Operators have a “calling” to develop artificial intelligence (AI) powered voice services and should take immediate action, argued Oren Jacob, CEO of PullString, warning that technology giants – Amazon, Apple, Microsoft and Google – now have their own relationship with mobile customers through their assistants.

Jacob, co-founder of the AI-voice application developer, said in yesterday’s keynote that a voice-based interface was an “essential part of the future”, and described the fact that the four aforementioned companies had built voice assistant characters as “interesting”, given the reach these services now have.

“Apple, Google, Microsoft and Amazon have built characters that are connecting to a larger audience than any other characters in history because all those things are present on mobile devices,” he said. “That therefore presents a new connection between those companies and customers on the device. For telcos and the audience here that have a connection with their customers, someone else has just stepped into that space.”

Jacob continued: “It’s essential that companies that have a customer relationship through devices, either as a carrier or a manufacturer, also have an offering in language, because there is nothing more connected than talking to someone. Siri, Alexa and Google are walking away with it.”

In a thought-provoking presentation, Jacob explained the differences between the current way people interact with computers and technology, and how that will change through voice.

He said that communicating through keyboards, a mouse and touchscreens was “very precise, very accurate and unambiguous, generally speaking,”, while representing “a very small aperture compared to the spoken language”.

“As we talk in all the human languages of the world, those languages are much broader than the normal and typical input we get from computer language, he said. “We speak in words, sentences, or just on expressions. All of the human condition is loaded up into the language we use to talk to each other. It is also imprecise, inaccurate, ambiguous, and amazingly, has infinite aperture. It can also be shockingly efficient.”

In addition, Jacob noted the challenges and limitations that today’s smart speakers have, particularly if the assistant has to deal with hearing two different people speaking about two different things at the same time. For those planning to build AI-voice intelligence, he said it was important to spend as much time in developing what the voice experience is supposed to do, in addition to what it is not supposed to do.

“Random topic changes happen all the time and this puts the computer conversation under direct stress and pressure. How will it answer? Will it understand if I change topics? Will it get mad at me? Will it error out? The moment this happens, the pressure is on, and how you design for that moment is absolutely critical,” he added.