Findings

How students actually use AI

2,000 responses · Data last updated 20 Aug 2026

The PDF is the version for teachers and judges. This page is the one to send to everyone else.

Want to see where you land in all this? Answer the five questions — it takes about a minute, and nothing you answer is stored.

Collection has passed the publication threshold. The write-up is being drafted.

There are 2,000 responses in the dataset. Nothing is written here yet — the charts below are the raw results, published without interpretation.

Set NEXT_PUBLIC_FORM_URL to the Form’s public /viewform link to show a link back to the survey here.

What the data shows so far

Published without commentary. The written analysis follows once collection closes.

How have you learned to use AI?

2000 students

35% taught themselves; 13% learned in school or an educational workspace.

How dependent are you on AI for your school/exam prep?

2000 students

Not at allVery dependent

Mean 6.5, median 7 out of 10.

What stops you from getting better at AI?

2000 students

Methodology

How responses were collected: A self-completed Google Form of 15 questions, filled in online and unsupervised. Respondents were not paid; the form offered a free AI toolkit on completion, which may have drawn people already interested in the subject.
How it was distributed: Through personal and school networks — shared directly and passed on by recipients. This is a convenience sample, not a random one. It skews toward a particular kind of school and a particular set of cities, and toward students socially connected to the people who distributed it. Results describe the people who answered. They do not generalise to students nationally.
Sample size: 2,000 responses in total. Every chart on this site shows the number of responses behind that specific question, which is not always the total: dependence is the one question the Form does not require, so its base is smaller. Where a group is too small to support a comparison, the site compares against the whole sample instead and says so on the chart.
Collection window: Collection is still open; the window will be stated here once it closes.
Privacy: Respondents are largely 15–17. Email addresses and free-text answers are never read by this site. An offline script aggregates the responses to counts and publishes only those; the response sheet is never fetched by your browser, and no individual response can be recovered from what is published here.
Comparison method: Your answers are never matched against an individual respondent. For each question independently, your answer is placed within the distribution of all answers to that question. Percentiles use the proportion of responses below yours plus half of those equal to it.

Questions about the method? Take the quiz to see how the comparison works on your own answers.