BALTIMORE, Md. — We've all done it - shouted "customer service representative!" at the phone until we were finally given the chance to speak to a real, live person instead of a robot.
But what if those robots could tell when you were getting annoyed and respond accordingly? Apparently, some already can.
"They can gauge how the customer is responding, like a good human, but not all humans are like this, and try to de-escalate situations where the customer can be upset. That has a lot of promise if humans don't mind being kind of manipulated by AI to not explode," Anton Dahbura, co-director of Johns Hopkins Institute for Assured Autonomy, told WMAR-2 News.
Dahbura says pretty soon, calling a business and getting an AI agent on the other line will become as normal as hearing a machine tell you to "press 1 for English." And you probably won't even know it's not a real person at first.
You can listen to an example in the video above.
The benefits to businesses are obvious - it saves them time and money. But whether AI agents will improve the customer experience is up for debate.
"The smart companies will keep the human in the loop," Dahbura said. "That’s the concern I have. If they just set it and forget it, would be the term, it’s gonna be problematic for both sides [...] So companies really need to figure out how to make the best out of it, and not just view it as an immediate cost-savings by getting rid of humans, and leaving the customers to flail around with the robots."
If you're one of those people who dread talking on the phone in the first place, there's a world in which you might not have to ever again. The technology works both ways. For example, Google already provides a service where an AI assistant will call a restaurant to make a reservation for you. There's an example of that in the video above too.
Of course, there's a dark side to all this technology too, when it winds up in the wrong hands, as our area saw recently with the Pikesville High School case.
But even without humans using it for nefarious purposes, the technology itself could prove to be just as flawed as we are. AI learns by using data from real-world conversations, biases, and prejudices included.
"Because of all the training, do you really want an AI agent that learned from all its training data that it should treat women in one way, and men in another way, or people with different racial backgrounds, things like that? Bias is just all over the place, and the machines are learning very quickly from humans. That’s very difficult to get rid of," Dahbura said.
Still, Dahbura says the technology becoming the norm is inevitable, and businesses and customers have to learn how to adapt with it.
"I think we’ll get used to it, as long as things are handled the right way," he said.