When kids hide or miscalculate how much their project partners rely on AI tools, team scores drop—and students who are already struggling take the biggest hit.
Hiding or failing to communicate AI use during group projects drags down final scores. This transparency gap is particularly damaging for kids who are already struggling with the subject matter.
Group work now regularly involves tools like ChatGPT or AI coding assistants. If your child misjudges how much a partner relies on AI, they cannot accurately evaluate the quality of the work, divide tasks fairly, or catch critical errors.
When digital short-cuts replace clear communication, lower grades and missed learning opportunities follow.
Generative AI makes it effortless to output polished text or working code in seconds, blurring the line between actual human understanding and machine output. Researchers wanted to know whether students can accurately spot their peers' AI habits, and whether failing to read a partner's workflow hurts the final grade.
A major gap between what students think their partner is doing with AI and what the partner is actually doing correlates with lower final project grades.
- In-person sessions do not fix the gap: Sitting side-by-side during group work failed to clear up misunderstandings about who was using AI and how much.
- Struggling kids suffer most: The negative impact of AI misperceptions was strongest in teams with lower prior performance.
- Silent work creates confusion: Students gauge a partner's skill and effort through the work submitted, which heavy AI use renders difficult to evaluate.
The core problem is not the technology itself; it is the breakdown of peer-to-peer learning. When AI obscures who actually understands the concept, weaker students lose the chance to learn from a stronger partner's authentic thought process—assuming instead that computer-generated output reflects personal mastery.
This study focused on 206 undergraduate software engineering students in a specialized university course. Results may differ for younger students or non-technical subjects like history or English. Additionally, this is an unreviewed preprint and observational study, meaning it shows a clear correlation, not definitive proof that misperceptions directly caused the drop in grades.
- If your child is starting a group project where digital tools are allowed... Tell them to establish an explicit "AI ground rule" with their partner on day one so both kids agree on when tools like ChatGPT can be used.
- If your child struggles with a subject and is paired with a high-achieving peer... Encourage them to ask their partner to walk through their thinking out loud rather than just accepting finished work, ensuring AI output isn't hiding a lack of real understanding.
- If you assume working side-by-side in person prevents miscommunication... Remind your child to actively ask what tools their partner used off-screen, since sitting together at a table does not automatically reveal who wrote the work.
Secret or unspoken AI use undermines teamwork and hurts the students who need collaboration the most. Teaching your child to be upfront about their digital workflow protects their learning and their grades.
Laura Graf, Ramona Beinstingel, Stephan Kusche et al. (2026). Students' Perception Accuracy of Partners' AI Use and its Relation to Collaboration Performance. arXiv (preprint). — arxiv.org