Integration of Artificial Intelligence Technologies into the Educational Process: Uzbekistan's Experience and Future Prospects

Uteulieva Aynura Audanbaevna1*, Otepbergenov Jetkerbay Sakbergenovich2

^1&2 Faculty of Nukus State Pedagogical Institute, Uzbekistan

^*Corresponding author’s e-mail: aynurauteulieva@gmail.com

Received: 29 Jul 2026

Revised: 15 Oct 2026

Accepted: 23 Oct 2026

DOI: doi.org/10.60072/ijeissah.2026.v4i04.002

ABSTRACT

This article reports a mixed-methods field study on how AI-based educational technologies affect learning in Uzbekistan. It was carried out in 15 schools and 3 higher education institutions across the Tashkent, Samarkand, and Fergana regions during the 2023–2024 academic year. An intervention group was compared with a comparison group over 16 weeks (920 students in total), and interviews were held with teachers and students. We looked at knowledge acquisition, learning motivation (ARCS), task-completion time, teachers’ administrative workload, and knowledge retention 12 weeks later. On the descriptive means, the intervention group showed about 27.5% higher knowledge acquisition than the comparison group, along with stronger motivation, better retention, and faster task completion. Teachers’ administrative workload was 34.6% lower, which reflects reduced workload rather than a general gain in teaching effectiveness. These differences still need to be confirmed with inferential analyses that account for clustering and baseline values. The article also considers the policy setting for AI in Uzbek education, keeping policy goals separate from the empirical findings. The results should be read as associations observed under the study conditions, not as proof of a universal causal effect.

Keywords: Artificial Intelligence; Adaptive Learning; ARCS Motivation; Digital Pedagogy; Learning Effectiveness; Teacher Workload; Uzbekistan

This is an open access article under the CC BY NC ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/?ref=chooser-v1); © 2026.

Published by Asia Pacific Higher Learning Sdn Bhd., Malaysia

Background

Over the last decade, the integration of artificial intelligence (AI) technologies into education has been bringing about fundamental changes in the global pedagogical paradigm. The traditional teaching model rests mainly on giving all students the same educational materials, whereas AI technologies make it possible to build individualized learning trajectories, analyze students' cognitive characteristics, and adapt educational content in real time. This approach helps move the educational process toward a student-centered model. Within the education system, AI is becoming not only a technological tool but also an intelligent mechanism that supports pedagogical decision-making. International experience shows that AI-based adaptive learning systems, intelligent tutors, and automated assessment tools make it possible to identify individual student needs, give real-time feedback, and reduce the administrative burden on teachers. In the USA, South Korea, Singapore, and China, for example, adaptive learning platforms are in wide use, and effective results have been recorded in improving students' subject mastery. More specifically, AI-based intelligent systems analyze students' previous activity, response statistics, and learning pace and form personalized recommendations from them. Students' rate of knowledge acquisition and their independent work skills improve significantly as a result.

According to research by Holmes et al., artificial intelligence technologies enable increased student engagement, knowledge reinforcement, and the prediction of learning outcomes. Furthermore, research conducted by Luckin has shown that AI-based systems enhance educational effectiveness by automatically identifying student errors and developing tailored recommendations for them. Hwang and Xie's research also noted that the adaptive management of the educational process using AI technologies not only improves students' academic performance but also has a positive impact on their learning motivation.

A 2023 UNESCO report states that between 2022 and 2023, AI-powered educational solutions were tested or integrated into national programs in 94 countries. These technologies have the potential to raise learning outcomes by 18–27% on average. Reports from the OECD and the World Bank have likewise stressed the economic and pedagogical effectiveness of artificial intelligence in a digital learning environment. Some analyses suggest that AI technologies can cut the time teachers spend on assessment and reporting by up to 30–40%, which in turn gives educators more opportunities to work individually with students and engage in creative activities.

At the same time, the development of generative artificial intelligence systems has opened a new stage in education. Platforms built on large language models such as ChatGPT, Gemini, and Copilot enable students to create educational materials, explain complex topics, analyze programming code, and support scientific research. Generative AI tools are regarded as essential for advancing scientific research, distance learning, and independent study, especially in higher education institutions. Yet the use of such systems also raises urgent questions of academic integrity, plagiarism, data privacy, and compliance with ethical standards.

Under Uzbekistan's digital transformation policy, integrating AI technologies into the education system has been made a priority. Presidential Resolution PQ-358 of the Republic of Uzbekistan, dated October 14, 2024, "On the Approval of the Strategy for the Development of Artificial Intelligence Technologies until 2030," requires the implementation of AI solutions in educational institutions, the creation of national AI content, and the improvement of the digital competencies of pedagogical staff. On the basis of this resolution, a series of organizational measures is under way to build a national artificial intelligence infrastructure, set up AI laboratories, and train specialists.

In addition, the Decree of the President of the Republic of Uzbekistan No. PQ-4884, dated November 6, 2020, "On Additional Measures to Further Improve the Education System," and Resolution No. PQ-3832, "On Measures for the Development of the Digital Economy and the Sphere of Crypto-Asset Turnover in the Republic of Uzbekistan," provide the legal basis for implementing modern educational technologies and for modernizing the educational infrastructure, respectively. Decree No. PF-6079, "On the Approval of the 'Digital Uzbekistan — 2030' Strategy and Measures for its Effective Implementation," set out the first formal steps in this direction. As a result, the institutional conditions needed for the wide introduction of e-learning platforms, distance learning systems, and digital educational resources were put in place.

Even so, the country's education system still faces several challenges. The effective integration of AI technologies is complicated by the uneven development of internet infrastructure across regions, a lack of technical resources in some educational institutions, gaps in educators' competencies with AI technology, and differing levels of digital literacy among students. International experience shows that simply having technological tools available does not automatically raise the quality of education; how effective they are depends closely on pedagogical methods, teacher preparedness, and the adaptability of the learning environment.

Objectives

Despite the growing international literature, the present study addresses a specific empirical gap by examining AI-supported learning in a multi-site Uzbek educational sample that includes both schools and higher education institutions. The study focuses on five outcome domains: knowledge acquisition, learning motivation, task-completion time, teacher administrative workload, and delayed knowledge retention. It also examines how implementation experiences and reported barriers may help explain differences in the quantitative findings. Accordingly, the study asks three research questions: (RQ1) What differences in learning and workload outcomes are observed between the intervention and comparison groups? (RQ2) How do the observed outcomes vary descriptively by institution type, region, and subject? (RQ3) How do teacher and student experiences, including implementation barriers, help explain the quantitative patterns?

Literature Review

The integration of artificial intelligence (AI) in education is supported by national strategic frameworks and digital transformation initiatives in Uzbekistan. Within this regulatory context, national policy instruments prioritize AI development, infrastructure expansion, and digital skill acquisition to modernize pedagogical practices (Ministry of Justice of the Republic of Uzbekistan, 2024). International research underscores that AI-driven interventions, particularly adaptive learning systems and intelligent tutoring platforms, can personalize instruction, optimize feedback mechanisms, and improve educational outcomes (Kabudi et al., 2021; Wang et al., 2024). Furthermore, meta-analytic evidence confirms significant positive effects of AI tools on student learning achievement, while highlighting the moderating influence of subject domains, intervention duration, and instructional contexts (Tlili et al., 2025). Concurrently, global policy frameworks emphasize that scaling educational AI requires human-centered practice, robust data governance, and comprehensive teacher competence frameworks (UNESCO, 2023; Miao & Cukurova, 2024). Although current literature documents the theoretical potential and quantitative efficacy of AI technologies, an empirical gap remains regarding multi-site implementation and localized pedagogical dynamics within Uzbekistan's educational institutions, necessitating rigorous empirical examination across secondary and higher education environments.

Methods

The The study used a mixed-methods field design with a quantitative comparison of an intervention group and a comparison group, followed by a qualitative explanatory phase. Because the supplied manuscript does not document the exact unit of allocation, the design is described here as quasi-experimental rather than as a fully randomised controlled trial. The final manuscript should state explicitly whether individual students, intact classes/academic groups, or another unit was assigned to each condition and should align the statistical model with that unit of allocation.

The research was conducted during the 2023–2024 academic year in 15 general education schools and 3 higher education institutions in the Tashkent, Samarkand, and Fergana regions. The manuscript reports that site selection considered urbanisation, internet infrastructure, digital-technology availability, and technical capacity. For reproducibility, the authors should add the exact site-selection criteria and identify the number of participating institutions, classes/groups, and students within each region and institution type.

Sample and Group Formation

A total of 920 students aged 14–21 and 78 educators participated. The manuscript describes a multi-stage sampling procedure in which educational institutions were selected first and classes or academic groups were then sampled. The current text also states that participants were randomly assigned to the two conditions, but it does not identify the unit or mechanism of randomisation. This wording must be replaced with the exact procedure documented in the study records:

The respondent-allocation table supplied with the study materials specifies 920 participants across 15 schools and 3 higher education institutions, with an overall allocation of n = 462 to the intervention group and n = 458 to the comparison group. The institution-type and regional subgroup sizes are reported in Table 3; subject-specific allocation remains to be verified from the original study records. Baseline comparability was examined using the procedures described in the original study. However, the actual baseline means, standard deviations, group-specific sample sizes, and test statistics are not included in the supplied manuscript. The final version should report these values for each outcome included in the primary model and should not rely only on a statement of p > 0.05.

The documented functions of the three platforms were as follows: Adaptive STEM adjusted the complexity of subsequent tasks in response to student responses and error patterns; the AI-Language Assistant analysed pronunciation, grammar, and semantic errors; and SmartFeedback provided automated assessment and individualized feedback. The authors should specify the adaptive rules or model family used by each platform and state whether the systems were developed by the research team or obtained from external providers.

Data Collection Tools

A multi-component data collection system was developed for the study. The following tools were used to assess learning outcomes and educational effectiveness:

1. •  standardized pre- and post-tests to measure knowledge levels (Cronbach's α = 0.89);

2. •  the ARCS (Attention, Relevance, Confidence, Satisfaction) motivation scale (α = 0.86);

3. •  a digital workload diary for teachers;

4. •  semi-structured interviews;

5. •  platform logs and activity data;

6. •  observation protocols.

Knowledge tests assessed conceptual understanding, practical skills, and problem-solving competencies. The manuscript states that the instruments underwent a three-stage validation process, but does not identify the three stages. The final version must name each stage and its procedure (for example, expert/content review, pilot testing, and psychometric analysis only if those procedures were actually conducted), identify the test wave associated with Cronbach's α = 0.89, and report the number of items and scoring range.

The ARCS motivation measure followed Keller's model of Attention, Relevance, Confidence, and Satisfaction [14]. Each component was measured on a 5-point Likert scale (α = 0.86). The final manuscript should report the number of items, scoring procedure, source/adaptation of the scale, language of administration, and whether the instrument was translated or culturally adapted for the study context. A digital gradebook system was developed for teachers, which recorded the time spent on lesson preparation, assessment processes, and administrative activities. This method allowed for determining the impact of AI technologies on the pedagogical workload. Semi-structured interviews were conducted with 32 teachers and 48 students. The protocol addressed experience with the AI platforms, usability, implementation challenges, and perceived educational effects. To strengthen the qualitative strand, the final manuscript should specify participant selection, interview duration, interview language, recording/transcription procedures, the main protocol domains, and the coding workflow.

The following metrics were automatically recorded through platform logs:

1. •  learning activity index;

2. •  task completion frequency;

3. •  response speed;

4. •  error patterns;

5. •  degree of individualization;

6. •  number of platform logins;

7. •  adaptation coefficient of recommendation algorithms.

This log data enabled the analysis of the adaptation dynamics of AI algorithms to students.

Analysis Methods

Quantitative data were processed using SPSS Statistics 28. Descriptive statistics, distribution diagnostics, and reliability estimates were calculated. The original analysis used ANCOVA with prior knowledge as a covariate, together with t-tests and Cohen's d. Because the study is multisite and the manuscript indicates sampling or assignment through classes/academic groups, the final analysis must first establish the unit of randomisation/allocation and then account for clustering, for example through a multilevel model or another design-appropriate approach. The final model should report the baseline covariate, adjusted group difference, model-based effect size, exact p-value, and 95% confidence interval. The 78 teachers must be analysed at teacher level; student-based degrees of freedom must not be used for the teacher-workload outcome.

Cohen's d was interpreted using the conventional benchmarks of approximately 0.20 (small), 0.50 (medium), and 0.80 (large), with the original source cited as Cohen. These thresholds are heuristic rather than absolute and should not replace interpretation of the magnitude and uncertainty of the study-specific estimates. Qualitative data were coded in NVivo 14 using inductive thematic analysis. The reported coding sequence included open coding, categorisation, and theme identification, with an inter-coder reliability statistic of Cohen's κ = 0.84. The revised manuscript should report how many coders independently coded the transcripts, how disagreements were resolved, the codebook-development process, and the frequency or prevalence of each final theme. These details are necessary for the qualitative findings to function as an explanatory component rather than as illustrative quotations only.

Ethics and participant protection: For school participants aged 14–17, report how parental/guardian consent and student assent or consent were obtained. Specify how platform log data were de-identified, where they were stored, who had access, the retention period, and the safeguards used for privacy and confidentiality.

Baseline and subgroup reporting required for the final submission: report pre-intervention means and standard deviations for the outcomes entered into the primary model for both groups; report group sizes and baseline comparisons; and provide descriptive outcome summaries separately for school versus university participants, for the Tashkent, Samarkand, and Fergana regions, and for mathematics, physics, chemistry, and foreign language.

Results and Discussion

The reported descriptive results show higher knowledge, motivation, and retention values and lower task-completion time and teacher administrative workload in the intervention group. The originally reported F-statistics and p-values are not reproduced here because the supplied manuscript does not contain the raw observations needed to verify the inferential analysis and the reviewer identified internal inconsistencies between the reported means, standard deviations, F-values, degrees of freedom, and Cohen's d. Final significance testing should therefore be completed after the clustering and baseline structure are verified.

Table 1. Descriptive Comparison of Intervention and Comparison Groups

Indicator	Comparison group (M±SD)	Intervention group (M±SD)	Descriptive change	d*
Knowledge Acquisition Score (0–100)	67.3±10.1	85.8±8.4	+18.5 points; ≈ +27.5%	1.99
ARCS Motivation Index (1–5)	3.31±0.58	4.24±0.49	+0.93; ≈ +28.1%	1.73
Task Completion Time (min.)	25.4±3.9	17.1±2.8	−8.3 min.; ≈ −32.7%	2.44
Teacher Administrative Load (hours/week)	15.6±2.7	10.2±2.0	−5.4 h; ≈ −34.6%	N/A†
12-Week Knowledge Retention (%)	61.2±7.3	78.9±6.1	+17.7 percentage points; ≈ +28.9%	2.63

The manuscript reports that platform logs were used to examine error patterns, individualized exercises, additional explanations, and adaptation dynamics. It also reports an 81% rate for cases receiving individualized exercises or explanations and a 38% reduction in selected teacher-review activities at School No. 28 in Samarkand. These figures should be retained only if the underlying log extract and calculation rule are available. The final Results section should define the denominator, time window, and calculation procedure for each percentage. The interviews produced three reported themes:

Unadjusted d values are computed from the reported group means and standard deviations using the pooled-SD formula and, for student outcomes, n = 462 and n = 458. These are descriptive standardized mean differences, not ANCOVA- or multilevel-model-adjusted effects. † Teacher workload cannot be recalculated as d from the supplied manuscript because the number of teachers in each condition is not reported. The final table must replace these descriptive calculations with raw-data-based effect sizes, exact p-values, adjusted estimates, and 95% confidence intervals after the unit of analysis and clustering are established.

Table 5. Descriptive Outcome Summary Used in the Present Manuscript

Outcome	Comparison group (M ± SD)	Intervention group (M ± SD)	Descriptive change	Unadjusted d
Knowledge acquisition score (0–100)	67.3 ± 10.1	85.8 ± 8.4	+18.5 points; ≈ +27.5%	1.99*
ARCS motivation index (1–5)	3.31 ± 0.58	4.24 ± 0.49	+0.93; ≈ +28.1%	1.73*
Task-completion time (min.)	25.4 ± 3.9	17.1 ± 2.8	−8.3 min.; ≈ −32.7%	2.44*
Teacher administrative load (hours/week)	15.6 ± 2.7	10.2 ± 2.0	−5.4 h; ≈ −34.6%	N/A†
12-week knowledge retention (%)	61.2 ± 7.3	78.9 ± 6.1	+17.7 percentage points; ≈ +28.9%	2.63*

Note. * These d values are unadjusted descriptive standardized mean differences based on the reported aggregate means and SDs; they are not adjusted for baseline or clustering. † Teacher workload cannot be standardized as d without teacher-group sample sizes. Final inferential estimates, p-values, and 95% CIs should be recalculated from the raw data after the allocation unit and clustering structure are verified.

1. Individualized approach: Teachers described greater visibility of individual weaknesses and more immediate adaptation of learning tasks. The quotation may be retained only with the participant's consent and the journal's anonymisation requirements.

2. Technical and pedagogical barriers: Interviewees reported internet instability, device shortages, and difficulties integrating AI-generated information into lesson planning.

3. Transformation of the teacher's role: Participants described changes in the teacher's role toward facilitation, motivation, and interpretation of learner data.

The manuscript reports that knowledge retention remained 28.9% higher in the intervention group at the 12-week retention assessment. The exact timing of that assessment relative to the end of the 16-week intervention is not specified in the supplied text and must be added. The previous causal statement about “deep learning and repetition algorithms” has been removed because the available data do not isolate algorithmic mechanisms as the causal explanation.

The observed descriptive differences are compatible with the broader literature showing that AI-enabled adaptive learning can have positive educational effects, but the magnitude of the present study's summary-statistic effects should be interpreted cautiously. Zawacki-Richter et al. (2019) mapped the research landscape rather than providing a universal 15–30% causal effect. More directly, Wang et al. (2024) reported an overall Hedges' g = 0.70 for AI-enabled adaptive learning and found that effect sizes varied with duration, discipline, educational level, and research design. The much larger unadjusted d values derived from the present manuscript's reported means and SDs therefore require careful discussion and raw-data verification before they can be compared directly with meta-analytic estimates.

A recent meta-analysis by Tlili et al. (2025), based on 85 quantitative studies, reported a large overall effect of AIEd on learning achievement (g = 1.10) but also found moderation by educational field, educational level, learning mode, intervention duration, and geographic distribution. Wang et al. (2024) similarly found meaningful moderators in AI-enabled adaptive learning research. These findings suggest several plausible reasons why the present study may show unusually large summary-statistic differences: differences in time on task, alignment between intervention and assessment, novelty effects, the use of multiple AI functions in the intervention condition, or residual methodological differences between groups. These possibilities should be treated as limitations and hypotheses for further testing, not as explanations established by the current data.

In Uzbekistan, the policy context supports continued digital and AI development, but the empirical study does not itself evaluate compliance with national implementation targets. The discussion therefore focuses on infrastructure, teacher preparedness, data governance, and curriculum alignment. The barriers reported by participants are consistent with broader evidence that digital transformation into less-resourced settings can be shaped by connectivity, device access, and stakeholder capacity.

Limitations. First, the sample is geographically limited to three regions and includes institutions selected under a multi-stage procedure whose exact selection criteria should be documented. Second, the manuscript currently does not establish the unit of allocation or account for the nested structure of classes, institutions, and regions; this limits the interpretation of the reported inferential statistics. Third, the absence of the raw dataset from the supplied materials prevents independent verification of the F-values, p-values, confidence intervals, and final effect sizes. Fourth, the study reports teacher and student interviews but does not yet provide them with prevalence or sufficient methodological detail for full qualitative auditability. Fifth, the duration of the intervention and the 12-week retention assessment do not establish effects beyond the study window. Finally, algorithmic fairness, privacy, and human oversight should be treated as explicit dimensions of future investigation, consistent with recent AI-in-education ethics research.

Practical Recommendations:

1. Consider establishing a national AI-in-education research and validation hub to coordinate curriculum-aligned pilots, Uzbek-language educational resources, teacher training, and independent evaluation. This recommendation follows from the study's emphasis on curriculum alignment and the need to document intervention fidelity; it is a proposal, not an action mandated by PQ-358.

2. Strengthen teacher professional development in AI pedagogy, data literacy, AI ethics, and human oversight. This recommendation is linked to the reported technical and pedagogical barriers and is also consistent with UNESCO's AI Competency Framework for Teachers.

3. Prioritise infrastructure and access measures where instability of internet connectivity and device shortages were reported. Possible responses include institution-level connectivity plans, device-sharing protocols, and offline-capable learning resources. The recommendation is directly linked to the barrier theme reported in the interviews.

4. Develop transparent procedures for evaluating and reporting AI educational solutions, including evidence of curriculum alignment, privacy safeguards, teacher oversight, and measurable learning outcomes. This follows from the study's documentation gaps and from the literature's emphasis on responsible and evidence-based AI adoption

Implications

The findings demonstrate that successful AI integration in education extends beyond technology adoption, requiring a holistic alignment with pedagogy, infrastructure, educator competence, and institutional capacity. While AI tools effectively reduce administrative workload and support personalized learning, they serve to enhance rather than replace human instruction. Consequently, educational policymakers and academic leaders must design comprehensive ecosystems that ensure data governance, privacy, and ethical standards. Prioritizing targeted professional development and teacher-AI collaboration will allow institutions to maximize instructional efficiency while fostering students' critical thinking, creativity, and academic independence within increasingly digital learning environments.

Conclusion

This study highlights the potential of AI-driven interventions incorporating adaptive recommendations, automated feedback, and individualized support. Descriptive results indicate meaningful improvements in the intervention group over the comparison group across key metrics: knowledge acquisition (+27.5%), ARCS motivation (+28.1%), and delayed retention (+28.9%), alongside reductions in task-completion time (−32.7%) and teacher administrative workload (−34.6%).However, until raw-data verification, baseline adjustments, and clustered inferential statistics are finalized, these findings should be interpreted strictly as descriptive group differences. Furthermore, claims regarding specific ARCS subcomponents (e.g., Attention and Satisfaction) and detailed workload reductions require dedicated supporting data in the results. While these multi-site findings in Uzbekistan contribute valuable empirical insights to international literature, scaling AI integration extends beyond technology adoption—it demands a holistic approach encompassing pedagogy, infrastructure, educator competence, data governance, and institutional capacity. Future research should employ longitudinal designs, expanded rural sampling, and multilevel statistical modeling. Priority should also be given to examining algorithmic fairness, data privacy, teacher-AI collaboration, and the broader impact of generative AI on students' critical thinking and academic independence.

Declarations

Ethics Approval and Consent to Participate: All participants gave informed consent before taking part in the study. Their participation was voluntary, and their responses remained confidential.

AI Assistance: The authors reviewed and validated all scientific content. The authors declared that generative AI tools were used only for grammar and language editing.

Conflicts of Interest: No competing interests.

Funding: No funding.

Authors' contributions: Uteulieva Aynura Audanbaevna designed the study, collected and analysed the data, and drafted the manuscript. Otepbergenov Jetkerbay Sakbergenovich supervised the research and revised the manuscript. Both approved the final version.

Acknowledgement: The researchers gratefully acknowledge all participants for their valuable involvement and contributions to this study.

References

Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum Associates. https://doi.org/10.4324/9780203771587

Holmes, W., Bialik, M., & Fadel, C. (2019). Artificial Intelligence in Education: Promises and Implications for Teaching and Learning. Boston: Center for Curriculum Redesign.

Hwang, G. J., & Chang, C. Y. (2023). A review of opportunities and challenges of chatbots in education. Interactive Learning Environments, 31(7), 4099-4112. https://doi.org/10.1080/10494820.2021.1952615

Kabudi, T., Pappas, I., & Olsen, D. H. (2021). AI-enabled adaptive learning systems: A systematic mapping of the literature. Computers and Education: Artificial Intelligence, 2, 100017. https://doi.org/10.1016/j.caeai.2021.100017

Keller, J. M. (2010). Motivational Design for Learning and Performance: The ARCS Model Approach. Springer. https://doi.org/10.1007/978-1-4419-1250-3

Kukulska-Hulme, A., Popenici, S. A. D., & Koivunen, P. (2021). Ethics and artificial intelligence in education: A review of the literature. British Journal of Educational Technology, 52(4), 1389–1405. https://doi.org/10.1111/bjet.13089

Luckin, R., Holmes, W., Griffiths, M., & Forcier, L. B. (2016). Intelligence Unleashed: An argument for AI in Education. London: Pearson.

Miao, F., & Cukurova, M. (2024). AI competency framework for teachers. UNESCO. https://doi.org/10.54675/ZJTE2084

Ministry of Justice of the Republic of Uzbekistan. (2024). National database of legislation of the Republic of Uzbekistan (LexUZ). LexUZ. https://lex.uz/

Mouta, A., Pinto-Llorente, A. M., & Torrecilla-Sánchez, E. M. (2024). Uncovering blind spots in education ethics: Insights from a systematic literature review on artificial intelligence in education. International Journal of Artificial Intelligence in Education, 34, 1166–1205. https://doi.org/10.1007/s40593-023-00384-9

Oʻzbekiston Respublikasi Prezidentining 14.10.2024 yildagi PQ-358-son qarori “Sunʼiy intellekt texnologiyalarini 2030-yilga qadar rivojlantirish strategiyasini tasdiqlash toʻgʻrisida”. Qonunchilik maʼlumotlari milliy bazasi, 17.10.2024-y., 07/24/358/0825-son; 27.10.2025-y., 06/25/189/0978-son; 22.11.2025-y., 06/25/223/1076-son https://lex.uz/docs/-7158604

Oʻzbekiston Respublikasi Prezidentining qarori, 06.11.2020 yildagi PQ-4884-son Farmoni “Taʼlim-tarbiya tizimini yanada takomillashtirishga oid qoʻshimcha chora-tadbirlar toʻgʻrisida”. Qonun hujjatlari maʼlumotlari milliy bazasi, 07.11.2020-y., 07/20/4884/1484-son; 26.03.2021-y., 07/21/5040/0243-son https://lex.uz/docs/-5085887

Oʻzbekiston Respublikasi Prezidentining, 03.07.2018 yildagi PQ-3832-son Qarori “Oʻzbekiston respublikasida raqamli iqtisodiyotni va kripto-aktivlar aylanmasi sohasini rivojlantirish chora-tadbirlari toʻgʻrisida”. Qonunchilik maʼlumotlari milliy bazasi, 24.07.2021-y., 06/21/6268/0700-son; 28.04.2022-y., 06/22/121/0354-son; 19.09.2024-y., 06/24/140/0728-son; 28.11.2025-y., 07/25/359/1106-son; 20.04.2026-y., 07/26/143/0386-son https://www.lex.uz/docs/-3806053

Oʻzbekiston Respublikasi Prezidentining, 05.10.2020 yildagi PF-6079-son Farmoni “Raqamli oʻzbekiston — 2030” strategiyasini tasdiqlash va uni samarali amalga oshirish chora-tadbirlari toʻgʻrisida”. Qonun hujjatlari maʼlumotlari milliy bazasi, 06.10.2020-y., 06/20/6079/1349-son; 31.10.2020-y., 06/20/6099/1450-son; 02.04.2021-y., 06/21/6198/0269-son https://lex.uz/docs/-5030957

Roll, I., & Wylie, R. (2016). Evolution and revolution in artificial intelligence in education. International Journal of Artificial Intelligence in Education, 26(2), 582–599. https://doi.org/10.1007/s40593-016-0110-3

Sapiah, S., Ulfah, S. M., Saputra, A. N., & Hardi, R. (2025). Smart education in remote areas: Collaborative strategies to address challenges in Majene Regency, Indonesia. Frontiers in Education, 10, 1552575. https://doi.org/10.3389/feduc.2025.1552575

Selwyn, N. (2019). Should Robots Replace Teachers? AI and the Future of Education. Cambridge: Polity Press.

Tlili, A., Saqer, K., Salha, S., & Huang, R. (2025). Investigating the effect of artificial intelligence in education (AIEd) on learning achievement: A meta-analysis and research synthesis. Information Development, 41(3), 825–842. https://doi.org/10.1177/02666669241304407

UNESCO IITE. (2022). AI Competency Framework for Teachers. Moscow: UNESCO Institute for Information Technologies in Education. https://iite.unesco.org/publications/ai-competency-framework-for-teachers/

UNESCO. (2023). AI and Education: Guidance for Policy-makers. Paris: United Nations Educational, Scientific and Cultural Organization. https://unesdoc.unesco.org/ark:/48223/pf0000380789

Wang, X., Huang, R. T., Sommer, M., Pei, B., Shidfar, P., Rehman, M. S., Ritzhaupt, A. D., & Martin, F. (2024). The efficacy of artificial intelligence-enabled adaptive learning systems from 2010 to 2022 on learner outcomes: A meta-analysis. Journal of Educational Computing Research, 62(6), 1348–1383. https://doi.org/10.1177/07356331241240459

World Bank. (2022). Digital Education in Central Asia: Opportunities and Challenges for AI Integration. Washington, DC: World Bank Group. https://doi.org/10.1596/38941

Wu, J., Tlili, A., Salha, S., Mizza, D., Saqr, M., López-Pernas, S., & Huang, R. (2025). Unlocking the potential of artificial intelligence in improving learning achievement in blended learning: A meta-analysis. Frontiers in Psychology, 16, 1691414. https://doi.org/10.3389/fpsyg.2025.1691414

Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019). Systematic review of research on artificial intelligence applications in higher education. International Journal of Educational Technology in Higher Education, 16(1), 39. https://doi.org/10.1186/s41239-019-0171-0