TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Google published findings from two studies of Gemini in education: a 10-week UK math-tutoring trial involving more than 1,100 students and a yearlong Italian study across 31 schools. The company reported results comparable to human tutors on some UK measures and more collaborative lessons in Italy, but the UK trial found no gains on STAR standardized test scores.
Google published two studies on October 29 examining Gemini’s use in schools: a 10-week UK math-tutoring trial with more than 1,100 students and a yearlong Italian study spanning 31 schools. The company said students using AI tutoring performed similarly to students working with human tutors on platform tasks and quizzes, while Italian teachers reported more collaborative classroom activities; the UK trial did not find gains on STAR standardized test scores.
The UK randomized controlled trial involved students aged 12 to 15 using the Eedi school platform. When students answered a question incorrectly, they received a static hint, help from a human tutor, or interactive AI tutoring that combined Eedi’s teaching resources with Gemini. Google said teachers monitored progress and provided classroom support. Human experts reviewed every AI response in real time before it reached a student.
Google reported that expert approval of the AI tutor’s responses rose from 76% to 95% compared with the previous year’s study, describing this as a nearly 25% improvement in pedagogical quality. In nearly 99% of sessions, the tutor reportedly guided students rather than giving them answers; students worked through the math themselves in more than 85% of chats. On platform tasks and intermediate quizzes, students receiving AI tutoring performed similarly to those paired with human tutors.
The Italian study collected classroom activities and teacher self-reports over a year across 31 schools in the Salesians of Don Bosco network. It covered 1,600 teachers and more than 24,500 students aged 13 to 18. Google reported that teachers who used Gemini to co-design lessons said they felt more creative in planning and teaching than those who did not use it, and that lecture-style teaching decreased as workshops, group work and peer discussion became more common.
Tutoring Results and Classroom Practice
The findings address two practical questions for schools: whether AI can offer learning support at a scale that resembles human tutoring, and whether teacher-facing AI tools can change how lessons are organized. In the UK trial, comparable performance on platform assignments and quizzes suggests Gemini-supported tutoring may help students work through problems, but the lack of improvement on the standardized test limits what can be concluded about broader learning gains.
Google’s Italian results point to a possible change in classroom activity when teachers use Gemini during lesson planning. More workshops and student collaboration could create additional opportunities for peer learning, but the reported findings come from a study described by Google and rely in part on teachers’ self-reports. They do not establish that Gemini alone caused the changes or that the same outcomes will occur in other school systems.
The research also highlights the role of human oversight. In the UK trial, expert review of AI replies and active teacher involvement were part of the study design. The results therefore describe AI tutoring operating within a supervised educational setting, not a test of unsupervised use.
Products Worth Considering
Apple iPad Air (2025) 13‑inch Wi‑Fi 256 GB – Space Gray (M3 chip)
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
Anker SOLIX S2000 Portable Power Station, 2,010Wh, 1,500W
As an affiliate, we earn on qualifying purchases.
How the Two Studies Were Run
Google said it had previously conducted an exploratory study with Eedi, finding that an AI tutor could operate in real classrooms. The newer UK trial was designed to examine learning over time and compare different kinds of support after incorrect answers. Researchers tracked progress using platform assignments, intermediate quizzes and independent STAR standardized tests.
The UK results did not move in the same direction across all measures. Google said AI tutoring improved performance on daily platform tasks and quizzes, but students using it solved about 14 problems per session, compared with about 19 for students receiving static hints. The company suggested that step-by-step tutoring took longer per question, leaving students with less time to cover curriculum material during the 10-week trial. That explanation is Google’s interpretation of the findings, not a demonstrated cause of the STAR result.
In Italy, researchers worked with the Salesians of Don Bosco school network and the Pontifical Salesian University in Rome. The study included primary, secondary and vocational educators, while the reported student age range was 13 to 18. The study compared classroom activities and teacher reports from lesson planning with and without Gemini for Education.
“AI tutoring using Gemini capabilities can deliver support on par with human tutors for UK students when used responsibly with human oversight.”
— Google, in its report
Limits of the Reported Evidence
The UK trial found no improvement in STAR standardized test scores over its 10-week period, despite gains on platform tasks and intermediate quizzes. Google said the AI group completed fewer questions and may have covered less curriculum, and it plans longer trials to investigate that possibility. The report does not establish whether a longer period would produce standardized-test gains.
The findings supplied by Google do not provide enough detail to assess every aspect of study design, including how participants were assigned, the size of measured effects, or the full statistical results. The Italian figures are reported outcomes from a study involving classroom activities and teacher self-reports; the material does not establish whether changes in lesson format led to lasting improvements in student achievement or wellbeing. Findings from these schools may not apply to other settings.
Longer Trials and Broader Measures
Google said it is exploring follow-up UK trials lasting a full academic year and plans to expand the measures used to study learning and later outcomes. Those studies could help clarify whether the slower pace of AI-guided tutoring affects curriculum coverage and whether short-term quiz results correspond to standardized-test performance over time.
The company’s report does not give a schedule for those trials or specify when further results will be released. For now, the findings offer evidence about Gemini use in the particular supervised UK trial and participating Italian schools, while longer-term effects remain open.
Key Questions
What did the UK study test?
A 10-week randomized controlled trial compared static hints, human tutoring and AI tutoring using Gemini with Eedi resources among more than 1,100 students aged 12 to 15.
Did Gemini tutoring improve students’ standardized test scores?
No gains appeared on STAR standardized test scores during the trial, according to Google. The company reported better results on platform tasks and intermediate quizzes, which are different measures.
How was the AI tutoring supervised?
Google said teachers monitored student progress and supported students in classrooms, while human experts reviewed each AI response in real time before it was shown to a student.
What did the Italian study report?
Across 31 schools, teachers using Gemini to co-design lessons reported more creativity in planning and teaching. Google also reported fewer lecture-style lessons and more workshops, group activities and peer discussion.
Will Google conduct more research?
Google said it is exploring full-academic-year follow-up trials in the UK and broader measures of learning outcomes. It did not provide a timetable for those studies.
Source: blog.google
