evidoria

← Back to browse

Good practice Imported

GPT-3-Based Admissions Chatbot at Pontificia Universidad Católica del Ecuador

Ecuador · Quito · See the Ecuador profile · See the Quito profile

Evidence: Descriptive / self-reported Top 100% 7/100 · Ask Evidence Copilot about this practice

A GPT-3 and NLU chatbot built by PUCE researchers, with partners in Ecuador and Spain, answers applicants' questions during admissions. Researchers report 90% satisfaction, though sample size and method are undisclosed.

90 %
Applicant satisfaction with the chatbot's ability to understand questions and hold a useful conversation
GPT-3-Based Admissions Chatbot at Pontificia Universidad Católica del Ecuador

Details

Maturity
Pilot
Promoter
Pontificia Universidad Católica del Ecuador (PUCE) — Artificial Intelligence Research Group
Period
2024
Keywords
higher education, admissions, chatbot, GPT-3, NLU

Context

Researchers from PUCE's Artificial Intelligence Research Group, working with Universidad Politécnica Salesiana (Ecuador) and the University of Vigo (Spain), built a conversational chatbot for PUCE's admissions office.

Objectives

The chatbot combines OpenAI's GPT-3 autoregressive language model with natural-language-understanding techniques so applicants can ask admissions questions in ordinary Spanish rather than navigating rigid menus.

Activities

The chatbot was deployed to answer applicants' admissions questions directly, in place of rigid menu-based interfaces.

Results

The published paper reports a 90% satisfaction rate among applicants for the chatbot's ability to understand questions and hold a useful conversation. The sample size and evaluation method behind that figure are not disclosed.

Conclusions

This should be read as a single self-reported metric from an early-stage deployment rather than an independently audited result; the paper states it documents limitations and risks for future improvement.

Implementation
Indicative cost
Low (< €50k) — Built using OpenAI's GPT-3 API plus NLU techniques by a university research group; no public budget figures disclosed.
Time to results
Short (< 1 year) — Reported as an early-stage admissions-support deployment; no rollout timeline disclosed beyond the study period (2024).
Staffing & skills
PUCE Artificial Intelligence Research Group, Partners at Universidad Politécnica Salesiana (Ecuador) and University of Vigo (Spain)

Conditions for success

  • Combining GPT-3 with NLU techniques for natural Spanish-language interaction
  • Cross-institutional research partnership (Ecuador and Spain)

Common failure modes

  • Sample size and evaluation method behind the 90% satisfaction figure are undisclosed
  • No independent audit of the reported result

Commonly funded by

Own resources / municipal budget

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

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