Students Use AI a Lot, but Many Don’t Trust It
Most uni students now use AI, but almost half of them don’t really trust it. That’s what a new Australian study found.
What is happening?
On 10 November 2026, researchers from four Australian universities shared new results in The Conversation. The universities are Monash, the University of Queensland, Deakin and the University of Technology Sydney. The University of Wollongong republished the article. The results come from their AIinHE.org project.
The team first ran this survey in 2024, with over 8,000 students. In early 2026, they ran it again and heard from more than 10,000 students. Here is what they found:
- Editing comes first: 69% of students use AI mostly to improve their writing. Another 64% use it to brainstorm, and 60% use it to learn how to do tasks.
- Worry about wrong answers: 67% limit their AI use because they don’t trust it to give true answers.
- Right and wrong: 47% have moral concerns about AI. The share of students who never use AI for classwork dropped a little, from 17% in 2024 to 15% in 2026.
- Not enough help: students who say their university gives enough AI guidance went up from 36% to 47%. But more than half still feel left on their own.
Why does this happen?
I think part of the reason is that universities are changing assignments faster than they explain the rules. Some classes now include AI in assessments, which can push use up even when students feel unsure.
AI language models guess the next likely word. They don’t check a list of facts. So they sometimes make up false details that sound true, which is called a hallucination. Many students seem to know this, so they use AI for safer jobs like fixing grammar or checking rough drafts, not writing whole essays.
A March 2026 UK survey (HEPI) found that 94% of UK students use generative AI for graded work. So almost everyone uses it. But the Australian study shows that using AI a lot doesn’t mean students like it or trust it.
Why does it matter?
Using a tool isn’t the same as trusting it. Many adults think students just copy whatever a chatbot writes. But these results show that students know AI makes things up, worry about fairness, and get confused when one teacher wants AI and another bans it.
This poll covers only four universities in Australia, so the numbers don’t speak for the whole world. Still, the lesson is clear: asking students to use AI without clear rules causes stress, not better learning.
What are the challenges?
Rules change from class to class. Students waste time guessing what counts as cheating, instead of studying. And more than half of students don’t feel their university gives them enough help.
What happens next?
Universities should stop sending mixed messages and include students when they make AI rules. Teachers should set clear rules for each class and teach students how to fact-check AI answers before handing in work.