Ceibal's AI-Assisted Feedback Tool for Online Teacher-Training Tutors (Uruguay)

Fundación Ceibal, Uruguay's public digital-education agency, working with the national education authority ANEP, built an AI-assisted correction tool for tutors on its online courses for teachers in training and practising educators, using OpenAI models as the underlying AI service within Ceibal's own workflow and code.

The tool aims to speed up feedback turnaround for tutors and shift written feedback away from generic remarks toward specific, personalised comments delivered close to submission time, while keeping a human tutor in control of the final grade and comments.

When a student submits a task, the AI drafts a correction with detailed comments and a suggested grade; the tutor then reviews it, can edit both the grade and the written comments, and only then sends it to the student, a human-in-the-loop design rather than an autograder. Ceibal is extending the same model to other assessment contexts, including a planned 2026 platform, Acredita, for adults seeking secondary-school equivalency certification, and a separate OEI-funded AI writing assistant for ninth-graders' essays.

Ceibal reports the tool sped up turnaround time and moved feedback from generic remarks ("very good, continue") to specific, personalised comments. The project was recognised as best academic work at a regional educational-technology conference and was a finalist for a PwC innovation award; Stanford University researchers are separately studying its outcomes, though no results had been published as of reporting.