.png)
Over the past few months, a run of education reports kept circling the same question: now that a child can hand schoolwork to an AI, are they learning more, or just finishing faster? A survey on how teens use chatbots. A second one tracking how fast that use is climbing. An OECD review of what it does to understanding. A tutoring study with a sobering result. And, closer to home, a New Brunswick audit of how children are reading. Five reports, one thread. The output is getting easier. The learning is not following on its own.
These reports did not all come out at once. They span from last August to early June, but read together they point somewhere useful for parents. Tools and shortcuts can lift what a child produces today while doing little for what sticks tomorrow, and the help that builds real understanding keeps turning out to be human, skilled, and aimed at one child's actual gap. Here is what each report found, the real numbers, the date it was published, and what it means if you have a child in a Canadian classroom. Every figure links back to its source.
How Are Teens Actually Using AI for Schoolwork?
Pew Research Center put out "How Teens Use and View AI" on February 24, 2026, drawing on 1,458 American teens aged 13 to 17 surveyed in the fall of 2025 (Pew Research Center, 2026). The headline is that AI is now ordinary. Sixty-four percent of teens say they use AI chatbots, about three in ten of them daily. Fifty-four percent have used a chatbot to help with schoolwork, and 57 percent to look something up.
The trickier number is about trust. Fifty-nine percent of teens think using AI to cheat happens at least somewhat often at their school (Pew Research Center, 2026). That does not mean most teens are cheating. It means the norms are still being written, in real time, in classrooms. Pew's sample is American, but Canadian classrooms are seeing the same shift, and Canadian provinces and teachers' federations are working through the same questions about where the line sits.
For a parent, the useful move is not to ban the tool or wave it through. It is to ask a sharper question: what did the AI do, and what did your child do? A chatbot that explained a concept your child then worked through is a study aid. One that produced the finished answer your child copied is something else. The difference is invisible on the page and obvious in conversation.
How Fast Is AI Homework Use Climbing?
If Pew shows where things stand, a RAND study shows how fast they are moving. Drawing on its American Youth Panel and reported on March 20, 2026, RAND found that the share of students using AI for homework jumped in a single school year. Among middle schoolers it rose from 30 percent to 46 percent between May and December 2025. Among high schoolers it climbed from 49 percent to 63 percent over the same months (RAND, 2026).
What has not kept pace is guidance. Only about one in three students said their school had a schoolwide rule on using AI for schoolwork, and the share of students who believe heavy AI use harms their critical thinking actually rose over the same period (RAND, 2026). Students are adopting the tools faster than the adults around them are setting expectations, and many of the students themselves are uneasy about it. These are self-reported figures, so treat the exact numbers as a snapshot rather than a precise census, but the direction is hard to miss.
The practical takeaway is a household norm, set early. Children are not waiting for permission to use these tools, so the useful conversation is not whether they will, but how. A simple family rule, AI can help you understand and check, but you do the thinking and you can explain your answer, does more than a ban a child will quietly route around.
Does AI Help Students Actually Learn, or Just Finish?
The OECD took the longer view. Its Digital Education Outlook 2026, published in January, reviewed the research on generative AI in classrooms and landed on a careful distinction (OECD, 2026). Students using general-purpose AI tools often produced higher-quality work. But that advantage tended to fade, and sometimes reverse, once the tool was removed, for example on a closed-book exam. Producing a good answer with AI, in other words, is not the same as having learned the thing.
The OECD's explanation is intuitive once you see it. When a chatbot does the hard parts, the steps where learning actually happens get skipped: noticing what you do not understand, testing your reasoning, and revising when you are wrong. The report does not argue against AI in schools. It argues for the right kind, recommending purpose-built educational tools designed with teachers and grounded in how children learn, over off-the-shelf chatbots used as shortcuts.
For families, the lesson lines up with the classroom one. AI is most useful when it makes a child do more thinking, not less: explaining a wrong answer, generating practice questions, or checking reasoning step by step. Used that way, it is a study partner. Used as a finisher, the gain disappears the moment the child has to perform on their own.
Does Tutoring Actually Help Kids Catch Up?
The most sobering result of the set was about tutoring, the very thing schools leaned on hardest after the pandemic. An analysis from the University of Chicago Education Lab and MDRC, covered by The Hechinger Report on August 25, 2025, followed more than 20,000 students, with about 9,000 assigned to tutoring and 11,000 not, across several US cities during the 2023-24 school year (Hechinger Report, 2025). The average result was modest: only one or two months of extra learning in reading or math, a fraction of the up-to-a-year gains that high-dosage tutoring produced in pre-pandemic studies.
Read quickly, that looks like tutoring stopped working. It did not. The researchers found that each minute of tutoring was still as effective as before. What fell short was dosage and consistency. Many post-pandemic programs simply did not deliver enough sessions, with enough regularity, from people prepared to teach them. The promise of tutoring is real, but it depends entirely on quality and frequency, not on the label.
That is the through-line connecting this study to the AI findings. Help only helps when it is good, regular, and aimed at the right thing. A child who meets a skilled tutor twice a week, on the specific skill they are missing, is in a different situation from one who logs a few scattered sessions or leans on a chatbot for answers.
What a New Brunswick Literacy Audit Reminds Us
The freshest report, and the closest to home, is a reminder that the fundamentals still need steady attention. On June 2, 2026, New Brunswick's Auditor General, Paul Martin, released an audit of how the province measures and supports student literacy. Its central finding was blunt: no provincial literacy assessment target was met in any year of the audit period (Auditor General of New Brunswick, 2026).
The audit was careful about why. It found that while literacy data was collected, it was not always monitored or acted on consistently, which can mean some students do not get extra support in time to matter. This is a measurement-and-follow-through finding, not a verdict on any teacher or any child, and literacy gaps are a Canada-wide challenge, not a New Brunswick one. The point for a parent is the timeline, not the blame.
Reading develops on a clock. A year spent waiting for a gap to be noticed and addressed is a long time in a young reader's life, which is exactly why early, targeted help matters. If a child is struggling to read at grade level, the constructive move is to keep momentum going now, on the specific skill that has stalled, rather than waiting for a system-wide number to improve.
What This Means for Ontario Families and Teachers
Put the five together and they tell one story. Teens have folded AI into schoolwork, and they are doing it faster than schools can set rules. The OECD says the tools lift output without guaranteeing learning. The tutoring research says even direct help only works when it is frequent and skilled. And the New Brunswick audit is a reminder that reading gains come from steady, early attention, not from waiting. None of this points to more tools or more hours. It points to the right help, aimed at the one thing a child is actually missing, from someone trained to spot it and teach it.
That is the work Connect Education is built for. Connect Education works exclusively with Ontario Certified Teachers and ECEs in good standing, who build a learning plan around your child's real gaps and meet online, in person, or at a public library, and who use tools to deepen understanding rather than replace it. The table below sets that next to a typical tutoring option. If your child is leaning on shortcuts or stalling on a specific skill, you can book a free 15-minute assessment or browse certified tutors to get started.
Questions Parents Ask Most
Does using AI for homework hurt learning?
It can, if it does the thinking for the child. The OECD's Digital Education Outlook 2026 found that students using general-purpose chatbots often produced higher-quality work in the moment, but that advantage tended to fade or reverse once the tool was taken away, such as on a closed-book exam. The tools are not the problem. The risk is skipping the steps where learning actually happens, like working out what you do not understand and reasoning through it. Used as a study partner that makes a child explain their thinking, AI can help. Used as a finisher, it tends not to.
Is it cheating if my child uses ChatGPT for homework?
Using a chatbot is not automatically cheating. The line is whether your child does the thinking or hands it off. A 2026 Pew Research Center survey found 54 percent of teens have used chatbots for schoolwork, and 59 percent believe AI-assisted cheating happens at least somewhat often at their school, so norms are still forming. A practical household rule helps: AI can explain a concept, check work, or suggest an approach, but the child writes the answer and can explain how they got it. Ask your child to walk you through their reasoning, not just show the result.
Does tutoring actually help students who are behind?
Yes, but only when it is frequent, consistent, and skilled. A large 2025 study of post-pandemic tutoring across more than 20,000 students found only one to two months of extra learning on average, far below the up-to-a-year gains seen before the pandemic. The researchers traced the gap to dosage and quality, not to tutoring itself: each minute of good tutoring was still effective, but many programs did not deliver enough of it. Tutoring works best when it is regular, taught by someone qualified, and aimed at the specific skill a child is missing.
How can I help my child build real understanding instead of just getting answers?
Focus on the thinking, not the output. Build short, regular practice into the week on the one or two skills your child is missing, and ask them to explain their reasoning out loud. Treat AI as a tool that has to make them show their work, not one that hands over the answer. If the gap is specific, such as fractions or reading fluency, a certified teacher can pinpoint it and rebuild it directly, which is more durable than general homework help or a quick AI answer.
Sources
- Office of the Auditor General of New Brunswick (2026), report on student literacy (released June 2, 2026), via Government of New Brunswick, gnb.ca
- Pew Research Center (2026), How Teens Use and View AI (February 24, 2026), pewresearch.org
- RAND Corporation, American Youth Panel (2026), on students using AI for homework, reported by K-12 Dive (March 20, 2026), k12dive.com
- OECD (2026), Digital Education Outlook 2026 (January 2026), reported by Digital Watch Observatory, dig.watch
- The Hechinger Report (2025), Proof Points: post-pandemic tutoring results (August 25, 2025), hechingerreport.org
