Many aren't yet, and the gap is fixable. In a KPMG Canada survey of people 18 and over, published in September 2026, 75% of students said they use generative AI for schoolwork, but only 53% said they had received formal instruction on using it responsibly, and 59% of educators said the education system isn't preparing students for a workplace where AI is a standard tool. Students have the tool. What many lack is practice using it with judgment, and proof that they can.
Colleges and universities can close that gap without rebuilding their programs. Here is what the data says, what employers now ask for and where to start.
What KPMG Canada found in 2026
KPMG Canada surveyed 1,009 students and educators aged 18 and over at Canadian universities, colleges, vocational schools and high schools between July 22 and August 5, 2026, and published the results on September 4. The 75% usage figure is based on the students in its full screening sample of 1,864 people.
| Finding | Share |
|---|---|
| Students who use generative AI for schoolwork | 75% |
| Students who say they received formal instruction on responsible AI use | 53% |
| Educators who say students get enough education on responsible use at their institution | 46% |
| Educators who say the system isn't preparing students for an AI workplace | 59% |
| Educators who consider using generative AI for schoolwork cheating | 72% |
| Students who agree it's cheating | 50% |
Two things stand out to me. First, use itself is no longer the question. KPMG's October 2025 release put student use at 73%, up from 59% in 2024 and 52% in 2023. Second, the cheating split is as much a policy problem as a values one. Educators are right to worry about shortcuts. But when nearly three in four of them call it cheating and half of students don't, students can meet different rules from one course to the next, and the likely result is hiding AI use instead of getting good at it.
Students see the cost too: 51% said their critical thinking has declined since they started using AI. Schools are adapting, with 77% of educators saying AI has already changed how they assess learning. The open question is whether that work reaches the career outcome.
What employers now expect from new graduates
- In Canada, 80% of hiring managers look for at least basic AI skills, and 34% call AI proficiency very important or absolutely essential, according to an Express Employment Professionals and Harris Poll survey of 508 Canadian hiring decision-makers, fielded in May and June 2026.
- In the United States, NACE reported in July 2026 that employers in its spring survey, 185 in all, said more than one third of their entry-level jobs require AI skills, nearly triple the share in its fall survey six months earlier.
- PwC's 2026 Global AI Jobs Barometer, released in June 2026, analysed 2.4 million US entry-level job postings. The entry-level roles most exposed to AI were seven times more likely to ask for traditionally senior, human-intensive skills such as leadership, creativity and face-to-face interaction. Job openings for entry-level roles that ask for these senior skills grew 35% since 2019, while other entry-level roles shrank 10%.
The detail I'd put in front of every program head comes from the Express survey. When Canadian hiring managers named the AI capabilities they want, writing usable prompts came last of the five listed, at 27%. Ahead of it were awareness of data privacy and security concerns at 31%, the ability to evaluate AI outputs for quality and reliability at 30%, understanding bias in AI at 29%, and communicating AI findings clearly at 28%. Prompting still counts, but it isn't what employers rank first. They want people who can tell when the output is wrong and explain why.
KPMG's education lead, Rob Clayton, made the same point in the 2026 release: "Proper AI use is about more than knowing how to prompt."
Meanwhile, 88% of Canadian job seekers in the same Express research said they already have the AI skills employers want. When nearly everyone claims the skill, evidence sets a candidate apart.
What institutions can do
These four steps fit inside the courses you already run.
Teach responsible use inside the discipline
An AI ethics module helps, but judgment is learned where the stakes are concrete. In an accounting course, that can mean asking AI to draft a reconciliation note and then finding its errors. In marketing, it can mean checking a generated campaign for unsupported claims and for customer data that never should have gone into the tool. Put each course's AI rule in the syllabus in a sentence or two, so students stop guessing.
Assess judgment and applied skill, not recall
If an assignment can be finished by pasting the prompt into a chatbot, it measures access to a chatbot. Assessments that hold up ask a student to pick the better option in a realistic work situation, find what's wrong in a draft, or explain a tradeoff. They take longer to write, and they are much closer to what the PwC data says entry-level work now demands.
Give students evidence employers can check
A transcript doesn't tell a hiring manager what a graduate can do as a junior data analyst. Students need something an employer can open and check: results tied to a role, work samples and a profile that holds together. Our guide on how employers verify candidate skills covers what hiring teams look for.
Track readiness across the cohort, before graduation
Outcome surveys arrive months after students leave. A program head who sees in second term that a cohort is weak in data literacy still has time to add a workshop or rework an assignment.
Where START fits
We built the Employability Intelligence Platform for the last three of those steps, and a college can run it under its own name. Students sign in to a white-label portal with your branding and take role assessments that test them against a specific job. The platform draws on more than 10,000 questions across technology, business and 17 soft skills, at Junior, Intermediate and Expert levels, and students see their strengths and gaps as soon as they submit.
START builds custom question banks with your team, so questions can reflect your discipline and the applied judgment described above. Every new bank is reviewed before learners see it. Staff get an admin portal with completions, average scores, weakest skills, risk segments and trends over time, with CSV and PDF exports for program reviews.
Graduates can take their results to employers through a Professional Profile and a Professional Authenticity Certificate, which a START reviewer checks before it is marked START verified. When the data shows a gap, START's bootcamps and events on AI, technology, business and job search, run through StartJobs.ca, give students a next step. Most are free and online.
It doesn't replace your teaching. It gives your teaching a measurement, and it gives your graduates something to show.
Frequently asked questions
What percentage of Canadian students use AI for schoolwork?
KPMG Canada's survey, released in September 2026, found that 75% of students aged 18 and over use generative AI for schoolwork. KPMG's earlier surveys put the figure at 73% in 2025, 59% in 2024 and 52% in 2023.
Are Canadian colleges teaching students to use AI responsibly?
Some are, but not consistently. In the 2026 KPMG survey, 53% of students said they had received formal instruction on responsible AI use for academic work, and only 46% of educators said students get enough of it at their institution.
What AI skills do employers want from new graduates?
Canadian hiring managers surveyed for Express Employment Professionals in 2026 ranked data privacy awareness, evaluating AI outputs for quality and understanding bias above prompt writing. In that ranking, checking AI output came ahead of operating the tool.
Is using AI for schoolwork cheating?
It depends on the course rule, and that is the problem. In KPMG's 2026 survey, 72% of educators called generative AI use for schoolwork cheating, while 50% of students agreed. A clear rule in each syllabus, saying which uses are allowed, would remove much of the confusion.
How can a college measure whether students are ready for an AI workplace?
Use assessments tied to the roles each program prepares students for, and track results by cohort while students are still enrolled. Grades and outcome surveys alone show readiness too late.
If you lead career services or a program area, the institutions page shows how a branded portal, your own question bank and a cohort view work together, and you can talk to our team about starting with one learner group.

Boris Poludo
Founder and CEO, StartNew Technologies
Boris Poludo is the Founder and CEO of StartNew Technologies, the company behind START Canada. START Canada grew out of the career training of the Poludo Institute of Technology and Media, which prepared students, newcomers and career changers for work in Canada.

