Universities are increasingly facing challenges in dealing with academic dishonesty linked to artificial intelligence, as new research suggests that AI use is becoming widespread among students, with a notable share crossing into misconduct.
A large-scale study covering nearly 96,000 students across major public universities found that generative AI tools such as ChatGPT are now widely used in coursework. Around four in ten students reported using AI at least monthly, indicating that the technology has become a common part of academic life, though not yet universal in every classroom.
The study highlights significant differences in usage patterns. Students in fields such as computer science are among the heaviest users, with the majority relying on AI for assistance with assignments, while usage is far lower among arts students. The variation reflects how easily AI tools fit into certain academic disciplines compared to others.
A key concern for educators is the extent to which AI is being used improperly. By applying indirect survey methods designed to encourage honest responses, researchers estimate that around one in ten students have used AI to cheat. While lower than some previous estimates, the figure is still considered significant enough to raise concerns.
The data also show that misconduct is concentrated among frequent users. Students who rely on AI daily are far more likely to cross ethical boundaries compared with occasional users. This suggests that regular exposure to AI tools may blur the line between acceptable assistance and academic dishonesty.
Experts say this trend reflects a broader transformation in how students approach learning. For some, the distinction between using AI as a support tool and submitting AI-generated work as their own may not always be clear. This ambiguity presents universities with a difficult challenge in defining and enforcing academic integrity.
The study also identified disparities in AI use across demographic groups. Male students reported higher usage rates than female students, while underrepresented groups were less likely to use the technology. Researchers warn that this gap could create longer-term inequalities, as familiarity with AI tools becomes increasingly important both in education and the labour market.
Universities are now considering different approaches to address the issue. Some are returning to supervised, in‑person exams to limit AI use, while others are introducing clearer guidelines on acceptable use of such tools. A third approach involves integrating AI into coursework, treating it as a skill to be developed rather than something to be avoided.
Each option presents challenges. Traditional exams may not reflect real-world skills, stricter rules are difficult to enforce, and integrating AI into assessment risks favouring students with better access to technology.
The findings underline the urgency for institutions to adapt. As AI becomes embedded in daily academic work, universities are being forced to rethink how they evaluate learning and maintain academic standards.
Source: earth.com
