By Steve Outing
A few years ago I learned about some academic researchers in the UK who had developed a next-generation “lie detector.” They had been working on video-analysis algorithms that detected tiny changes in facial movements — such that the human eye typically wouldn’t notice — and could analyze a video of a person talking and identify moments containing certain minuscule facial movements that tend to occur when someone is not being truthful.
The group of academics was working with a British relative of mine, hoping to commercialize the technology. As you can imagine, a “better lie detector” would have many applications across many fields. Beyond the obvious (law enforcement, human resources and recruiting, politicians’ and CEOs’ speeches, etc.), my mind immediately imagined how useful this would be for journalists — not as a definitive “he/she was lying when answering my question about X,” but as a reasonably accurate indicator that an interviewee probably is not to be taken at face value when the video algorithm detects minute facial movements associated with an individual not telling the truth.
|Could smart glasses be the vehicle of live video lie detection soon?|
Last I heard from my relative, the transition from academic research to commercial product ran into problems, apparently having more to do with academic politics than with the technology’s efficacy.
Of course, others have been working on this problem, and researchers at MIT have developed what they call “Eulerian Video Magnification” (EVM). A simple explanation is that this technology can take a standard video sequence (recorded or live!) and look for a series of small motions or visually imperceptible body changes, amplifying each to show when during a statement by an individual these changes occur.
The EVM technology is not restricted to uses for “lie detection,” but it is well suited for it. For example, the filtered-and-amplified video analysis might show when a flow of blood fills the face, which probably is a sign of the person becoming uncomfortable, and perhaps not telling the truth. The same goes for certain minute movements of the eyes, eyebrows, or lips which researchers have determined accompany saying something untruthful.
Researchers have found many facial changes that can indicate probable lying — in fact, so many that an accomplished liar is unlikely to be able to avoid all of them. The talented liar might fool the human interviewer, but amplified visual algorithm analysis will reveal the telltale facial ticks and movements and quick eye movements that otherwise would go unnoticed. A good liar could be adept at fooling a traditional electronic lie-detector (with wire sensors attached to the person to monitor various body states), but not a really good video algorithm measuring a host of facial movements that would be impossible for a lying person to completely squelch.
Can you see where this is going?
Where this will get really interesting in the years ahead is when EVM and competing technologies are incorporated into mobile and wearable devices.
First, we’ll need to get EVM-like apps onto smartphones, which will analyze data captured by the phone’s video camera. Of course, it’s a bit awkward in an interview situation for a journalist to use a smartphone to video-record the session. It probably would be necessary to later review the video to spot points in the conversation where the app indicated the possibility that the interviewee was lying. That’s a powerful tip-off for the reporter to subsequently fact check the suspected false statements.
|The glasses could send the journalist live alerts via the glasses whenever the software detects subtle indicators of potential lying by the interviewee|
Here’s a much better solution, and it’s not far away: Use the video camera in a pair of “smart glasses” (e.g., Google Glass or the inevitably coming wave of competitors). Connected wirelessly to a smartphone running an EVM app, the glasses could send the journalist live alerts via the glasses whenever the software detects subtle indicators of potential lying by the interviewee.
There should be no assumption that the “lie-detector” app is completely or even mostly accurate in spotting a falsehood, but a journalist tipped off in real time would be automatically prompted to question what was just said. (I’d love to see TV-news interviewers and news talk-show hosts using this technology, so that they might push back more on statements by politicians and business leaders rather than let them state an untruth and go unchallenged, which has too often been the norm.)
Ethics? Law? TBD
While this all sounds wonderful from the perspective of a journalist or a consumer of news, “live” lie detection presents all sorts of ethical and legal questions and issues. It’s beyond the scope of this article to go deep into those woods. But here are just a few things to consider before reporters and TV interviewers get this “super-power” of spotting when someone is probably lying:
- Is it ethical to turn on a wearable lie detector during an interview without telling the interviewee? Is it legal?
- Is it OK if you inform the person what you’re doing and get his/her verbal permission?
- Would it be appropriate, and legal, for a shareholder or a financial reporter to do live video lie-detection as a CEO speaks at an annual shareholders’ meeting and accuse the executive of lying based on this technology?
- Will presidential-debate moderators begin to utilize video lie-detection live in order to catch candidates in being dishonest? Is that shoddy ethics, or good journalism?
- Will this new technology, when released into the wild, cause such a backlash by privacy advocates that it is not allowed to mature and evolve?
One thing is certain: Live video lie-detection will keep some lawyers busy and well paid, and ethicists and philosophers up late at night pondering this technology’s impact on our society, culture, business, government, and more.