evidoria

← Back to browse

Good practice

Elige Educar 'Kai' — AI counsellor for higher-education career guidance in Chile (IDB RCT)

Chile · Santiago · See the Chile profile

IDB/Elige Educar WhatsApp AI chatbot for career guidance to Chilean secondary students. Pre-registered RCT (43,000+ students, 2024–25): AI raised initiation rates (35% vs 28% human counsellors) but produced shorter conversations — evidence of AI's trade-offs in student guidance.

35 %
Kai chatbot conversation-initiation rate (2024–25 admissions cycle)
28 %
Human-counsellor conversation-initiation rate (2024–25 admissions cycle)
3.7 messages
Mean messages per student with Kai
5.9 messages
Mean messages per student with human counsellors
43,000+ students
Students enrolled in the RCT (2024–25)

Details

Maturity
Pilot
Promoter
Elige Educar / Inter-American Development Bank
Period
2024–2025
Keywords
career guidance, AI chatbot, higher education, admissions, NGO

Context

Elige Educar's 'Quiero Ser Profe' programme provides personalised WhatsApp-based counselling to Chilean secondary students considering teaching as a career. In 2024, the Inter-American Development Bank (IDB) partnered with Elige Educar to pilot 'Kai', a generative-AI WhatsApp chatbot designed to replicate the personalised guidance of skilled human counsellors at much lower cost and at scale, adapting to student responses to provide tailored information about teaching careers, application processes and scholarships.

Results

A pre-registered RCT during the 2024–25 Chilean university admissions cycle enrolled over 43,000 secondary students across three arms: the Kai AI chatbot, human Elige Educar counsellors, and a no-intervention control. Kai achieved a higher initiation rate (35% of assigned students began a conversation) than human counsellors (28%), suggesting the AI lowered psychological barriers to engagement. However, conversations with Kai were markedly shorter (mean 3.7 messages per student) than with human counsellors (mean 5.9 messages), and the IDB evaluation concluded the AI chatbot was cheaper and more scalable but less impactful per interaction, owing to limited conversational depth and personalisation. Enrolment and career-choice outcomes for the full cohort had not yet been published as of mid-2026.

Conclusions

This RCT is one of the most rigorously evaluated AI counselling pilots in Latin America. Its candid findings — that AI can extend reach while trading depth for breadth — offer replicable insights for any jurisdiction facing a shortage of qualified student counsellors.

Implementation

Indicative cost
Low (< €50k) — Explicitly described as cheaper than human counsellors; cost_band set to low.
Time to results
Short (< 1 year) — Piloted across a single admissions cycle (2024–25); timeline_band set to short.
Staffing & skills
Human Elige Educar counsellors form the comparison arm and continue operating alongside Kai

Conditions for success

  • WhatsApp delivery, widely accessible to Chilean secondary students
  • Pre-registration and IDB institutional oversight ensuring methodological rigour

Common failure modes

  • Shorter, less personalised conversations than human counsellors
  • Requires smartphone/WhatsApp access, partially limiting equity
  • Enrolment and career-choice outcomes not yet published

Data sources

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

Similar practices you may find useful