AI Speech Tricks: Humans Struggle to Distinguish Real from Fake

0:00It's March 23rd, Canada News Today starts now, AI-powered and ready. Chinese scientists have uncovered strong evidence that people struggle to distinguish real human speech from AI-generated fakes. Researchers from Tianjin University and the Chinese University of Hong Kong tested 30 volunteers hooked up to brain scanners. The group listened to audio clips and tried to spot the artificial ones, but they performed poorly overall. The team then gave the participants training tips to improve their detection skills. This effort offered only slight behavioral gains, meaning folks still couldn't reliably tell the difference. This lines up with other recent studies raising alarms. For instance, work from Queen Mary University of London last September showed deep fake voices are now nearly impossible to spot by ear. A study last month from Australian universities found similar overconfidence when judging fake faces. On a promising note, the brain scans revealed something key. After training, neural responses became clear between human and machine speech, picking

1:02up subtle sound differences even if decisions didn't follow. Suit As AI audio tricks grow more sophisticated, these findings underscore the need for better tools and awareness to combat rising risks in areas like finance and impersonation.