When trying to communicate with future customers, it’s important to strike the right tone. When it isn’t right, as in the case of Microsoft’s Tay, things can spin out of control quickly. That’s why more companies are trying to understand how to integrate emotional intelligence into the machines that interact with people.

Broad Listening, an AI platform in beta mode, uses an “Artificial Emotional Intelligence” system as a human resources tool. By analyzing online behavior, Broad Listening can help companies at every step of the hiring process–finding the right candidate, communicating the job in an appealing way, aiding in the interview process, and even reassigning employees to better-fitting roles, if needed.

“When people are writing text and they want feedback on how they’re coming across, they want Artificial Emotional Intelligence,” (AEI), said Sarah Austin, founder of Broad Listening. This is especially important, Austin said, in terms of job posting. Broad Listening can provide information for recruiters and people working in HR.

SEE: Why Microsoft’s ‘Tay’ AI bot went wrong

“It’s about bringing in people to an organization,” Austin said. “You need to be hyperaware of how you’re coming across when you’re writing your job postings,” she said. “There’s not one person that’s going to be an authority on how to attract the right talent to a specific job, if they have no experience in that job.” Broad Listening can generate a posting, and then the AEI can look at the post and say, “This is missing, this is good, this is bad, this comes off a little sexist,” said Austin.

It’s important, Austin said, to pay attention to the fine details in written communication, the micro-decisions that go into our word choices. “If you said, ‘skiing is fun,’ or ‘I enjoy skiing, they sound like the same thing,” said Austin. “The difference is that when you say skiing is fun, you’re making a judgment. You’re saying skiing is fun for everybody. But it might not be fun for your grandma in a wheelchair, right? No, skiing is not fun for everybody.”

So how does the AEI engine work? It draws from thousands of real sources, everything from manager-employee interviews and surveys to job postings to Monster.com. “There’s this huge amount of data that we can sift through and organize to understand rules of jobs and personalities, or the best personalities and types of people for certain jobs,” said Austin. “I’m focused on what makes somebody great at their job and how to speak to those kinds of people and those kinds of jobs.”

Broad Listening claims that “from Watson to Hogan, no other system even comes close to the depth of understanding Broad Listening provides.” So how does Broad Listening stand apart from the others?

SEE: Why making AI sound human is a bad idea

“Watson is only looking at keywords,” said Austin. “What they do is they tag all of these words, that this is a good word or a bad word. Positive word, negative word.” But the reality, Austin said, is that the context determines what the word means. “Maybe you bought a small car because it’s environmentally-friendly, so it’s a positive thing for you. But Watson will say small is bad,” she said. Instead, they look at the whole sentence and sentences around it to analyze the context.

When working in human resources, said Austin, it’s important to make sure the language in written material is friendly and appropriate. “No one is consciously trying to be bad,” said Austin. “It’s that we have a lot of information in our subconscious that we’re not aware of. We need a little bit of that Emotional Intelligence to give us that feedback.”

Also see…