AI Tutors, Algebra Gaps and Youth in Care

Give 2,379 university students an AI tutor built right into their course, and you might expect their grades to rise. In a new randomized trial at the University of Maryland, they fell (Liu et al., 2026).

This digest gathers five pieces of research and reporting published between September 30 and October 21, 2026. Two are Canadian: a CBC News look at why experts want math taught with more structure, and a CBC investigation into how few children in government care finish high school. Three come from the United States: the AI tutor trial, a national study of students who were ready for Grade 8 algebra but never took it, and a new report on how few school districts have recovered since the pandemic. The thread running through all five is the same: access to help is not the same as getting it. A tool, a course or a program only works when a student actually uses it, with the right footing underneath.

Does an AI Tutor Help Students Learn? A Large Trial Found the Opposite

Researchers at the University of Maryland, College Park, led by Jing Liu, ran what they describe as the first large randomized trial of AI tutoring across many subjects at a US public university. In the fall semester of the 2025-26 school year, 30 instructors teaching sections in STEM, the social sciences, the humanities and education were randomly assigned to give their students access to a GPT-4o study assistant built into the course website. It drew on materials the instructor chose. The paper is dated September 30, 2026 and was released as Annenberg Institute EdWorkingPaper 26-1598 in October (Liu et al., 2026).

Among sections of the same course, access to the tutor lowered final grades by 0.37 standard deviations and cut recorded participation in the course's online activities, such as discussion posts and quizzes, by 0.90 standard deviations. Homework results showed no gain. Estimated losses were larger for first-generation students. In a survey, students who used the tutor were more likely to say they asked their instructor fewer questions about course content (Liu et al., 2026). The authors' reading is that the tutor may have replaced the effort and contact that normally drive learning, rather than adding to them.

The caveats are real. Only about 15 per cent of students in the tutor sections used it at least once, so the result describes what happened when a class was offered the tool, not what happens to every student who uses one. The outcome is course grades, not an independent test, and the student survey had an 18 per cent response rate. These were university students, not children. Still, the finding lines up with something families see at the kitchen table: a tool that hands over answers can quietly take the place of the thinking. Our post on how students themselves view AI shows many teens already notice it.

How Should Math Be Taught? Canadian Experts Make the Case for Foundations

On October 3, 2026, CBC News reporter Jessica Wong looked at a question that recent provincial and international results keep raising: whether the way math is taught needs another look. The piece notes that the slide in Canadian students' math skills predates the pandemic, and that some experts want numeracy to get the same attention reading has received in recent years, when many provinces moved toward explicit, phonics-led literacy teaching (CBC News, 2026).

John Mighton, the mathematician who founded the education charity JUMP Math in 2002, told CBC that much instruction over the past 20 years has treated children as experts before they were ready, starting with complex, wordy problems. His alternative is more guidance, more scaffolding and more practice, with regular low-stakes checks, and he is clear that this is not a return to rote learning. Daniel Ansari, a neuroscientist and professor of psychology and education at Western University, made a similar point: asking young learners to explain everything at once can overload them and keep new ideas from reaching long-term memory. He suggests breaking math into smaller steps, and he sees a place for timed fact practice once children have had time to learn the facts first (CBC News, 2026).

The story also shows what this looks like in a school board. In Ontario's Upper Grand District School Board, a multi-year plan added central math facilitators, new teaching resources, an early math screener, updated training for Grade 1 and Grade 3 teachers and, in the schools with the greatest need, support teachers who deliver focused tutoring. The board's math lead told CBC it saw improvement in its latest EQAO results. This is reporting, not a controlled study, and the board's results cannot be credited to any one change. For parents, the practical takeaway is the one both experts share: number sense comes first. If a child can count to 10 but cannot say what 10 means, later topics will feel shaky. Our guide on helping a child who hates math has ideas for building that base at home.

Ready for Algebra but Not Enrolled: Why Course Placement Matters

On October 9, 2026, The 74 reported on two studies about a quiet gap in middle school math. The first, a national study by the testing organization NWEA using a sample of 55,000 students from 29 states, found that in 2025 roughly one in four seventh graders had spring scores at or above a benchmark for algebra readiness. Yet almost a quarter of those ready students were not taking algebra in Grade 8. The study found that students with stronger math backgrounds did better in algebra when they took it in Grade 8 rather than Grade 9 (The 74, 2026).

The second study, from the New York Equity Coalition convened by The Education Trust-New York, found that 58 per cent of New York seventh graders scored proficient on the state's Grade 7 math test in 2023-24, but only 37 per cent of eighth graders in schools serving both grades were enrolled in Algebra I. The gaps were wider for some groups. Only 43 per cent of students from low-income families were proficient, compared with 71 per cent of more affluent peers, and among Asian students, 75 per cent were proficient but 14 per cent were enrolled. Researchers quoted in the story disagree on the fix. Some favor automatic enrolment for every student who qualifies; others warn that placing students in algebra regardless of readiness has not raised achievement on its own (The 74, 2026).

These are US findings, and Canadian systems handle middle school math differently. Ontario, for example, now has a single de-streamed Grade 9 math course, so the key choices come a little later, when students choose the senior courses that keep university, college and apprenticeship paths open. The useful lesson travels well: a student can be ready for a harder course and still miss it because nobody asked. A short conversation with the teacher about what your child has mastered helps. Our guide to what Grade 9 math actually tests in Ontario explains the starting point. NWEA's research also found that students who are not yet ready do no better by simply waiting a year. They need targeted help, not just more time.

Why Do So Many Children in Care Not Finish High School?

On October 16, 2026, CBC News reporter Shaina Luck published a story in the Atlantic Investigative Unit's series on children in government care. Jacquie Gahagan, associate vice-president of research at Mount Saint Vincent University and lead of the Atlantic Tuition Waiver project, said fewer than 50 per cent of children from care in Canada finish high school, and fewer than 10 per cent of former youth in care go on to finish post-secondary education. By comparison, data from Nova Scotia and Statistics Canada show an overall high school graduation rate above 90 per cent (CBC News, 2026).

Tuition waivers for former youth in care began appearing at Atlantic institutions around 2020, and at least 45 students started post-secondary programs in Nova Scotia this fall with one. Gahagan's research finds the waivers help, but that students also need mental health support, mentoring and sometimes help with food or housing. Nova Scotia's Path program, launched in 2024, gives former youth in care biweekly financial support up to age 24. Provincial briefing notes released through access-to-information requests said that in the 10 years before 2021, 730 young people aged out of care in Nova Scotia, and nearly 60 per cent sought income assistance at some point (CBC News, 2026).

The national figures are given as approximate thresholds, and better outcome data is one of the changes researchers are asking for. The point that matters for teachers and caregivers is about stability during school. Gahagan described how each move between placements can mean a new school, a new peer group and a race to catch up on missed work. The provincial department responsible told CBC it works with the education system on these transitions, including a consultant role shared with the Department of Education to help children move between schools. Consistent, one-to-one academic support that follows the child, whatever the school, is one practical way to keep learning on track through those moves.

Only 1 in 10 US Students Is in a District That Has Recovered

On October 21, 2026, the Center on Reinventing Public Education (CRPE), a research organization based at Arizona State University, released its fifth annual State of the American Student report. It found that only about 1 in 10 US students attends a school district that has returned to pre-pandemic achievement levels in both reading and math. Drawing on sources including the Harvard and Stanford Education Scorecard, the report estimates lifetime earnings losses nearing 9 per cent for students in the states hardest hit by learning loss, compared with about 2 per cent in the least affected states (The 74, 2026).

The report also looks at state policy. It counts 43 states with science of reading laws and 18 with laws on effective math instruction, but finds that far fewer have built the training and accountability needed to put those laws into practice. Only one state, Maryland, requires AI tools to be checked for their effect on learning, and no state's attendance has returned to pre-pandemic levels, although Alabama and Iowa have made gains against chronic absenteeism (The 74, 2026).

This is a US report, built largely on US test data, so it says nothing directly about Canadian districts. Its central message still applies here. A law or a strategy on paper is the start, not the finish; what changes outcomes is the teaching that reaches each student. CRPE's own summary puts it plainly: the issue is not only the pandemic, and returning to the old normal will not be enough. Recovery happens one child at a time.

What This Means for Families and Teachers

Put side by side, this week's research describes help that was available but did not land. An AI tutor sat in the sidebar of a course and grades went down. Students qualified for algebra and sat in a different class. Tuition is waived for youth leaving care, yet fewer than half finish the high school diploma that would let them use it. In each case, what made the difference was a person paying attention: an instructor students kept asking questions, a teacher who explained the next course, a steady adult across school moves.

For families, that suggests three small habits. Ask how your child is using AI, and whether it explains steps or simply supplies answers. Check that math basics are solid before a harder course, and ask the teacher directly about readiness. And if your child has changed schools or fallen behind, look for support that stays consistent. Every tutor on Connect Education is a licensed teacher in good standing, and families can book a free 15-minute assessment or browse certified tutors by subject and grade.