Technical and pedagogical limitations of artificial intelligence in undergraduate EFL higher education: a systematic review

Abstract

Artificial intelligence (AI) has become increasingly integrated into undergraduate English as a Foreign Language (EFL) education; however, evidence regarding its technical and pedagogical limitations remains fragmented. This systematic review synthesized empirical evidence on the technical and pedagogical limitations of AI-supported EFL instruction and examined patterns of co-occurrence between these limitations across studies. Following the PRISMA 2020 guidelines, a qualitative evidence synthesis was conducted using Scopus as the sole database. The final search was conducted on March 24, 2026, and identified 284 records. After database filtering, 104 records were screened, 75 full-text reports were assessed for eligibility, and 54 empirical studies met the inclusion criteria and were included in the synthesis. The included studies, published between 2021 and 2026, were analyzed through a higher-order analytical coding framework. The synthesis identified seven technical limitations grouped into two higher-order categories. Unreliable AI-generated output was reported in 22 studies (41%), and limited contextual understanding in 16 studies (30%). Eight pedagogical limitations were grouped into five higher-order categories, with overreliance on AI reported in 27 studies (50%) and reduced critical thinking in 25 studies (46%). The most frequent co-occurrence involved AI system performance limitations and learner cognitive and autonomy risks, reported together in 25 studies (46%). These patterns indicate association rather than causal direction. The findings indicate that the educational challenges associated with AI-supported EFL instruction extend beyond technological performance and reflect recurring associations among technological capabilities, instructional design, learner behavior, teacher mediation, and institutional conditions. By integrating technical and pedagogical limitations into a higher-order analytical framework, this review provides an evidence-based foundation for teacher education, instructional design, institutional policies, and the responsible implementation of AI in undergraduate EFL higher education.