Surveys show that AI tools are already common in schools, but they measure use and self-reported time saved, not learning. A June 2025 Gallup survey of 2,232 U.S. public K-12 teachers found that six in 10 used an AI tool for work in the 2024-25 school year and about three in 10 used one weekly, while a fall 2025 survey by KPMG in Canada found that 73 percent of Canadian adults who are students used generative AI for schoolwork. ChatGPT is one product among those counted, and most of these surveys ask about AI tools in general.
Key takeaways
- Gallup reported that 32 percent of U.S. public school teachers used AI at least weekly in 2024-25, and that weekly users estimated saving 5.9 hours per week.
- The 5.9-hour figure is a self-reported estimate built from task-by-task answers, not a timed measurement.
- A 2026 Gallup report found that only 18 percent of teachers had received formal guidance from school administrators on using AI tools.
- In Canada, a KPMG survey reported student use of generative AI for schoolwork at 73 percent in 2025, up from 52 percent in 2023, and no Statistics Canada figure on the topic was found.
- None of these surveys tests whether students learn more or less with ChatGPT; that needs a comparison group, as in a trial.
What U.S. teachers report using AI for
The Walton Family Foundation-Gallup study, published by Gallup on June 24, 2025, surveyed 2,232 teachers from the RAND American Teacher Panel between March 18 and April 11, 2025. Six in 10 said they had used an AI tool for their work during 2024-25. Across nine tasks, 32 percent used AI at least weekly and 28 percent used it monthly or less. Among tasks done at least monthly, preparing to teach (37 percent), making worksheets or activities (33 percent), and modifying materials for student needs (28 percent) led. The margin of error for the full sample is 2.5 percentage points.
Weekly users estimated that AI saves them 5.9 hours a week, which the report converts to about six weeks over a 37.4-week school year. The Gallup write-up explains the method: each task was estimated in half-hour steps and non-overlapping tasks were added together, and only teachers who said AI changes their time on a task were included. That makes the figure a reasonable estimate of what a weekly user perceives, and a weak basis for projecting savings across all teachers.
The guidance gap reported in 2026
A second Walton Family Foundation-Gallup report, "Closing the Expectations Gap," drew on 2,069 U.S. public K-12 teachers surveyed from February 9 to March 2, 2026. Gallup's article on it, dated May 26, 2026, reports that 18 percent of teachers had received formal guidance from school administrators on AI tools. Across 10 tasks, 34 percent got no guidance at all and 48 percent got only informal guidance, defined as verbal conversations or shared norms. For one-on-one instruction or tutoring, 69 percent received none, and for grading and student feedback, 58 percent. Encouragement was more common than discouragement: the highest share told not to use AI was 10 percent, for grading. In practice, many individual teachers are deciding for themselves.
What Canadian and U.S. student figures show
For students, the Pew Research Center reported in January 2025 that 26 percent of U.S. teens had used ChatGPT for schoolwork, double the 2023 share; that survey asked about ChatGPT by name. In Canada, KPMG published a survey on October 9, 2025, of 3,804 Canadians aged 18 and older, including 684 students, fielded from August 15 to September 15, 2025. It reported that 73 percent of students used generative AI for schoolwork, against 59 percent in 2024 and 52 percent in 2023. Here "student" covers university, college, vocational, and high school attendance. Of the students, 48 percent said their critical thinking had deteriorated since they began using it, and 66 percent said their results improved but they did not think they were learning or retaining as much.
These are opinions of the respondents, from a consulting firm's poll rather than a government statistical program.
What each survey can and cannot show
| Survey | Who answered | Headline figure | Cannot show |
|---|---|---|---|
| Walton Family Foundation-Gallup, 2025 | 2,232 U.S. public K-12 teachers | 32 percent weekly AI use; 5.9 hours saved by weekly users | Measured time savings or student results |
| Walton Family Foundation-Gallup, 2026 | 2,069 U.S. public K-12 teachers | 18 percent received formal guidance | Whether guidance changes outcomes |
| Pew Research Center, fall 2024 | 1,391 U.S. teens aged 13 to 17 | 26 percent used ChatGPT for schoolwork | How the tool was used or whether work improved |
| KPMG in Canada, 2025 | 684 students among 3,804 adult Canadians | 73 percent used generative AI for schoolwork | Use by students under 18, or learning effects |
Questions to ask before quoting a ChatGPT-in-schools statistic
- Does the question name ChatGPT, or does it say AI tools in general?
- Who is in the sample: teachers or students, public schools only, adults only?
- Is the number measured, such as logged use, or self-reported?
- Does the survey have a comparison group, or does it only record opinions?
- Who paid for it, and does the publisher say how the key figure was calculated?
FAQ
Do these surveys show that ChatGPT improves learning?
No. They record how many people use AI tools and what they say about them. Testing learning needs a design that compares students who used a tool with similar students who did not.
Is "six weeks saved a year" a measured result?
It is an estimate. It applies only to teachers who use AI weekly, comes from their own answers, and multiplies 5.9 hours by a 37.4-week school year. Monthly users reported much smaller savings.
Does Statistics Canada track student AI use?
No Statistics Canada release on students' use of AI for schoolwork was found while preparing this page. The Canadian figures here come from a private poll, so they should be read as that poll's results.
Do most teachers get instructions on using AI?
The 2026 Gallup report suggests not. Only 18 percent reported formal guidance from administrators, and nearly half received only informal guidance.
Related reading
- Teaching ChatGPT in a classroom: what works
- OpenAI products for schools and their age rules
- Impact and potential of AI in education
- How AI is used in the learning experience
More in AI in education.
Last verified: October 2026
Sources
- Three in 10 Teachers Use AI Weekly, Saving Six Weeks a Year, Gallup, June 24, 2025 (Walton Family Foundation-Gallup, Teaching for Tomorrow)
- Most Teachers Receive No Formal Guidance on AI Use, Gallup, May 26, 2026 (Walton Family Foundation-Gallup, Closing the Expectations Gap)
- About a Quarter of U.S. Teens Have Used ChatGPT for Schoolwork, Pew Research Center, January 15, 2025
- Generative AI boom among Canadian students raises dilemmas, KPMG in Canada, October 9, 2025