Chat and smile with virtual assistants
On the other end of the phone, a warm voice that receives questions, as has happened all his life. But that voice does not come from a human throat. It has been designed by goldsmiths of all disciplines to fulfill a double task: to resolve customer inquiries and retain them.
Although it sounds like the future, Marcos Obed, director of the Digital Experience area and head of Conversational Services Europe at the Everis technology consultancy, part of the NTT DATA Group, insists that we are already talking about a palpable reality. His company, for example, offers technological solutions such as eva, a platform for bots to act as a center of attention for customers. Both written and spoken, the bots they can take on millions of customers and analyze data from their conversations to perfect themselves over time as virtual assistants. It is a sample of a gigantic market, which will grow at an average of 37.7% between now and 2027 to exceed 36,000 million euros, according to data from Grand View Research.
The technological layer that makes this new generation of virtual assistants possible is not new. It is as old as humanity: it speaks. Natural language, more precisely, its creation through technological means – a discipline of artificial intelligence known as NLP (Natural Language Processing) -, is the starting point. In conjunction with another human capacity, vision. “What we are talking about, in terms of virtual assistants, is the ability to interpret the emotions of a client when he talks to the assistant. And not something as simple as ‘this is happy or angry’. We are talking about compositions as complex as detecting if a person is nostalgic ”, points out Obed.
The speed of computing has been the great technological leap in the sector. Of machine learning and the deep learning has been passed, since 2017, to models transformer, that have had a key impact in tackling tasks such as translation. “Google T5, for example, is trained with 11 trillion parameters. GPT-3 reaches 175 billion, after making a web-scraping —Technique for extracting information from the internet using software automated system that emulates a person’s digital browsing — to Wikipedia among other sources. It means that, basically, it is capable of answering you about everything. This allows us to visualize the great changes that will take place in our relationships with machines ”, summarizes Obed.
Another key technology in the field is neural machine translation networks, which have gone from robotic voices generated by phonemes to directly synthesizing sound waves that the human ear perceives as a genuine voice.
The creation of virtual assistants is not focused on technology, but on the user experience. Our conversation designers come from the humanities
Marcos Obed, director of Everis, of the NTT DATA Group
The impact for companies of this type of virtual assistants is enormous. First of all, in the obvious, reduce the great cost of the switchboards for customer service. But beyond this cost cut, there is a much more precise calibration of what experience you want to offer the user in order to retain them. In this objective, multidisciplinary teams play a key role where the humanities find a surprising central place in the technological bosom. “An important part of our teams for the creation of these virtual assistants are not focused on technology, but on the user experience. Our professionals, whom we call conversation designers, come from disciplines such as philosophy, the humanities, and linguistics. In fact, we, with the brands that hire us, do not talk about technology, but about telling a story, because the voice of the virtual assistant is going to be their voice for the customer ”, says Obed.
The impact of these technologies will reach far beyond the customer-company relationship. The University of Cambridge already published in January 2017 an extensive compilation article of all natural process language technologies and their impact on mental health. But also such a key and delicate process as the job interview can be affected by this revolutionary technology that could act as an accurate polygraph to evaluate the candidate. “We have already had some pilot experience in this regard. Although, effectively, this will require legislation that defines the legal framework and protects the privacy of people in all areas of their interaction with machines ”, Obed adds. A framework that should accelerate its implementation to respond to a technology of unstoppable evolution and profound consequences: in 2024 there will be 8,400 million devices with voice assistants. That is, a number analogous (de facto, slightly higher) to the world population.