AI in EFL: A Critical Guide for the Arab World & Libya

Artificial Intelligence–Mediated EFL/ELT: A Critical, Context‑Sensitive Introduction (with a Focus on the Arab World and Libya

Artificial intelligence (AI) is no longer a peripheral add‑on to Computer‑Assisted Language Learning (CALL); it now shapes learner–content, learner–teacher, and learner–learner interactions across the EFL/ELT ecosystem—from adaptive practice and automated feedback to conversational agents and large language models (LLMs) (Son et al., 2023; Crompton et al., 2022).  State‑of‑the‑art reviews consistently show measurable learning benefits alongside non‑trivial risks that must be addressed through pedagogy, policy, and professional learning (Kundu & Bej, 2025; Sharadgah & Sa’di, 2022).  In empirical syntheses specific to EFL/ELT, AI interventions have produced gains across reading, writing, vocabulary, listening, and speaking, while also surfacing concerns about bias, explainability, teacher role reconfiguration, and digital inequality (Kundu & Bej, 2025; Crompton et al., 2022).  Cambridge’s recent evidence brief similarly argues that AI can enhance motivation and feedback when embedded in well‑designed learning sequences, yet highlights the need for principled tool selection and ethical safeguards (Macinska & Vinkler, 2024). 

From CALL to AI‑mediated language education 

Compared with rule‑based CALL, contemporary AI affords dynamic learner modeling (e.g., NLP‑driven feedback, ASR‑based pronunciation analytics) and scalable formative assessment (Son et al., 2023).  Overviews in CALL and language‑education journals map this shift to AI‑supported tutoring, automated writing evaluation, chatbots, and data‑driven learning pipelines (Son et al., 2023; Crompton et al., 2022).  Systematic reviews focused on ELT confirm a steady rise in AI studies post‑2016, with higher education remaining the most researched sector and writing/speaking the most targeted skills (Crompton et al., 2022; Sharadgah & Sa’di, 2022).  At the same time, reviews caution that the promise of personalization can be undermined by under‑representation of low‑resourced languages and by infrastructural inequities—concerns that are salient in parts of the Arab region (Albedah, 2025). 

Generative AI and conversational agents (e.g., ChatGPT) 

Since late 2022, LLM‑based tools have rapidly entered ELT classrooms for brainstorming, scaffolded drafting, and low‑stakes oral rehearsal. Systematic reviews of EFL writing and broader ELT applications report improvements in fluency, feedback timeliness, and learner autonomy, alongside risks of over‑reliance, hallucinations, and academic integrity issues (Teng, 2024; Al‑Khresheh, 2024).  A comprehensive review of ChatGPT in ESL/EFL education synthesizing 70 empirical studies further documents gains in learning opportunities and personalization but flags persistent gaps in rigorous designs and coverage of skills beyond writing (Lo et al., 2024).  An Annual Review of Applied Linguistics article characterizes research on ChatGPT as nascent but fast‑maturing, urging attention to methodological robustness and ethics (Fang & Han, 2025).  

Automated Writing Evaluation (AWE) and AI‑mediated feedback 

Multi‑level meta‑analysis shows that AI‑based automated feedback yields medium positive effects on writing performance, though heterogeneity is substantial and design quality matters (Fleckenstein et al., 2023).  Recent quasi‑experimental work indicates that combining automated and teacher feedback can elevate EFL learners’ self‑efficacy and performance relative to teacher‑only feedback, while not uniformly reducing anxiety—implications for blended feedback design (Sari & Han, 2024).  Studies with Arab EFL learners report positive impacts of AWE on peer/self‑editing behaviors and favorable learner attitudes, yet also note limitations in higher‑order feedback (e.g., coherence) without teacher mediation (Al‑Inbari & Al‑Wasy, 2022/2023).  Contemporary reviews emphasize differential effectiveness by proficiency level and the continued need for human orchestration of revision cycles (Aldosemani et al., 2024/2025). 

ASR‑enabled pronunciation and speaking practice 

Meta‑analytic evidence across 2008–2021 studies shows a medium overall effect (g≈0.69) of ASR on ESL/EFL pronunciation, with larger gains for segmental features, explicit feedback, and peer‑mediated practice over longer interventions (Ngo et al., 2024).  A systematic review focused on EFL contexts corroborates effectiveness while highlighting design tendencies (quasi‑experimental, in‑class, segmental focus) and the need for research on suprasegmentals and classroom integration (Liu et al., 2025).  Mixed‑methods trials also report improvements in comprehensibility and speaking performance when ASR is paired with structured peer correction (Sun, 2023). 

Intelligent Tutoring Systems (ITS), adaptivity, and mobile AI 

Evidence syntheses across language learning position ITS as most effective when aligned with communicative/task‑based pedagogies and when combining NLP, ASR, and automated feedback for adaptivity (Adhami & Moore, 2025).  The AI100 (Stanford) report documents mainstreaming of AI‑based language training (e.g., Duolingo) and the shift from lab prototypes to real‑world tutoring at scale (AI100, 2016).  In parallel, mobile‑assisted learning reviews and meta‑analyses show sustained advantages for vocabulary growth—especially with longer treatments and gamified, feedback‑rich designs—underscoring mobile+AI synergies relevant to bandwidth‑variable contexts (Okumuş Dağdeler, 2023; Zhou & Zhou, 2026).  Systematic reviews on self‑directed mobile learning emphasize links to sustainable learning and lifelong learning competencies, provided digital abilities are scaffolded (Roy & Gandhimathi, 2024/2025).  

AI‑assisted assessment: validity, washback, and safeguards 

Operational AI scoring for writing and speaking (e.g., ETS’s e‑rater and SpeechRater) has multi‑decade research bases; studies show human–machine agreement comparable to inter‑rater reliability when systems are used with human oversight (ETS, 2018; ETS, n.d.).  Automated scoring can influence preparation behaviors (washback), sometimes narrowing practice to machine‑salient features; mixed‑methods work on TOEFL SpeechRater highlights both efficiency benefits and risks to communicative practice when misused (Gong, 2023).  Independent predictive‑validity studies of the Duolingo English Test (DET), which employs human‑in‑the‑loop AI, show positive associations with postgraduate attainment and provide transparency on scoring reliability—useful for placement and admissions decisions (Isaacs et al., 2023; Duolingo, 2020). 

Teacher roles, competencies, and professional learning 

AI repositions teachers from sole feedback providers to designers of AI‑supported learning experiences and critical AI literacy mentors. Systematic reviews and surveys in the region (e.g., Saudi Arabia, Jordan) identify urgent needs for AI competencies (data literacy, ethics, prompt design, tool orchestration) and caution against perceived teacher “de‑centering” without institutional support (Laoha et al., 2025; Metwally & Bin‑Hady, 2025).  Qualitative work with Jordanian undergraduates underscores that students still prize the irreplaceable human elements of empathy, cultural knowledge, and relational pedagogy—arguing for a balanced, human‑centered integration (Almashour et al., 2025). 

Ethics and governance: aligning classroom innovation with policy 

Global guidance now offers a clear policy frame: UNESCO’s Guidance for Generative AI in Education and Research recommends human‑centered, age‑appropriate use, data‑privacy protections, and teacher training; it also flags the lack of institutional policies worldwide (UNESCO, 2023).  OECD resources on trustworthy AI echo requirements for transparency, accountability, and risk management, linking educational AI to broader AI principles and literacy frameworks (OECD, 2023–2025).  Regional initiatives (e.g., UNESCO Arab States seminars on AI competencies) point to growing capacity‑building momentum across ministries and institutions (UNESCO, 2024; RCEP/UNESCO, 2022). 

The Arab World—and Libya in Particular 

Regional patterns. Research across Arab EFL contexts documents positive educator/learner dispositions toward AI tempered by concerns about training, access, and ethics; targeted professional development is repeatedly recommended (Altamimi, 2025; Jamal et al., 2025).  In Saudi Arabia and Yemen, large‑scale surveys of EFL educators call for structured PD aligned with UNESCO’s competency frameworks, emphasizing privacy, bias awareness, and pedagogical integration (Metwally & Bin‑Hady, 2025).  In Morocco, qualitative studies on ChatGPT in ELT similarly surface both engagement gains and integrity concerns—mirroring global findings (Bekou et al., 2024, as cited in Al‑Zahrani, 2025). 

Libya’s enabling environment. Despite political volatility, Libya’s recent connectivity indicators are strong at a macro level: DataReportal estimates ~88% internet penetration (January 2024) and high social‑media reach, while ITU’s 2024 “Facts & Figures” and reported ICT Development Index show notable advances in universal and effective connectivity (DataReportal, 2024; ITU, 2024; The Libya Observer, 2024).  Freedom House, however, documents uneven reliability (e.g., power‑related disruptions), rights concerns, and episodic shutdowns, warning that access and civic freedoms do not always move in tandem—an important caveat for edtech planning (Freedom House, 2024).  Oxford Business Group’s sector report highlights local digitalization initiatives (e.g., in Misrata) and a youthful demographic conducive to ICT uptake (Oxford Business Group, 2024). 

Libya’s AI in education: emerging policies and studies. In July 2024, ICESCO and Libya’s Ministry of Education launched an initiative to integrate AI into the educational process, including foresight training and system design—signaling policy intent and capacity building (ICESCO & Libya MoE, 2024).  In October 2025, the General Information Authority announced a national AI strategy (2025–2030) foregrounding training, ethics, and sectoral pilots (including education) (Libya GIA, 2025).  White‑paper proposals from national bodies further argue for unified AI governance to bridge regional disparities (NESDB Libya, 2026).  On the ground, Libyan EFL studies report generally positive teacher attitudes toward AI with calls for sustained support and training; instructors and students identify benefits for vocabulary learning, engagement, and content creation, while noting constraints in familiarity, quality control, time, and privacy (Almashrgy & Alburki, 2024; Alkurtehe & Rathakrishnan, 2025; Elkali & Abad, 2025).  Broader analyses of digital education and higher‑education infrastructure confirm momentum alongside persistent capacity gaps—especially in staff development and secure digital infrastructure (Yahya et al., 2025; Sultan & Elturki, 2023).  

Pedagogical Implications for AI‑Mediated EFL/ELT in Libya 

Adopt a “human‑in‑the‑loop” design for feedback and assessment. Blend AWE/ASR with teacher mediation to preserve higher‑order writing feedback, discourse‑level speaking goals, and academic integrity; leverage AI to free teacher time for coaching and dialogic feedback (Fleckenstein et al., 2023; Sari & Han, 2024; Gong, 2023). 

Prioritize teacher AI competencies and ethics. Align PD with UNESCO/OECD guidance—data privacy, bias awareness, prompt and task design, and critical AI literacy for students—integrating local policy initiatives (ICESCO/Libya MoE; GIA AI strategy) (UNESCO, 2023; OECD, 2023–2025; ICESCO & Libya MoE, 2024; Libya GIA, 2025).  

Exploit mobile+AI affordances for access and continuity. Given mobile penetration and intermittent power/connectivity, use mobile‑first, offline‑capable AI practice (especially for vocabulary and pronunciation) with clear learning analytics and teacher dashboards (Okumuş Dağdeler, 2023; Zhou & Zhou, 2026; Ngo et al., 2024).   

Strengthen validity, transparency, and washback controls. Where automated scoring is used (e.g., placement), ensure human oversight, publish rubrics, and monitor washback to avoid narrowing of communicative practice (ETS, n.d.; Chen et al., 2018; Gong, 2023). [ets.org],  

Localize content and attend to equity. Encourage Arabic–English bilingual prompts, culturally relevant materials, and support for under‑resourced varieties; plan for differential access across regions in line with ITU/UNESCO recommendations (Albedah, 2025; ITU, 2024).  

Conclusion 

AI can meaningfully augment EFL/ELT when its strengths (scalable feedback, adaptivity, dialogic practice) are channeled through principled pedagogy and robust governance—and when contextual constraints are taken seriously. In Libya, a rare combination of improving macro‑connectivity indicators, nascent national AI initiatives, and a motivated EFL community creates a window for responsible, equity‑minded innovation—provided that teacher capacity, ethics, and infrastructure reliability remain core priorities (The Libya Observer, 2024; ICESCO & Libya MoE, 2024; Almashrgy & Alburki, 2024). [libyaobserver.ly], [icesco.org], [mitec.unikl.edu.my] 

References  

Al‑Zahrani, A. A. (2025). Exploring the role of ChatGPT in English teaching and learning: Educators’ and language learners’ perspectives. EDUPIJ, 17, e2025330. https://doi.org/10.22521/edupij.2025.17.330 [edupij.com] 

Albedah, F. (2025). Artificial intelligence in language education: A systematic review of multilingual applications, large language models, and emerging challenges. Language Teaching Research Quarterly, 49, 247–268. https://doi.org/10.32038/ltrq.2025.49.13 [files.eric.ed.gov] 

Aldosemani, T., Lhothali, A., & Albsisi, M. (2024/2025). Automated Writing Evaluation in EFL contexts: A review of effectiveness, impact, and pedagogical implications. International Journal of Computer‑Assisted Language Learning and Teaching, 13(1). https://doi.org/10.4018/IJCALLT.329962 [pdfs.seman…cholar.org] 

Altamimi, D. H. F. (2025). AI integrating in EFL classrooms: A study on lecturers’ adaptation and student outcomes. Arab World English Journal, Special Issue on CALL(11), 364–379. https://dx.doi.org/10.24093/awej/call11.22 [awej.org] 

Crompton, H., Edmett, A., & Ichaporia, N. (2022). Artificial intelligence and English language teaching: A systematic literature review. British Council. https://www.britishcouncil.org/sites/default/files/ai_in_english_language_teaching_systematic_review.pdf [britishcouncil.org] 

Duolingo English Test. (2020). Analysis of the scoring and reliability for the Duolingo English Test. https://englishtest.duolingo.com/scores (see technical PDF) [englishtes…uolingo.cn] 

Fang, S., & Han, Z. (2025). On the nascency of ChatGPT in foreign language teaching and learning. Annual Review of Applied Linguistics, 45, 148–178. https://doi.org/10.1017/S026719052510010X [cambridge.org] 

Fleckenstein, J., Liebenow, L. W., & Meyer, J. (2023). Automated feedback and writing: A multi‑level meta‑analysis. Frontiers in Artificial Intelligence, 6, 1162454. https://doi.org/10.3389/frai.2023.1162454 [frontiersin.org] 

Gong, K. (2023). Washback of automated speech scoring (SpeechRater) in TOEFL iBT. Asian‑Pacific Journal of Second and Foreign Language Education, 8, 25. https://doi.org/10.1186/s40862-023-00197-2 [link.springer.com] 

Isaacs, T., Hu, R., Trenkic, D., & Varga, J. (2023). Examining the predictive validity of the Duolingo English Test. Language Testing, 40(3), 748–770. https://doi.org/10.1177/02655322231158550 [journals.sagepub.com] 

Kundu, A., & Bej, T. (2025). Transforming EFL teaching with AI: A systematic review of empirical studies. International Journal of Artificial Intelligence in Education, 35, 2281–2314. https://doi.org/10.1007/s40593-025-00470-0 [link.springer.com] 

Laoha, R., Chomthong, W., & Pongpanich, W. (2025). AI and EFL teachers’ competencies: A systematic review. Higher Education Studies, 15(3), 262–275. https://doi.org/10.5539/hes.v15n3p262 [files.eric.ed.gov] 

Liu, Y., Ab Rahman, F., & Mohamad Zain, F. (2025). ASR in EFL pronunciation: A systematic review. Cogent Education, 12(1), 2466288. https://doi.org/10.1080/2331186X.2025.2466288 [tandfonline.com] 

Lo, C. K., Yu, P. L. H., Xu, S., Ng, D. T. K., & Jong, M. S.‑Y. (2024). ChatGPT in ESL/EFL education: A systematic review of empirical studies. Smart Learning Environments. https://doi.org/10.1186/s40561-024-00342-5 [link.springer.com] 

Macinska, S., & Vinkler, A. (2024). The role of AI in language learning: Research evidence and strategies. Cambridge University Press & Assessment. https://www.cambridge.org/…/The_Role_of_AI_in_Language_Learning_Research_Evidence_and_Strategies.pdf [cambridge.org] 

Metwally, A. A., & Bin‑Hady, W. R. A. (2025). AI integration training for EFL educators in Saudi Arabia. Cogent Education, 12(1), 2472462. https://doi.org/10.1080/2331186X.2025.2472462 [tandfonline.com] 

Ngo, T. T.‑N., Chen, H. H.‑J., & Lai, K. K.‑W. (2024). The effectiveness of ASR in ESL/EFL pronunciation: A meta‑analysis. ReCALL, 36(1), 4–21. https://doi.org/10.1017/S0958344023000113 [cambridge.org] 

Okumuş Dağdeler, K. (2023). Mobile‑assisted vocabulary learning: A systematic review. Smart Learning Environments, 10, 19. https://doi.org/10.1186/s40561-023-00235-z [link.springer.com] 

OECD. (2023–2025). Trustworthy artificial intelligence (AI) in education; OECD.AI Observatory; AI literacy framework (review draft). https://oecd.ai/ ; https://ailiteracyframework.org [oecd.org], [oecd.ai], [ailiteracy…mework.org] 

Sharadgah, T. A., & Sa’di, R. A. (2022). AI in ELT: A systematic review (2015–2021). Journal of Information Technology Education: Research, 21, 337–377. https://doi.org/10.28945/4999 [A Systemat…igence …] 

Son, J.‑B., Ružić, N. K., & Philpott, A. (2023). AI technologies and applications for language learning and teaching. Journal of China CALL, 5(1). https://doi.org/10.1515/jccall-2023-0015 [degruyterbrill.com] 

Sun, W. (2023). ASR, peer correction, and L2 pronunciation/speaking. Frontiers in Psychology, 14, 1210187. https://doi.org/10.3389/fpsyg.2023.1210187 [frontiersin.org] 

Teng, M. F. (2024). ChatGPT for EFL writing: Opportunities and challenges. International Journal of TESOL Studies, 6(3), 36–57. https://doi.org/10.58304/ijts.20240304 [tesolunion.org] 

UNESCO. (2021). AI and education: Guidance for policy‑makers. https://unesdoc.unesco.org/ark:/48223/pf0000376709 [unesdoc.unesco.org] 

UNESCO. (2023). Guidance for generative AI in education and research. https://unesco.org/…/guidance_for_generative_ai_in_education_and_research.pdf [unesco.org.uk] 

UNESCO (Arab States). (2024). UNESCO supports Arab States in developing AI competencies for students and teachers. https://www.unesco.org/en/articles/unesco-supports-arab-states-developing-ai-competencies-students-and-teachers [unesco.org] 

Wang, Y., & Sun, Z. (2018). Human vs. automated speaking scores (SpeechRater). Psychological Test and Assessment Modeling, 60(1), 81–100. (Preprint). [researchgate.net] 

Libya‑focused and regional sources 

Aboseta, A. K. A., Alhaj Salem, A. A. M., & Alnagrat, A. J. A. (2025). Harnessing AI for education reform in Libya: Opportunities and challenges. PERFECT: Journal of Smart Algorithms. https://media.neliti.com/…/642350-…pdf [media.neliti.com] 

Alkurtehe, K. A. M., & Rathakrishnan, M. (2025). AI for Libyan EFL learners’ vocabulary: Practices, perspectives, challenges. Asian Journal of Language, Literature and Culture Studies. https://hal.science/hal-05247625v1 [hal.science] 

Almashrgy, A. A., & Alburki, H. F. (2024). Libyan teachers’ perceptions of AI in ELL. Malaysian Journal of Industrial Technology, 8(4). https://mitec.unikl.edu.my/…/TEACHERS-PERCEPTIONS-…LIBYAN-EFL-CLASSROOMS.pdf [mitec.unikl.edu.my] 

Elkali, H. A. A., & Abad, A. R. (2025). Benefits and challenges of integrating AI in Libyan EFL higher education: Instructors’ perspectives. IJHSS, Special Issue (Mar.). https://doi.org/10.61856/wp094e23 [ijhss.eventsgate.org] 

ICESCO & Libya MoE. (2024). Initiative to integrate AI into the educational process in Libya. https://icesco.org/en/2024/07/17/launch-of-new-initiative… [icesco.org] 

Libya General Information Authority. (2025). National Strategy for Artificial Intelligence 2025–2030 (public release notice). https://libyaherald.com/…/national-strategy-for-artificial-intelligence-2025-2030…/ [libyaherald.com] 

National Economic & Social Development Board (Libya). (2026). A unified AI governance framework to bridge the digital divide. https://nesdb.ly/…/Whitepaper-Libya-AI-Framework-1.pdf [nesdb.ly] 

DataReportal. (2024). Digital 2024: Libya. https://datareportal.com/reports/digital-2024-libya [datareportal.com] 

Freedom House. (2024). Freedom on the Net: Libya. https://freedomhouse.org/country/libya/freedom-net/2024 [freedomhouse.org] 

International Telecommunication Union. (2024). Measuring digital development: Facts and figures 2024. https://www.itu.int/itu-d/reports/statistics/facts-figures-2024/ [itu.int] 

The Libya Observer. (2024, July 16). Libya ranks 1st in 2024 ICT in Africa. https://libyaobserver.ly/tech/libya-ranks-1st-2024-ict-africa [libyaobserver.ly] 

Oxford Business Group. (2024). ICT and innovation: The Report—Misrata 2024. https://www.oxfordbusinessgroup.com/reports/libya/2024-report/ict-innovation-chapter/ [oxfordbusi…sgroup.com] 

Yahya, N., Abdullah, M., & Masuwd, M. (2025). Digital education in Libya: Progress, challenges, and future directions. International Journal of Education and Digital Learning, 3(5), 211–219. https://doi.org/10.47353/ijedl.v3i5.307 [ij.lafadzp…ishing.com] 

Sultan, A., & Elturki, K. (2023). Digital infrastructure standards for Libyan state universities. (Working paper). https://www.researchgate.net/…/The-Digital-Infrastructure-Standard…Libyan-State-Universities.pdf