evidoria

← Back to browse

Good practice

Eduten AI-Adaptive Mathematics — Nationwide Rollout in Mongolia

Mongolia · Ulaanbaatar · See the Mongolia profile

Mongolia's Ministry of Education rolled out the AI-adaptive Eduten maths platform nationwide, reaching 134,000+ students in 650+ schools by 2023 and expanding to grades 1-9 in 2025, after a 12-week pilot found 22% greater improvement than a control group.

22% greater improvement
Learning-outcome improvement vs control group (12-week pilot, early 2023)
94%
Students reporting the platform helped them understand maths better (12-week pilot, early 2023)
87%
Teachers satisfied with the platform (12-week pilot, early 2023)
134,000+ students, 650+ schools
Students reached (national rollout, grades 3-4) (from September 2023)
5,200+
Teachers trained (national rollout) (from September 2023)
230 million+
Submitted answers recorded
696,000+ students
Planned 2025 expansion (grades 1-9) (2025 school year)
Eduten AI-Adaptive Mathematics — Nationwide Rollout in Mongolia

Details

Maturity
Scaling
Promoter
Ministry of Education and Science of Mongolia; Eduten Oy; MNIER; University of Turku
Period
2023-ongoing (pilot Jan-Mar 2023; national rollout Sept 2023; expanded to grades 1-9 in 2025)
Keywords
adaptive learning, mathematics education, machine learning, K-12 education

Context

Eduten is an adaptive mathematics-learning platform built on roughly two decades of research at the University of Turku's Centre for Learning Analytics (the 'ViLLE' platform) and commercialised by the Finnish company Eduten Oy. Its machine-learning engine draws on a bank of about 200,000-250,000 exercises and roughly 2.5 billion prior student responses to build an individual profile for each learner, aiming to keep the share of correctly-answered items within an 80-85% 'optimal challenge' band, while also automating grading to reduce teacher workload.

Activities

Mongolia's Ministry of Education and Science (MoES) piloted Eduten with the National Institute of Education Research (MNIER) and the Education Evaluation Center over 12 weeks in early 2023, comparing students using the platform against a control group that also received extra weekly maths practice. Following the pilot, MoES rolled Eduten out nationally in September 2023 for grades 3-4 across public schools, training more than 5,200 teachers, with coverage subsequently planned to expand to grades 1-9 for the 2025 school year.

Results

MoES reported the Eduten group achieved 22% greater improvement in learning outcomes than the control group, alongside high self-reported satisfaction (94% of students said it helped them understand maths better, 87% of teachers were satisfied). The national rollout reached more than 134,000 students in over 650 schools and recorded over 230 million submitted answers; independent confirmation of on-the-ground use comes from Mongolia's state news agency MONTSAME, which reported a February 2024 event in Zavkhan aimag involving 29 general-education schools and 38 kindergartens.

Conclusions

The headline 22% relative-improvement result and enrolment figures are published through Eduten's own blog, attributing the underlying data to MNIER and the Education Evaluation Center; no independently published evaluation report or peer-reviewed study reproducing these figures was found. The pilot's own report flags school-infrastructure readiness as a risk for wider implementation, relevant given Mongolia's low population density and rural, partly nomadic population, and no independent evaluation of the national rollout's learning outcomes has been published.

Implementation

Indicative cost
High (€500k–€5M)
Time to results
Medium (1–3 years)
Staffing & skills
Ministry of Education and Science of Mongolia (MoES), National Institute of Education Research (MNIER), Education Evaluation Center, Eduten Oy, University of Turku Centre for Learning Analytics

Conditions for success

  • Large adaptive item bank (~200,000-250,000 exercises, ~2.5 billion prior responses) enabling personalisation
  • Algorithm tuned to keep students in an 80-85% correct-answer 'optimal challenge' band
  • Automated grading to reduce teacher workload
  • National teacher-training programme (5,200+ teachers trained)

Common failure modes

  • Pilot report flags school-infrastructure readiness as a risk given Mongolia's low population density and rural, partly nomadic population
  • No independent evaluation of the national rollout's learning outcomes has been published, only the original 12-week pilot

Where it fits

Governance type
national ministry-led rollout with academic/vendor partnership
Scale
national, expanding from grades 3-4 to grades 1-9
Income level
lower-middle-income country (Mongolia), rural/low-density context

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful