Where Vocational Counsellors Are Scarce, Universidad Técnica de Machala Builds a DeepSeek-Powered Career-Guidance Chatbot for Ecuadorian Students

Ecuador's secondary schools face a shortage of trained vocational-guidance counsellors and limited access to guidance technology. Researchers at Universidad Tecnica de Machala (UTMACH) built a chatbot powered by the DeepSeek language model, using prompt-engineering techniques to interpret students' open-ended answers and match them to a suggested academic path drawn from UTMACH's own programme offerings.

The team evaluated it with a quasi-experimental design combining a treatment group and a control group, involving 351 third-year secondary students. Students completed the government's official SENESCYT vocational aptitude test, then interacted with the AI chatbot, then completed a Likert-scale perception survey. The chatbot's suggestions matched the SENESCYT test results in 80.1% of cases, and 76% of students rated the conversational interaction, question clarity and relevance of recommendations positively.

The 80.1% concordance is a meaningful validation against an external government benchmark, but it also means roughly one in five recommendations diverged from the official test, and the chatbot's suggested pathways were drawn specifically from UTMACH's own course catalogue -- a caveat prospective adopters elsewhere should weigh before reusing the tool unmodified.