TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Google has published findings from a 10-week UK math-tutoring trial and a yearlong study in 31 Italian schools. The company reports that Gemini-supported tutoring produced results similar to human tutoring on classroom tasks and quizzes, while Italian teachers reported more collaborative lessons; the UK trial did not find gains on STAR standardized-test scores.
Google has reported findings from two studies of its Gemini education tools: a 10-week UK trial of AI-assisted math tutoring and a yearlong project across 31 Italian schools. The company says the tutoring performed similarly to human tutoring on classroom assignments and quizzes, while Italian educators reported more collaborative lessons when using Gemini to help plan; the UK trial found no corresponding improvement on standardized STAR test scores.
The UK randomized controlled trial involved more than 1,100 students aged 12 to 15, who used Eedi’s school platform. When students answered a math question incorrectly, they received either a static hint, help from a human tutor, or interactive tutoring combining Eedi’s teaching resources with Gemini. Teachers monitored progress and supported students, and human experts reviewed each AI response before it was shown to a student.
Google says students receiving AI tutoring performed similarly to students paired with human tutors on platform tasks and intermediate quizzes. Expert reviewers approved 95% of Gemini responses, up from 76% in the previous study, which the company describes as a nearly 25% improvement in pedagogical quality. Google also says Gemini guided students rather than supplying answers in nearly 99% of sessions, and students worked through the math themselves in more than 85% of chats.
The Italian study ran for a year with the Salesians of Don Bosco school network and the Pontifical Salesian University in Rome. It drew on reports and classroom activities involving 1,600 teachers and more than 24,500 students aged 13 to 18. According to Google, teachers who used Gemini to co-design lessons said they felt more creative in planning and teaching than those who did not use it: 79% compared with 30%. The company also reports a 38% drop in traditional front-of-class lectures and nearly three times as much student-led collaborative learning in the lessons studied.
Classroom Gains, With a Test-Score Gap
The findings speak to two practical questions for schools considering generative AI: whether it can support learning without simply giving students answers, and whether it can help teachers organize more active lessons. Google’s UK results suggest that, with teacher involvement and expert review, an AI tutor can support students through problems and match human tutoring on some classroom measures. The Italian findings suggest a possible role in lesson design that shifts classroom time toward group work and discussion.
But the UK results also show why classroom performance and broader academic outcomes should not be treated as interchangeable. Students using AI tutoring improved on daily tasks and intermediate quizzes, yet the trial found no gains on STAR standardized tests. That distinction matters to schools weighing the benefits of more guided practice against curriculum coverage, test outcomes, and the time teachers need to supervise AI use.
Products Worth Considering
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Bose Lifestyle Ultra Wireless Home Speaker
As an affiliate, we earn on qualifying purchases.
How the Two Studies Were Run
The UK trial builds on an earlier exploratory study by Google and Eedi. In the newer study, researchers compared three kinds of help after an incorrect answer: a hint, a human tutor, and an AI tutor combining Gemini with Eedi’s pedagogical resources. Student progress was tracked through platform assignments, intermediate quizzes, and independent STAR tests. Google says the work was designed to examine tutoring in structured lessons over time, rather than rely only on student or teacher impressions.
Google links the lack of STAR score gains to the pace of AI tutoring. Students solved an average of about 14 problems per session with AI, compared with 19 when given static hints. The company says the step-by-step approach took longer per question, so students covered less curriculum over the 10-week period, even as their pacing and understanding improved. The Italian project, by contrast, focused on classroom activity, lesson planning, student engagement, and teacher wellbeing across primary, secondary, and vocational settings.
“Technology best serves education when it creates space for inspiring teaching and deeper learning.”
— Google, in its October 29, 2026 report
Long-Term Learning Remains Untested
The UK trial lasted 10 weeks, and Google says that period may not have been long enough for students to cover the range of material assessed by the STAR tests. The company has not reported evidence from a full academic-year follow-up, so it remains unclear whether the classroom and quiz gains would lead to improvements on standardized tests over a longer period.
The Italian findings are based on classroom activities and teacher self-reports, as described in Google’s account. The report does not establish that Gemini alone caused the changes in teaching style or student collaboration, nor does the supplied material give detailed comparisons of academic outcomes between classrooms using the tool and those not using it. The results therefore do not establish that the reported patterns will occur in other schools or education systems.
Google Plans Longer UK Trials
Google says it is exploring follow-up tutoring trials across a full academic year and plans to broaden its measures of learning to understand what contributes to meaningful progress and later outcomes. Longer-term results could help show whether students can retain the benefits of step-by-step guidance while covering enough curriculum to affect standardized assessments.
For now, the evidence covers specific programs, age groups, and school settings in the UK and Italy. Schools and researchers will need further data on academic outcomes, classroom implementation, and the time required for human oversight before drawing wider conclusions about Gemini’s effectiveness in education.
Key Questions
What did the UK study find about Gemini math tutoring?
Google says students using AI tutoring performed similarly to students paired with human tutors on platform tasks and intermediate quizzes. The trial did not find gains on STAR standardized-test scores.
Did Gemini give students the answers?
Google says Gemini guided students rather than handing over answers in nearly 99% of sessions, and students worked through the math themselves in more than 85% of chats. Human experts reviewed each response before it reached students in the trial.
What did teachers in Italy report?
In Google’s account, 79% of teachers using Gemini to co-design lessons said they felt more creative in their teaching and planning, compared with 30% of teachers not using it. Google also reported fewer lecture-style lessons and more student-led collaboration.
Why were there no STAR test gains in the UK trial?
Google says AI tutoring took longer per question, and students averaged about 14 problems per session, compared with 19 for students receiving static hints. The company suggests that students covered less curriculum during the 10-week trial, but the explanation has not yet been tested in a full-year study.
What further research is planned?
Google says it is exploring follow-up trials across a full academic year and broader measures of learning. The company has not provided a timetable for those studies in the supplied report.
Source: rss
