Unreliable AI-generated output is the most frequently reported technical limitation of AI-supported undergraduate EFL instruction, appearing in 22 of 54 studies (41%)¶

Claim · All claims
Evidence · 1 study · review r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? Across 54 empirical studies, unreliable AI-generated output was the most frequently reported technical limitation, reported in 22 studies (41%). → Espinoza-Celi V and Vargas-Saritama A 2026

Evidence¶

Espinoza-Celi V and Vargas-Saritama A 2026¶

Espinoza-Celi V and Vargas-Saritama A (2026) Technical and pedagogical limitations of artificial intelligence in undergraduate EFL higher education: a systematic review. Front. Educ. 11:1948091. https://doi.org/10.3389/feduc.2026.1948091

q2 · i? · review · r2

Qualitative evidence synthesis of 54 Scopus-indexed empirical studies (2021–2026) on AI-supported undergraduate EFL instruction; study-level frequencies from the cross-study coding show "unreliable AI-generated output was reported in 22 studies (41%)". Percentages are descriptive prevalence, not effect sizes.

"Within this category, unreliable AI-generated output was reported in 22 studies (41%), followed by limited contextual understanding in 16 studies (30%), while prompt sensitivity, speech-recognition limitations, hallucinations, and feedback limitations were reported less frequently across the reviewed studies."

Discussion¶

Related Claims¶

• Limited contextual understanding of pragmatic, cultural, and discourse information is a major technical limitation, reported in 16 of 54 studies (30%) — related
• Academic integrity concerns were the third most prevalent pedagogical limitation, reported in 20 of 54 studies (37%) — related
• The most frequent co-occurrence pattern pairs AI system performance limitations with learner cognitive and autonomy risks, reported together in 25 studies (46%), as association rather than causation — related
• Infrastructure and access constraints were reported in 18 of 54 studies (33%) as the second higher-order technical limitation category — related
• Overreliance on AI is the most prevalent pedagogical limitation (27 studies, 50%), followed by reduced critical thinking (25 studies, 46%) — related
• Reviewed literature reports risks of overreliance, AI errors in complex tasks, and students' uncritical acceptance of AI-generated outputs — a broader claim this one bears on
• Unstructured ChatGPT use in programming education is linked to overreliance, reduced persistence, unreliable outputs, and academic integrity risks — related
• Persistent challenges include feedback inaccuracy, learner overreliance, and limited transparency — related
• The most critical challenge in integrating AI into instructional design is pedagogical alignment (14 of 28 studies, 50.0%) — related
• The most frequent limitation of AI in instructional design is content reliability and transparency (16 of 28 studies, 57.14%) — related
• Technical difficulties were the most frequently reported implementation limitation, mentioned in ten of the thirteen studies — related
• Most participants articulated technical and contextual limitations of AI evaluation, including handwriting misreading and unawareness of the instructor's rating system — related