AI Vanguardest. 2024
Vanguard Open judging underway · Results October 3About the Open →
Research · 2025 Student AI Policy Survey

What did 447 students say about AI at school?

Students are not asking for a ban. Four in five already use AI for schoolwork, nine in ten think it is acceptable with conditions, and the most common thing they want from their schools is to be taught how to use it well. Every figure on this page comes with its count, its denominator, and the exact question that produced it.

447 responses · 6 schoolsFielded Aug 25 – Dec 17, 2025Convenience sample · Southern CaliforniaMethods updated September 2026
Findings

Four numbers, with their receipts.

All percentages are shares of all 447 submitted responses, rounded to the nearest whole number. Each one lists the exact question and which response options were counted.

  1. 80%
    Used AI for schoolwork
    357 of 447
    The question, as asked
    “Have you used AI tools (e.g., ChatGPT, Grammarly, Khanmigo, QuillBot) for schoolwork?”
    Counted as yes
    “Yes, frequently (weekly or more)” (84) + “Yes, occasionally (a few times)” (273)
    Everyone else
    52 had heard of AI tools but not used them, 24 had never used them, and 14 wrote in a custom answer, most describing limited or teacher-directed use.
  2. 87%
    Find AI helpful
    389 of 447
    The question, as asked
    “How helpful do you find AI tools for your learning?”
    Counted as yes
    “Very helpful” (182) + “Somewhat helpful” (207)
    Everyone else
    35 said not helpful, 7 said harmful, and 16 wrote in an answer, mostly non-users.
  3. 90%
    Say AI is acceptable
    404 of 447
    The question, as asked
    “Do you think using AI for schoolwork should be considered…”
    Counted as yes
    “Acceptable, as long as it’s used responsibly” (222) + “Acceptable only for certain tasks” (182)
    Everyone else
    16 (4%) called it cheating even if just for help, 22 were not sure, and 5 wrote in an answer.
  4. 74%
    Want schools to teach responsible use
    330 of 447
    The question, as asked
    “What role should schools play in guiding AI use?”
    Counted as yes
    “Teach students how to use AI responsibly”, in a select-all-that-apply question
    Everyone else
    Students could choose more than one role, so the options below total more than 100%.
What students want schools to do

The dominant preference isn't a ban. It's guidance.

“What role should schools play in guiding AI use?” was select-all-that-apply, so the bars total more than 100%. Only 22 students (5%) wanted AI to do most of the work; 158 (35%) asked to help write the rules themselves.

  1. 01
    Teach students how to use AI responsibly
    74%n = 330
  2. 02
    Involve students in shaping AI rules and policies
    35%n = 158
  3. 03
    Create stricter rules to limit misuse
    26%n = 116
  4. 04
    Allow free use with minimal restrictions
    24%n = 106
  5. 05
    Do nothing; AI use should be up to individuals
    14%n = 63
  6. 06
    AI should complete most of the work
    5%n = 22
Share of all 447 responses selecting each option. Source: AI Vanguard 2025 Student AI Policy Survey, Aug 25 – Dec 17, 2025. Downloadable SVG and CSV above carry the same values.

Horizontal bar chart of what 447 students said schools should do about AI. Teach students to use AI responsibly, 74% (330). Involve students in shaping AI rules, 35% (158). Create stricter rules, 26% (116). Allow free use with minimal restrictions, 24% (106). Do nothing, 14% (63). AI should complete most of the work, 5% (22). Select-all-that-apply, so bars total more than 100%. AI Vanguard 2025 Student AI Policy Survey.

Equal access

Two in three say access is equal. One in five disagree.

“Do you feel all students at your school have equal access to AI tools?” Single choice. The 18% who said some students lack access is the number our equity commitment is built on.

  1. Yes, most students have equal accessn = 29666%
  2. No, some students do notn = 8218%
  3. Don’t known = 6715%
  4. Othern = 20%
Who responded

Six schools, two of them almost entirely.

This is a convenience sample: students reached through AI Vanguard representatives at their own campuses, not a random draw. 96% of responses came from two schools. The findings describe these students, not all students.

By school
  1. Downey High Schooln = 24755%
  2. Cerritos High Schooln = 18141%
  3. Cypress High Schooln = 164%
  4. Beckman High Schooln = 10%
  5. Sunny Hills High Schooln = 10%
  6. Troy High Schooln = 10%
By grade
  1. Freshman (9th)n = 14031%
  2. Sophomore (10th)n = 13330%
  3. Junior (11th)n = 12528%
  4. Senior (12th)n = 4911%
Methods and limitations

How the survey was run, and what it can't tell you.

Instrument

Google Form, nine items, distributed by AI Vanguard student representatives at their campuses. Nine items: eight structured questions and one optional comment field. The exact wording of every question is listed below.

Recruitment and field period

Responses were collected from Aug 25 – Dec 17, 2025. Representatives shared the form at their own schools, so participation depended on where AI Vanguard had a representative and how actively the form circulated there.

Sample

A convenience sample of high school students in Southern California. It is not representative of any school, district, or state. Two schools account for 96% of responses; seniors are under-represented (11%).

Denominators and rounding

Every percentage divides by all 447 submitted responses, including write-in answers, and is rounded to the nearest whole number. The school-role question was select-all-that-apply, so its options total more than 100%.

Duplicates

12 rows are exact duplicates of another response (same school, grade, and answers, different timestamp), most likely accidental double submissions. Removing them (n = 435) changes no headline figure by more than half a percentage point, so the published analysis keeps all 447 and discloses this here.

Corrections

In September 2026 the first headline figure was corrected from 84% to 80%. The earlier number was the count of frequent users (84) mistaken for a percentage. No other figure changed.

Privacy

No names or emails are published. Quotes from the comment field are attributed by grade and school only. Student-level records are not released.

Analysis and authorship

The survey was designed and distributed by AI Vanguard and analyzed by AI Vanguard. The analysis script is public in our site repository, so anyone can check the counts against the aggregate method.

Every question, as worded on the form
  1. 1What school do you attend?
  2. 2Grade level
  3. 3Have you used AI tools (e.g., ChatGPT, Grammarly, Khanmigo, QuillBot) for schoolwork?
  4. 4What have you primarily used AI tools for? (select all that apply)
  5. 5How helpful do you find AI tools for your learning?
  6. 6Do you think using AI for schoolwork should be considered…
  7. 7Do you feel all students at your school have equal access to AI tools?
  8. 8What role should schools play in guiding AI use?
  9. 9Comments or request for policy advocation? (optional)

Analysis script: scripts/analyze-survey.mjs on GitHub ↗

For educators and journalists

Use it. Cite it.

Findings, the chart, and the table may be reproduced with attribution to AI Vanguard and a link to this page. For the underlying aggregate counts, a student source, or a founder interview, use the contact form; we answer.

Suggested citation
AI Vanguard. (2026). 2025 Student AI Policy Survey: what 447 students said about AI at school. https://aivanguard.org/research/student-ai-survey
Quotable, organizational
“Students are not asking for a ban. They are asking to be taught, and to be consulted.”AI Vanguard, on the 2025 Student AI Policy Survey. An organizational statement, not a recorded interview.
Downloads
Chart (SVG) · Table (CSV) · Policy briefChart alt text: Horizontal bar chart of what 447 students said schools should do about AI. Teach students to use AI responsibly, 74% (330). Involve students in shaping AI rules, 35% (158). Create stricter rules, 26% (116). Allow free use with minimal restrictions, 24% (106). Do nothing, 14% (63). AI should complete most of the work, 5% (22). Select-all-that-apply, so bars total more than 100%. AI Vanguard 2025 Student AI Policy Survey.
Permission
Reproduce freely with attribution. Please do not describe the sample as national or representative; it is neither. Contact: info@aivanguard.org.

This survey is why we now study whether student voice changes anything.