evidoria

← Back to browse

Good practice Imported

Eligiendo Mi Camino — World Bank's AI Math Tutor, Career Coach and Teacher AI-Training in Lima's Public Schools

Peru · Lima · See the Peru profile · See the Lima profile

Evidence: Randomised controlled trial Top 17% 67/100 · Ask Evidence Copilot about this practice

The World Bank, Peru's DRELM, Anthropic and Microsoft launched an AI maths tutor and career-guidance coach across 110 Lima public schools (6,600 students); a related RCT of 1,270 teachers found brief training raised active AI use, while student-outcome results remain pending.

390 schools
Schools in teacher AI-adoption RCT (2025)
1,270 teachers
Teachers in RCT (2025)
85 schools
Schools at programme launch (March 2026)
~4,500 students
Students at programme launch (March 2026)
110 schools
Schools by mid-2026 (mid-2026)
6,600 students
Students by mid-2026 (mid-2026)
142 occupations across 12 economic sectors
Occupations covered by AI Vocational Coach (n/a)
Eligiendo Mi Camino — World Bank's AI Math Tutor, Career Coach and Teacher AI-Training in Lima's Public Schools

Details

Maturity
Scaling
Promoter
World Bank Group (Development Research Group) / DRELM Lima / uDocz / Anthropic / Microsoft
Period
Teacher AI-adoption RCT 2025; Eligiendo Mi Camino launched March 2026, ongoing
Keywords
secondary education, career guidance, teacher training, EdTech

Context

In Peru's public school system, only 1 in 10 secondary students finishes with adequate maths proficiency and barely 3 in 10 go on to higher education; Peru ranks fifth worldwide for its NINI rate (youth neither in education, employment nor training), per UNICEF. The World Bank's Development Research Group partnered with DRELM (Lima's regional education authority), ed-tech platform uDocz, Anthropic and Microsoft to test whether AI could help close these gaps.

Activities

A randomized controlled trial across 390 Lima schools tested whether brief, practice-oriented training helped teachers adopt AI in their classrooms. Building on this, the World Bank and DRELM launched 'Eligiendo Mi Camino' (Choosing My Path) in March 2026, combining an AI Math Tutor that gives fifth-year secondary students personalised support with real-time difficulty detection, and an AI Vocational Coach covering 142 occupations across 12 economic sectors, built around three pillars: self-knowledge, exploration and decision.

Results

Active AI use increased among the 1,270 participating teachers in the RCT, who also learned to write stronger prompts and evaluate AI outputs more critically; a related validation exercise found that AI-based classroom observations scoring pedagogical and instructional quality produced results comparable to those of trained human observers. The programme itself reached roughly 85 public schools and 4,500 students at launch, expanding to 110 schools and 6,600 students by mid-2026, with teachers supervising AI use during class.

Conclusions

The World Bank is explicit that student learning outcomes remain an open question: a dedicated randomized trial is now underway to test whether the AI tutor builds lasting maths skills rather than short-term gains, and results were not yet public as of this review. The strongest verified evidence to date concerns teacher AI-adoption and the reliability of AI-assisted classroom observation, not confirmed student achievement gains.

Implementation

Indicative cost
Medium (€50k–€500k)
Time to results
Medium (1–3 years) — Teacher AI-adoption RCT conducted in 2025; Eligiendo Mi Camino launched March 2026 and expanded from 85 to 110 schools by mid-2026, with a student-outcome RCT now underway.
Staffing & skills
World Bank Development Research Group, DRELM (Lima regional education authority), uDocz (ed-tech platform), Anthropic, Microsoft, Classroom teachers supervising AI use

Conditions for success

  • Brief, practice-oriented teacher training shown (via RCT) to increase active, critical AI use before wider tool rollout.
  • Teachers supervising AI use during class rather than pupils using tools unsupervised.
  • A dedicated evaluation trial built in from the start to test whether tutor gains are lasting, not just short-term.

Where it fits

Governance type
multilateral-development-bank and regional-government partnership
Scale
regional (Lima), expanding

Commonly funded by

National / regional programmes Philanthropic / foundation funding

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Replication kit

Reusable artefacts from this practice — as published by their sources.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

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

Attachments

Similar practices you may find useful