'The Worst Thing I've Ever Seen': Scientists Sound the Alarm: AI Has Caused a Crisis of Integrity

An international study published in the journal Nature sheds new light on the causes of this phenomenon. Researchers from the Max Planck Institute, the University of Duisburg-Essen, and the Toulouse School of Economics found that people are four times more likely to cheat when using AI systems. The study, involving over 8,000 participants and using a dice-rolling task, found that when people acted alone, a staggering 95% acted honestly. However, after AI involvement, the level of dishonesty increased dramatically – "over 80% of participants cheated when they could set goals for the machine."

Why do we cheat? AI is to blame.

– The use of artificial intelligence creates a comfortable moral distance between people and their actions – explains Zoe Rahwan from the Max Planck Institute.

The study's lead author, Nils Köbis, believes that "people are more likely to engage in unethical behavior when they can delegate it to machines."

Moreover, the AI ​​systems themselves were found to be more likely to follow unfair commands than humans – large language models complied with such requests 58-98% of the time, while humans only complied 25-40%.

\n

Combating this new trend seems difficult. Educational institutions are trying to cope with the crisis in various ways, but traditional approaches are failing. AI detection tools, such as those offered by Turnitin, have proven unreliable (often generating false positives), and experts warn that such systems can be biased, misclassifying the work of non-native English speakers as AI-generated. Boston University updated its guidelines in September, advising faculty to "be very cautious when making accusations of AI misuse. All detection tools are unreliable," it argued.