mytetz › Study guides › How students study now
How students study now
Two recent surveys measure how students use an AI tool. Pew found that 54% of US teens have used an AI chatbot for help with schoolwork. HEPI found that 95% of UK undergraduates use AI in at least one way. Each number describes one narrow group. Read the sample before you read the percentage.
Key points
- Pew: 54% of 1,458 US teens aged 13 to 17, measured in autumn 2025.
- HEPI: 95% of 1,054 full-time UK undergraduates, polled in December 2025.
- In the US, 28.2% of postsecondary students took some courses at a distance in autumn 2024.
- Nobody reliably measures the share of study on a telephone, or the average length of a study session. This page states no number for either one.
What do the recent surveys measure?
Each row below names one measurement, one population and one date. A percentage without those three items tells you nothing.
| Measurement | Figure | Population and sample | Date of the data |
|---|---|---|---|
| US teens who have used an AI chatbot for schoolwork | 54% | 1,458 US teens aged 13 to 17. Margin of error 3.3 points. | 25 September to 9 October 2025 |
| UK undergraduates who use AI in at least one way | 95% | 1,054 full-time UK undergraduates. Margin of error about 3 points. | December 2025 |
| UK undergraduates who use generative AI for assessed work | 94% | The same 1,054 undergraduates. | December 2025 |
| US students who took some courses at a distance | 28.2% | 5,547,429 of 19,659,633 students at all US degree-granting institutions. | Autumn 2024 (provisional) |
| Registered learner accounts on one large platform | 197 million | Cumulative registered accounts on Coursera, up from 168 million. This is not a count of active learners. | 31 December 2025 |
How much of study now involves an AI tool?
The headline number hides the useful detail. Pew asked US teens how much of their schoolwork they do with help from a chatbot. The answers split like this:
| Share of schoolwork | Percentage of teens |
|---|---|
| All or most | 10% |
| Some | 21% |
| A little | 23% |
| None | 45% |
So "54% use AI" and "10% lean on it heavily" describe the same group. The first number sounds like a transformation. The second one describes it better.
The UK picture holds one useful change. In the HEPI survey, the share of undergraduates who use AI to generate text fell from 64% to 56% between the two years. Use of AI did not fall. The way students use it moved.
How many students study at a distance?
In autumn 2024, 5,547,429 of 19,659,633 US postsecondary students took some but not all courses at a distance. That is 28.2%. The figure comes from the federal IPEDS collection, and it is provisional.
A platform count is a different measurement, and it answers a different question. Coursera reported 197 million registered learner accounts on 31 December 2025, up from 168 million. That is a cumulative count of accounts. It does not say how many people studied last month.
What does nobody measure?
Three figures appear often on the web, and we found no reliable source for any of them. We state none of them.
- The share of study that happens on a telephone. No survey by a named institution measures it.
- The average length of a study session. No reliable measurement exists.
- A 2026 completion rate for an online course. Figures such as "15 to 20% for a free course" circulate widely. Every trail we followed ended at a site with no named primary source. The guide on why a person stops an online course gives the one measurement that does have a source.
How do you read a study statistic?
- Find the population. "Students" is not a population. "1,054 UK undergraduates" is.
- Find the date of the data, and not the date of the article.
- Find the exact question. "Have used" and "use every week" are different measurements.
- Find the denominator. A rate over registrants differs from a rate over active users.
- Follow the link to the primary source. Stop trusting a number that has no link.
Sources
- Pew Research Center (24 February 2026). How Teens Use and View AI. Survey of 1,458 US teens aged 13 to 17, fielded 25 September to 9 October 2025. pewresearch.org
- Stephenson, R., & Armstrong, C. (March 2026). Student Generative AI Survey 2026. HEPI Report 199, with Kortext. 1,054 full-time UK undergraduates, polled in December 2025. hepi.ac.uk
- National Center for Education Statistics. Digest of Education Statistics, table 311.15, prepared March 2026. Autumn 2024 IPEDS data. nces.ed.gov
- Coursera, Inc. Fourth quarter and full year 2025 results, 5 February 2026. investor.coursera.com
Common questions
Do these numbers describe students in my country?
Probably not. One survey covers US teens, and one covers UK undergraduates. The distance figure covers US institutions only. Treat each one as a local measurement.
Does an AI tool improve a result?
These surveys measure use, and not outcome. They tell you what students do. They do not tell you whether it helps.
Why does this page hold so few numbers?
Because few numbers survive a check. We state a figure only with a named source, a sample size and a date. We leave the rest out.
How often do you update this page?
We review it when a named institution publishes new data. The review date sits at the foot of the page.
Where do you go next?
- Why a person stops an online course — the completion data, and what it counts.
- How to study on your own — what to do with the time, whatever tool you use.
Start with a topic
A statistic is easier to judge when you understand the method behind it. These three topics cover that ground.
Open the catalogue