Empowering Search and Improving Investment in Human Capital in Colombia

AI Assisted Search Bot for College Major Choice

In this project we collaborate with the government of Colombia to investigate whether information and behavioral frictions affect the decision of whether and where to enroll in tertiary education. We test low-cost innovative, personalized approaches to overcome informational and behavioral obstacles at scale using a randomization research design. These approaches leverage existing large-scale government datasets and the ability to personalize information using cutting-edge, technology-based interactive approaches.

The chatbot is an implementation of text messaging that allows for dynamic interactions with students. This interactive chatbot treatment allows for personalization of the provision of information based on interaction with the student. Specifically, it allows for a low cost, guided-search, through the space of potential choices taking into account the student preferences. A dynamic version of the bot will learn over time about the best way to communicate with different types of students. One of our contributions with this approach is to test at scale a replicable innovation that is relatively cost-effective vis-à-vis college counselors or other informed members of students’ social network.

The experiment is designed to test the differential impact of information provision through traditional static channels and dynamic interactition and personaliztion provided by a chatbot. The data collected will also allow the researchers to test several models of information search behavior.