Gemini's Guided Learning: Results from a Randomized Controlled Trial in Sierra Leone

TL;DR

An eight-week trial in Sierra Leone involving 1,763 students evaluated the impact of Guided Learning in Gemini on math progress.
Students using Guided Learning showed a gain of +0.258 standard deviations in math scores, equivalent to 1.2 to 1.7 years of learning progress.
Students in classrooms where teachers integrated Gemini for roughly half their lessons saw even higher gains, equivalent to 1.8 to 2.5 years of progress.
The AI tool prioritized building understanding over providing direct answers, with students using it for conceptual understanding in 91.4% of interactions.
Gemini responded with scaffolding questions in 76% of messages and provided direct solutions in only 2% of cases.
Teachers reported the AI supported their professional growth and helped shift their roles from 'lecturers' to 'facilitators'.
Student engagement was high, with 69% meeting or exceeding usage targets, and students reported enjoying math more and engaging beyond regular instruction.
Skill-building queries increased from 68% to 90% over the trial, while solution-seeking questions dropped from 25% to 10%.
The trial highlighted that students who entered with stronger math skills benefited the most, indicating a need for tools that support students needing the most help.
Resources, including a teacher training guide and a playbook on RCTs, are being released to help others implement similar programs and studies.

Continuer la lecture de l'article original