The clearest measured benefit of AI in schools is teacher time, and the clearest drawbacks are thin training and unsettled effects on student thinking. A Gallup survey of 2,232 US public school teachers, fielded in March and April 2025 and reported in June 2025, found that those who use AI tools at least weekly estimate saving 5.9 hours per week, while only 19 percent of teachers said their school has an AI policy. The time figure comes from teachers' own estimates, and the survey did not measure what students learned.
Key takeaways
- Teachers who use AI weekly estimate 5.9 hours saved per week, and monthly users estimate 2.9 hours.
- Among teachers who use AI for a task, 64 percent report better quality when adapting student materials and 57 percent when giving feedback and grading.
- Sixty-eight percent of teachers had no school or district training on AI, and 52 percent taught themselves.
- Teachers named independent thinking and critical thinking as the areas most likely to be hurt, and 40 percent of teachers favor AI in K-12 schools while 28 percent oppose it.
- The survey covers US public schools only. No equally detailed Canadian teacher survey was verified for this article.
What the survey measured and how
The study, titled Teaching for Tomorrow, was produced by Gallup for the Walton Family Foundation. The full report describes a web survey run from March 18 to April 11, 2025, with teachers drawn from the RAND American Teacher Panel, a probability-based panel of US public K-12 teachers. The sample was weighted to national teacher and school characteristics, and the margin of sampling error is plus or minus 2.5 percentage points for the full sample and larger for subgroups.
Sixty percent of the teachers had used an AI tool for work during the 2024-25 school year. Use ran higher among high school teachers (66 percent) and early-career teachers (69 percent). Thirty-two percent used AI at least weekly, 28 percent used it less often, and 40 percent did not use it at all. Preparing to teach was the most common monthly use (37 percent), followed by making worksheets (33 percent).
Benefit one: time that teachers say they get back
Gallup calls the time saved an "AI dividend." Each teacher who used AI for a task at least monthly said whether it saved time, took more time, or made no difference, and then estimated hours to the nearest half hour. Net hours across six tasks were averaged, with extreme answers trimmed. For weekly users the average was 5.9 hours per week. Multiplied by an average of 37.4 contracted weeks and set against a 37.5-hour week, Gallup reports that figure as about six weeks per school year.
The arithmetic is transparent, but the inputs are recollections. A teacher who guesses high or low shifts the average, and the six-week figure applies only to weekly users, who are the 32 percent most comfortable with the tools. Teachers also described what they did with the time: more detailed feedback, individualized lessons, parent emails, and leaving work on time.
Benefit two: quality of everyday teaching work
Teachers who used AI for a task were asked whether quality changed. Sixty-four percent reported better quality in the materials they modified to meet student needs, 61 percent reported better insight into student learning or achievement data, and 57 percent reported better feedback and grading. The report says 16 percent or fewer saw lower quality on any task. Separately, 57 percent of teachers agreed that AI will improve the accessibility of learning materials for students with disabilities, and 65 percent of special education teachers agreed.
These are teacher judgments about their own work. Nobody compared the AI-assisted worksheets or feedback against a control group, so the numbers describe perceived quality, not demonstrated learning.
Ledger of reported gains and costs
| Item | Reported figure | What it does not show |
|---|---|---|
| Weekly users' time saved | 5.9 hours per week | Self-estimate; excludes the 68 percent who use AI less than weekly |
| Monthly users' time saved | 2.9 hours per week | Gains depend on frequency of use |
| Quality of adapted materials | 64 percent report higher quality | No independent quality rating |
| School AI policy | 19 percent of teachers have one | Most schools have no stated rules |
| Training received | 68 percent had none from school or district | Self-taught use varies in quality |
| Expected effect on grades | 32 percent expect an increase, 10 percent a decrease | Opinions, not outcomes |
Drawbacks: training gaps, policy gaps, and thinking skills
Teachers at schools with an AI policy were more likely to have used AI (70 percent versus 60 percent), and those schools showed a dividend of 2.3 hours per teacher per week against 1.7 hours without a policy. Because most schools lack a policy and most teachers lack training, the report warns that gaps between teachers with and without AI skills could widen.
On student outcomes, teachers were split or unsure. At least four in 10 were neutral or did not know the effect on engagement, motivation, grades, and readiness for college and work. Independent thinking and critical thinking were the top two areas where teachers predicted negative effects. The report also cites a companion Gallup study in which 49 percent of Gen Z respondents said AI will hurt their ability to think about information carefully, and 35 percent of high school students said most or all classmates use AI when they are not supposed to.
Opinion tracks experience. Among teachers who had used AI at least once, 56 percent favored it in K-12 schools, compared with 16 percent of those who had not. Whether that gap reflects what the tools do or who chooses to try them cannot be settled from a single survey.
Questions to ask before a school adopts an AI tool
- Which tasks would the tool take over, and who will check its output?
- Is there a written policy, and does it say what student data may enter the tool?
- What training is planned, given that most teachers report none?
- How will the school learn whether students are learning more, and not just whether teachers feel faster?
- Does the evidence come from a setting like this school, in this country?
FAQ
Does AI in schools improve student learning?
This survey does not answer that. It measures teacher time and teacher opinion. Student learning needs controlled studies, and the results so far depend heavily on how the tool is designed and supervised.
Do the findings apply to Canada?
Not directly. The sample is US public school teachers, and provinces set their own education rules. Canadian readers can treat the figures as one data point from a neighboring system, not a measure of Canadian classrooms.
Is 5.9 hours a week typical for any teacher?
No. It is the average for teachers who use AI at least weekly. Monthly users estimated 2.9 hours, and 40 percent of teachers surveyed did not use AI at all.
Why do teachers who use AI view it more favorably?
Possible reasons include direct experience, self-selection, and the time savings themselves. The report says the cause is likely multifaceted and that benefits and drawbacks will become clearer as more teachers adopt the tools.
Related reading
- Documented challenges and limits of AI in classrooms
- Impact and potential of AI in education
- What implementing AI actually costs
- How AI changes the learning experience
More in AI in education.
Last verified: October 2026
Sources
- Teaching for Tomorrow: Unlocking Six Weeks a Year With AI, Gallup and Walton Family Foundation, June 2025
- Voices of Gen Z: How American Youth View and Use Artificial Intelligence, Gallup and Walton Family Foundation, 2025 (figures as cited in the report above)