AI literacy divide threatens educational equity in diverse colleges

First-generation and minority students reported lower initial comfort with using the AI platform compared to non-first-generation and majority group students. These disparities were linked to differences in technological capital and prior digital exposure. Despite this, minority and first-generation students rated the simulation’s contribution to their understanding of the course material more highly than their peers, suggesting that when adequately supported, these students found value in the tool.

Institutions around the world are embracing artificial intelligence to enhance teaching, learning, and academic engagement. Beneath the surface of this AI-led digital transformation lies a critical question: Are these innovations truly inclusive, or are they silently reinforcing the very inequalities they aim to dismantle?

A groundbreaking new case study investigates whether artificial intelligence tools are leveling the educational playing field or unintentionally entrenching existing academic disparities. The study, titled "Artificial Intelligence in Higher Education: Bridging or Widening the Gap for Diverse Student Populations?" and published in Education Sciences (2025, Vol. 15, Issue 5), explores the impacts of AI-based simulations on learning experiences, outcomes, and engagement among diverse student groups at a peripheral Israeli college.

Do AI-based tools improve learning experiences for all students equally?

The study assessed 110 students in an introductory psychology course that integrated a generative AI simulation designed to teach psychological theory through interactive prompts. Initial findings revealed that all students reported improved understanding of course material after engaging with the AI tool. However, this general success masked stark differences beneath the surface.

First-generation and minority students reported lower initial comfort with using the AI platform compared to non-first-generation and majority group students. These disparities were linked to differences in technological capital and prior digital exposure. Despite this, minority and first-generation students rated the simulation’s contribution to their understanding of the course material more highly than their peers, suggesting that when adequately supported, these students found value in the tool.

Qualitative data from interviews reinforced this divide. While non-first-generation students approached the technology with excitement and experimentation, their marginalized peers expressed apprehension, self-doubt, and difficulty navigating setup and functionality. Instructor support played a critical role in overcoming these barriers, yet many students still felt embarrassed to ask for help.

A longitudinal pattern emerged: initial resistance gradually gave way to confidence as students adapted, though the nature of engagement remained stratified. While some students used the AI strictly as a course tool, others, particularly those with higher prior exposure, began to experiment with prompt creation and cross-disciplinary applications.

Do AI-based simulations translate to broader academic success?

The simulation proved effective in promoting course-specific comprehension, but the benefits did not extend uniformly across academic domains. Quantitative data showed that non-first-generation and majority group students were significantly more likely to use AI tools outside of the class. They applied the technology to semester assignments, exam preparation, and professional communication such as drafting formal emails. In contrast, only 14.7% of first-generation students used AI for assignments, compared to 93.4% of their non-first-generation counterparts.

A particularly striking disparity appeared in the application of AI knowledge beyond the psychology course. While majority students demonstrated high transference of skills to other fields, minority and first-generation students were far less likely to apply what they had learned, despite recognizing the value of the simulation for their coursework. This gap, quantified with a large effect size (Cohen’s d = 2.63), raises questions about whether AI literacy alone is enough if students lack the confidence or frameworks to use it meaningfully across disciplines.

The study also found engagement patterns diverging over time. Non-first-generation students often moved from structured interaction to independent exploration. In contrast, first-generation and minority students tended to stay within the predefined bounds of the simulation, hesitant to explore additional capabilities. The report frames this not as a lack of interest, but as a reflection of deeper social and educational inequities tied to cultural capital and academic self-efficacy.

What factors influence how students use and benefit from AI?

The study identifies three key determinants shaping the educational impact of AI tools: technological exposure, cultural-linguistic inclusivity, and student self-efficacy.

Students who had prior experience with technology or coding found the AI intuitive and were more likely to experiment beyond the simulation’s instructional boundaries. Meanwhile, students from households without computers or consistent internet access found the tool alien and intimidating at first. Language also played a role: students appreciated being able to use the simulation in their native tongues, such as Arabic, which helped bridge comprehension gaps and alleviate anxiety.

The simulation was designed to be culturally sensitive, but familiarity with the examples and academic norms still varied. Students from majority groups found the case studies relatable, whereas others struggled to connect them with lived experience. These cultural mismatches reduced the simulation’s accessibility and inadvertently limited the development of higher-order analytical skills for some students.

Academic self-efficacy emerged as another powerful influence. Students confident in their academic abilities were more likely to persist through challenges, explore advanced uses, and derive long-term benefits. Conversely, students with lower self-perceived competence often hesitated to engage deeply with the technology, even when support was offered.

A conceptual framework proposed by the study connects cultural and technological capital with three mediating factors: AI literacy, AI-enhanced cognitive flexibility, and AI engagement patterns. These variables jointly influence how students adopt and benefit from AI in academic contexts. Crucially, the study warns that without targeted interventions, AI tools risk reinforcing rather than reducing existing educational disparities.