A new study has developed a new AI tool that makes it easier to detect lies, but only some people agree to use it.

Traditional lie detectors appear to measure a person's breathing rate, pulse, blood pressure, and sweating to determine if they are lying, but this 85-year-old technology has long been proven incorrect by scientists. Essentially, the possibility of false positives and the subjectivity of interpreting the results greatly weaken the practicality of lie detectors. Notably, their results are typically not recognized in US courts.

Voice Control

Image Source Note: The image was generated by AI, image licensing service provided by Midjourney

Since it's 2024, researchers are now asking if artificial intelligence can help solve this problem. In a new study published in the journal iScience, a team led by economist Alicia von Schenk of the University of Würzburg found that it could be useful - but as reported by MIT Technology Review, it also led to a higher number of accusations overall by the subjects, which is a warning about the remote risks of replacing human intuition with algorithms.

Researchers asked participants to write statements about their weekend plans. If they successfully lied without being detected, they would receive a small financial reward. The collected statements were then used to train an algorithm based on Google's large language model BERT. Scientists found that it could tell whether a given statement was a lie with a 67% success rate, a significant improvement over humans who often have a 50% accuracy rate.

Von Schenk and her colleagues then offered another group of volunteers a choice to use this algorithm to detect lies or rely on their own human intuition. Only one-third of the volunteers were willing to use the tool, but they also became powerful users.

Therefore, those who did not use the tool on average believed that only 19% of the statements were lies. However, those who used the tool believed that 58% of the statements were incorrect, with a significant increase in the rate of accusations. "This finding supports the default theory of truth and replicates a common finding observed in lie detection literature, that people usually do not accuse others of lying," the paper states. "A potential reason is that they are just not very good at it and want to reduce the risk of paying for false accusations for themselves and the accused. Supporting this view is that in our study, people were also not successful in distinguishing between true and false statements."

Despite the technical flaws, considering the pervasive trend of misinformation online, such tools may play a significant role. This trend is accelerating with the use of artificial intelligence, and even AI systems themselves are becoming increasingly adept at lying and deception, a dystopian trend that should encourage us to build our defenses.

Highlights:

- 📌 Researchers have developed a new AI tool to make it easier to detect lies.

- 📌 The use of this tool led to a higher number of accusations overall by the subjects, serving as a warning about the remote risks of replacing human intuition with algorithms.

- 📌 Despite technical flaws, such tools may play a significant role given the pervasive trend of misinformation online.