for vulnerable youth. Furthermore, AI literacy can serve as a protective factor against digital risks such as cyberbullying or misinformation, by equipping children with the tools to evaluate content critically and respond effectively. By framing AI literacy as both a cognitive and emotional resource, its role in shaping mental health becomes more visible and significant.

4.2. Socioeconomic Disparities in Digital Exposure

Children from affluent families often enjoy consistent access to devices, high-speed internet, and parent-supported learning. In contrast, low-income children may share devices among family members, rely on mobile data plans, or attend schools with outdated technology [11,12,13]. These material gaps contribute to skill gaps, which then reinforce psychological disparities [11,26,27]. Moreover, the environments in which children use technology—quiet, supportive, and safe spaces versus chaotic or resource-limited ones—can greatly affect how they internalize digital experiences. Early digital frustrations without adult support can result in technostress, reduced digital curiosity, and even tech avoidance behaviors in under-resourced communities. In low-income households, parents may themselves lack digital literacy, making it difficult to provide guidance. This absence leaves children to navigate AI-driven environments alone, often resulting in heightened anxiety or dependence on peers who may be equally uninformed. Thus, socioeconomic disparities create layered disadvantages—material, educational, and emotional—that converge to widen the digital divide.

4.3. Disproportionate Mental Health Benefits Among Low-Income Youth

The reviewed studies indicate that while all children can benefit psychologically from AI literacy, those from low-income backgrounds experience more pronounced effects. Because they begin with less exposure and confidence, even moderate gains in digital competence are associated with improvements in emotional regulation, self-esteem, and social belonging. These children often face systemic disadvantages that undermine their sense of agency, inclusion in educational or digital environments, and confidence in learning or participating in AI literacy education programs [28]. Direct evidence on AI-specific literacy programs is still limited, so some conclusions should be viewed as theoretically grounded expectations rather than established causal effects. Gaining fluency in AI tools can reverse this dynamic, offering a concrete sense of progress and control. When low-income youth receive structured and supportive opportunities to engage with AI technologies, their levels of digital anxiety decrease, and their sense of competence rises. As they begin to see themselves as capable digital learners and creators, their psychological well-being improves through enhanced self-perception and a stronger connection to peers engaged in similar learning experiences [29]. These benefits are particularly significant for children who may otherwise feel excluded from technology-rich futures. In this way, AI literacy can function as both a skill and a catalyst for mental health resilience among disadvantaged populations.

While middle- and high-income students may refine existing skills when exposed to AI literacy programs, low-income students often make transformational shifts in both confidence and mental health. Such improvements can extend into other domains, such as academic motivation, peer interactions, and family dynamics. The ripple effect is therefore larger among disadvantaged youth: learning to master AI tools not only changes their relationship with technology but also alters their self-concept and emotional resilience. This suggests that targeted AI literacy interventions could serve as powerful equalizers in addressing mental health inequalities.

4.4. Holistic Intervention Strategies

Programs that successfully promote AI literacy among disadvantaged youth often pair technical instruction with emotional scaffolding. For example, community-based workshops that blend peer mentoring/learning, hands-on AI learning, and mental health check-ins show promise in building both skill and resilience [2,30,31,32]. These initiatives prioritize safe environments and emotional safety as much as curriculum design. Furthermore, interventions which are sustained, culturally responsive, and embedded within familiar community settings are important and may be more effective [31,33]. Schools, libraries, and youth centers can serve as hubs for AI education when equipped with the appropriate infrastructure and trained personnel [31]. Involving caregivers, community leaders, and older peers in the delivery of AI content can help build trust and normalize digital exploration in low-income contexts. The design of AI literacy programs should incorporate trauma-informed practices and acknowledge the unique stressors faced by low-income youth. Programs must be sensitive to the emotional labor required for marginalized children to engage with unfamiliar or intimidating technologies. Integrating expressive activities—such as storytelling, digital art, or collaborative projects—can foster a sense of belonging and promote both technological and emotional fluency/inquiry [2,24,30,31]. Moreover, linking AI literacy with real-world problem solving may enhance motivation and relevance [2,24,30,31]. For instance, youth-led projects that use AI tools to address community challenges (e.g., environmental monitoring, food access, neighborhood safety) not only develop technical skills but also empower participants to view themselves as agents of change. Such experiences can significantly boost self-esteem, peer connection, and purpose—factors known to buffer against anxiety and depression. Some researchers have stated that AI literacy education should combine digital skill-building with social-emotional learning [30]. These holistic models often incorporate teachers, counselors, and families as co-educators, ensuring that children are supported across all domains of development. Evaluations of such interventions have shown improvements not only in digital proficiency but also in classroom behavior, peer relationships, and self-regulation [2]. These holistic approaches demonstrate that AI literacy is not only a cognitive goal but also a relational and emotional one. Programs that respect this complexity are better positioned to produce long-lasting benefits for mental health and digital inclusion alike. Ultimately, holistic interventions remind us that AI literacy is not simply about preparing children for future jobs or academic success. It is about equipping them with the emotional and social resources needed to thrive in a technology-saturated world. By recognizing the intersection of digital skills and psychosocial development, such programs can deliver enduring benefits that extend far beyond the classroom.

5. Discussion

This review highlights how AI literacy functions not only as a technical skill but also as a form of psychological empowerment. For children, especially those from marginalized backgrounds, learning to navigate AI technologies can contribute to a stronger sense of agency, competence, and belonging in a rapidly digitizing world [14,25]. At the same time, it is important to note that AI use can carry risks, including overreliance on automated systems, exposure to harmful or misleading content, and the psychological impact of algorithmic prejudice. AI literacy is therefore valuable not only for fostering resilience and confidence but also for helping children identify and manage these potential stressors. Yet access remains deeply unequal, shaped by systemic socioeconomic factors that create barriers to engagement [11,12,13]. Digital Capital Theory helps explain how access to devices, supportive environments, and guided instruction enables some children to flourish in digital contexts while others are left behind [18]. Simultaneously, Cumulative Advantage/Disadvantage Theory shows how these early inequities compound over time, resulting in long-lasting mental health and developmental differences [16,17]. AI literacy, then, is not only about preparing children for a digital future but also about redressing social and emotional inequalities rooted in systemic disadvantage. Interventions must be inclusive, sustained, and emotionally supportive [31,33]. They should focus on building not only cognitive skills but also digital confidence. Programs targeting low-income communities should include reliable access to devices, engaging and age-appropriate instruction, and environments that foster curiosity without judgment. Without such intentional design, well-meaning digital education efforts risk deepening the very divides they aim to solve.

AI literacy also moderates emotional regulation through specific cognitive pathways. Children who understand how algorithms operate can better anticipate system outputs, evaluate unexpected recommendations, and manage ambiguous digital situations—reducing stress and increasing their sense of control. These domain-specific competencies distinguish AI literacy from general psychosocial variables such as depression or social stress.

One of the central insights from this review is that the psychological impact of AI literacy is not uniform but stratified by socioeconomic context. For middle- and high-income children, AI literacy often supplements an already rich set of digital skills, producing incremental benefits such as improved efficiency, academic performance, or digital confidence. For low-income children, however, the same interventions can produce transformative outcomes, altering their self-perception, emotional resilience, and capacity to participate fully in educational and social life. This suggests that the marginal utility of AI literacy is greater for disadvantaged groups, underscoring the need for targeted interventions.

Another important point concerns the psychosomatic dimension of inequality. Caregivers and educators in low-income contexts often face heightened stress when children struggle with digital tools, creating environments where frustration and anxiety become normalized. By strengthening AI literacy, both children and caregivers may experience reduced psychological strain, which in turn benefits family and community dynamics. Thus, the value of AI literacy extends beyond individual children to influence broader relational and psychosocial networks.

Beyond disparities in access and skills, children face additional stressors linked to the nature of AI systems themselves. Algorithmic bias, privacy concerns, and data surveillance can heighten feelings of vulnerability, particularly among marginalized groups. AI literacy that includes critical awareness of fairness, transparency, and data practices is therefore essential. These competencies also align with children’s rights perspectives, emphasizing that equitable AI literacy is both an educational and a rights-based issue. Developmental stage matters as well: for younger children, programs may focus on guided curiosity and safe exploration, while for adolescents the emphasis may shift toward autonomy, identity, and ethical reflection. At the same time, it is important to acknowledge concerns that AI-based educational systems could unintentionally create environments where children who struggle with AI are left behind or feel neglected. Some critics also raise the possibility that increasing reliance on automated systems to evaluate children may risk narrowing the space for human emotional development. For these reasons, we stress that AI literacy initiatives must remain human-centered and emotionally supportive, ensuring that efforts to close socioeconomic gaps do not come at the cost of diminishing children’s broader psychological and social growth.

This discussion highlights a broader societal responsibility. Governments, educational institutions, and technology developers all play a role in shaping the digital ecosystems that children inhabit. Without deliberate policies and sustained investment, market forces alone are unlikely to close the digital divide. Equitable access to AI literacy must be framed as a public good, essential for both mental health and social cohesion.

6. Conclusions

AI literacy is increasingly recognized as a foundational skill for thriving in contemporary society. Beyond its technical utility, it bears significant implications for emotional and psychological development—particularly among children and adolescents. This review has demonstrated that while AI literacy supports positive mental health outcomes across all socioeconomic groups, it holds disproportionately high promise for youth from low-income backgrounds. These children often experience compounded disadvantages [11,12,13], and the acquisition of AI literacy can serve as a meaningful pathway toward empowerment, confidence, and emotional resilience. However, realizing this potential requires more than distributing devices or offering sporadic digital workshops. A comprehensive, equity-driven framework must guide AI literacy promotion. Such a framework must account for disparities in access, learning environments, support systems, and cultural context. It must treat AI literacy not merely as a technical competency but as a developmental and psychological asset.

This review suggests that resources should be strategically allocated to support low-income youth who stand to gain the most from AI literacy interventions. Policymakers and funders should prioritize this demographic in educational programs designed to foster digital competence and emotional well-being. Sustainable infrastructure must also be developed, ensuring consistent access to devices, high-speed internet, and safe, supportive learning environments. In addition, mental health supports should be intentionally integrated into AI curricula. Rather than viewing technology and mental health as separate domains, programs should address children’s emotional experiences with AI, helping them develop self-efficacy, manage frustration, and build collaborative skills. Such integration fosters both technological fluency and psychological resilience.

Effective implementation will require cross-sector collaboration, bringing together educators, mental health professionals, technologists, and community stakeholders. These collaborative efforts can ensure that programs are grounded in both technical rigor and socio-emotional relevance. Importantly, participatory design should be central to all interventions. Engaging children and families from underserved communities in the development and delivery of AI literacy programs increases their cultural relevance, accessibility, and impact.

Policy and practice recommendations can be organized around two design principles: (a) building digital capital through equitable access and caregiver or teacher support, and (b) interrupting cumulative disadvantage through sustained and scaffolded exposure to AI. To enhance feasibility, we propose concrete actions: integrate AI literacy modules into existing school-day lessons rather than creating entirely new programs; designate libraries and youth centers as community hubs where children can access devices and trained facilitators; and develop short caregiver training workshops that can be delivered at schools or community centers at minimal additional cost. Ensuring device access and stable connectivity requires budget allocation, which could be supported by public education budgets supplemented by community partnerships and targeted technology grants. Ministries of education can set curricular frameworks, local school districts can prioritize budget lines for digital literacy infrastructure, and community organizations can contribute by tailoring programs to local cultural needs. Where evaluations indicate benefits, we specify the outcomes measured; where evidence is less direct, claims are framed cautiously to avoid overstatement.

In conclusion, AI literacy is more than an educational trend—it is a form of digital empowerment that intersects deeply with mental health, social inclusion, and developmental equity. As societies navigate the AI-driven future, we must ensure that every child, regardless of socioeconomic status, has the tools, support, and confidence to thrive. Equity must be the guiding principle, not the afterthought, in our collective efforts to build AI-literate and emotionally resilient generations.