New system lets AI take a piece of someone’s mind and passes on the rest to a human.

For years now there has been a lot of talk about the problem of “power harassment” in which employees are pushed to unhealthy extremes by their superiors. While there’s still a lot of work to be done there, attention has been gradually turning to the other end of the suffering spectrum, known as customer harassment,” in which employees are pushed to unhealthy extremes by their customers.

Customer harassment puts companies in a tricky situation since they don’t want to step on the PR landmine of telling off or pressing charges against the people who essentially support them. On the other hand, having employees with mental breakdowns as a result of dealing with people who seem to lack human decency is also bad for the old brand image.

It’s a lose-lose situation but luckily technology has found an answer in the form of an AI that takes angry voices and converts them into something more palatable to a listener’s ears.

▼ A news report demonstrating the system still in development, with an example of how it changes voices.

The demonstration of this system developed by SoftBank and the University of Tokyo in the video above shows how “IS THIS HOW YOU TREAT A CUSTOMER OF OVER FIVE YEARS?! HEY!!!” becomes “Is this how you treat a customer of over five years? Hey…” with the flick of a switch.

Readers of the news seemed to like the idea, though some online comments wondered: Why stop there when it comes to what AI can do at call centers?

“Just lowering the pitch seems to have a big effect.”
“The slow voice seems good.”
“It should make everyone sound like a classy lady too.”
“Will there be a warning that this call will be recorded and altered to make you cuter?”
“Just let the AI do the whole thing.”
“They could get an AI to do the talking on the call center end, but humans are too ambiguous for it to be effective enough.”
“I think it works because an old phone I had used a ‘slow voice mode’ for seniors, and I couldn’t never tell if anyone was angry because everyone sounded sleepy.”
“Is it less offensive to call someone an idiot in a soft voice?”

“Unfortunately, it’s not just a problem of yelling. I’ve had a lot of people use snark and muttering curse words to be offensive.”
“Just set it up that when a person sees it’s hopeless they can transfer the caller to the AI ‘manager.'”

AI would certainly seem to be every bit as good at handling the more emotionally toxic elements of working at call centers as it is at making kitschy art. On the other hand, anyone who’s dealt with one of these chatbots would have noticed that they’re almost pathologically apologetic which could backfire and be irritating for normal people calling in with simple questions and problems.

So, SoftBank’s solution that combines the strengths of AI and people is probably ideal. They are still fine-tuning the system so that operators can accurately understand how the callers feel while still being shielded from abusive language. Perhaps other features will also be added by the time it is expected to be put into use next year.

Source: NTV News, Golden Times
Featured images: Pakutaso
● Want to hear about SoraNews24’s latest articles as soon as they’re published? Follow us on Facebook and Twitter!