AI's prowess in gauging intoxication levels through tongue twisters is making headlines. Researchers have harnessed this tech by training it to analyze speech patterns and frequency changes indicative of varying stages of inebriation. In a study detailed in the Journal of Studies on Alcohol and Drugs, 18 consenting adults were dosed with vodka gimlets to intoxication levels, then tasked with reciting tongue twisters at regular intervals while their breath alcohol levels were monitored.


The study unveiled that alterations in voice pitch and frequency during intoxication could be decoded by AI, boasting a staggering 98% accuracy in determining legal sobriety limits for driving. Dr Brian Suffoletto, the lead author from Stanford's emergency medicine department, envisions practical applications for this technology, suggesting potential use as a safeguard in cars—employing a 'voice challenge' before ignition—or in high-risk workplaces like transportation or heavy machinery operation to ensure public safety.

Suffoletto also envisions restaurant and bar implementations to manage patrons' drink purchases effectively. However, while the concept exhibits promise, critics note the study's limitations: a small sample size of solely white participants. Petra Meier, a public health professor, acknowledges the potential of this approach but urges testing it in more diverse and extensive samples for a comprehensive evaluation.