Chile · Santiago · See the Chile profile · See the Santiago profile
Evidence: Observational / pre–post Top 14% 67/100 · Ask Evidence Copilot about this practice
UC Chile's AyudantIA trains professors to build course-specific AI tutoring chatbots; by 2026, 194 agents across ~100 courses reached 4,600+ students, with 60%+ reporting the tutors aided learning and two-thirds of professors rating the impact positively.
AyudantIA is an institution-wide initiative at the Pontificia Universidad Católica de Chile (UC Chile) that trains professors to build their own course-specific AI tutoring agents — chatbots populated with each course's own materials and bibliography — run jointly by UC Chile's Vice Presidency for Digital Intelligence and Vice Presidency for Academic Affairs on Microsoft Azure OpenAI infrastructure integrated with the Canvas learning platform.
Give students on-demand, 24/7, instructor-validated AI tutoring support as a safe alternative to unsupervised generative-AI use — UC Chile cites survey data showing 94% of its students already use AI tools informally.
By early 2026, 92 professors had built 171 agents (59 on GPT-4o, 112 on GPT-4o-mini) across 26 general-education and 68 discipline-specific courses; a later count reported 194 agents deployed across roughly 100 courses. UC Chile is collaborating with the University of Sydney and Microsoft to study responsible scaling toward a stated goal of 5,000 courses.
The agents reached over 10,000 students with access, with 4,600+ students actively engaging in sessions averaging 13–23 minutes. A first-semester evaluation found over 60% of students said the agents contributed to their learning, 90% wanted continued access, and two-thirds of professors rated the impact positively; one Microbial Biology course reported average seminar grades above 6 on Chile's 1–7 scale after adoption.
AyudantIA shows strong early engagement and largely positive self-reported perceptions at meaningful scale, but the evidence to date is a first-semester institutional survey without a comparison group, not a controlled evaluation of learning gains.
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.
Where this practice's information was retrieved from, and when.
Hungary
A 13-week controlled study at four Hungarian universities found a ChatGPT-based adaptive tutor using Bayesian Knowledge Tracing nearly doubled programming …
Denmark
Two experiments (175 university, 234 high-school students) found a learning-theory-based chatbot beat ChatGPT on conceptual knowledge, trust and enjoyment; gains …
United States of America
A randomized trial across four sections (140 students) at Wisconsin-La Crosse found students pairing peer study groups with Macro Buddy, …
Nigeria
A six-week World Bank RCT in nine Edo State public schools showed GPT-4 tutoring (Microsoft Copilot) delivered English gains equivalent …
Open full copilot Grounded in cited practices — always check the sources.