The Bot That Knows Your Feelings

It’s been more than 50 years since researcher Albert Mehrabian famously concluded that only 7 percent of human communication is delivered through words; the rest is body language (55 percent) and tone of voice (38 percent). Although the specific breakdown has been the subject of academic debate, what’s not controversial is the fact that all three combine to convey meaning.

That’s a challenge for artificial intelligence — but one that’s being solved. Bots have no problem understanding the meaning of words; vocal inflection and sarcasm can be trickier. Still, AI has advanced to the point where retailers, financial services companies and other businesses that deal directly with the public are deploying “sentiment analysis” technology to improve customer relationships.

Businesses spent more than $469 million on sentiment analysis technology in 2020, and the segment is growing at about 15 percent a year globally. Applications include chatbots that know when to kick something over to a human and call center technology that helps guide human call takers through challenging conversations, recommending specific phrases that can help keep emotions cool.

MORE AI: How artificial intelligence can solve multiple problems for retailers.

Now, even social media companies are getting into the act: Using sentiment analysis, Twitter might prompt you to think twice before posting a hotheaded tweet, and Facebook uses the tech to monitor conversations that might be getting out of hand. Research suggests that AI-mediated communication sometimes makes people uncomfortable, but nevertheless works in improving the quality of communication.

“People are creeped out by it, but it’s improving interpersonal perceptions of the people you’re communicating with,” Jess Hohenstein, a postdoctoral researcher at Cornell University, told Wired. “It’s counterintuitive.”