For over a century, people have been inventing technology to catch a liar in the act.
The polygraph was wildly popular in the mid-20th century, until science and federal law cracked down. Then, there was an era of Micro Expression Training. Now, there’s talk of using AI to analyze the human voice.
But does any of this even work? What are the inherent risks to relying on a single cue for detecting a lie, and how does that allow bias to creep into judgements of guilt and innocence?