What 11 Studies Reveal About AI's Real Impact on English Learners in Arab Universities

Out of 184 studies initially identified, only 11 survived the screening process for a new systematic review examining how AI tools are actually helping (or not helping) students learning English as a second or foreign language across Arab university classrooms. That steep drop-off, from nearly 200 candidate papers down to a handful meeting rigorous inclusion criteria, is itself a telling data point about how thin the empirical base still is for a technology now embedded in classrooms worldwide.

The review, posted to arXiv in June 2026, pulled its source material from three major academic databases: Google Scholar, Web of Science, and Scopus. Researchers narrowed their scope to studies published between January 2023 and August 2025, capturing the period in which generative AI tools moved from novelty to near-ubiquity in higher education.

Enthusiasm Doesn't Always Translate to Mastery

The synthesized findings paint a nuanced picture rather than a simple endorsement. Students report genuinely positive attitudes toward using AI for drafting text, revising drafts, and practicing language skills, suggesting these tools have found real footing as everyday study aids. But the review draws a sharp distinction between where AI clearly helps and where the evidence gets murkier.

Gains were most consistent and measurable for what the researchers describe as surface-level outcomes, the kind of mechanical improvements like grammar correction, spelling, and basic sentence structure that AI writing assistants are well-suited to catch. When it comes to deeper competencies, though, such as higher-order writing quality (organization, argumentation, critical thinking on the page) or spoken language proficiency, the results were mixed at best. Crucially, those gains often depended on whether a teacher was actively mediating the AI use rather than students working with the tools unsupervised.

A Call for Structure, Not Restriction

Rather than recommending universities pull back on AI adoption, the review's authors push in the opposite direction: they argue for more deliberate scaffolding. Their recommendations center on three pillars: structured, staged integration of AI tools into curricula, dedicated training for instructors on how to guide that integration, and reflective assignments designed to keep students from leaning on AI as a crutch rather than a learning aid.

That framing matters for how institutions read the data. The takeaway isn't that AI tools fail language learners, it's that the tools work best as a supplement to pedagogy rather than a replacement for it. Left unsupervised, students seem to get real value on the easy wins (cleaner grammar, faster drafts) while the harder skills, the ones that actually define fluency and command of a second language, still seem to require a human in the loop.

For an education sector under pressure to adopt AI quickly, the review offers a useful corrective: enthusiasm among learners is real and measurable, but so is the gap between what AI improves easily and what still demands deliberate teaching.