New Survey Reveals Student AI Use in College: 5% Admit Full Assignment Generation, 73% Fear False Accusations

Summary

A new survey from Packback, an educational technology platform, surveyed nearly 700 college students and found that approximately 5 percent of students report regularly using AI to generate complete assignments—a rate comparable to traditional academic dishonesty. The report reveals a striking disconnect between students' actual AI usage and their perceptions of peer behavior, with about 25 percent admitting to using AI to support coursework (summarizing materials, etc.), while roughly three times as many believe their peers are doing so. Most notably, nearly 73 percent of students expressed at least moderate concern about being wrongly accused of using AI, with over 40 percent describing it as a major concern.

Packback CEO Kelsey Behringer attributed the anxiety to strict institutional AI policies and reliance on "flawed technology" for detection, noting that fear of false accusations may be demotivating students from pursuing education. The survey also found significant variation by field of study, with business and management students using AI most frequently and reporting greater benefits, while humanities and social sciences students reported lower usage rates. Behringer emphasized that AI usage mirrors how knowledge workers use the technology—as a thinking and brainstorming partner—and that students who are academically dishonest today are likely the same ones who would have cheated before AI existed.

• Strict institutional policies and unreliable AI detection tools are creating student anxiety that may undermine educational motivation rather than deter dishonesty

Editorial Opinion

This survey highlights a critical tension in higher education: while institutions are implementing strict AI policies to prevent cheating, they may be inadvertently creating a chilling effect that discourages legitimate learning. The disconnect between actual usage (5% full assignment generation) and perceived usage suggests that institutional messaging around AI risks is amplifying student anxiety without necessarily addressing the real problem. Colleges need to move beyond blanket restrictions toward nuanced policies that distinguish between AI as a learning tool and AI as academic dishonesty, while simultaneously improving detection reliability to justify the stringent enforcement these policies entail.