Google CEO Sundar Pichai delighted the audience at the Internet giant's annual developer event Google I/O last year with a demonstration of an upcoming feature for Google Assistant currently called Duplex. Live in front of the Mountain View audience, Pichai showed Google Assistant making a telephone call to a hair salon, talking to the salon representative who answered the phone, negotiating the time of the appointment and making a booking for the user.
The Google Duplex demo gave the audience (and Youtube viewers around the world) a tiny glimpse into our artificial intelligence future: a future where our intelligent devices will be able to make our calls, restaurant reservations, flight bookings and buy us tickets for the theatre. It's a seemingly benevolent view of how AI will help ordinary people get the most out of life and how these technologies can be democratised for the greater good.
The future context of this was left largely uncommented on by Pichai at the time: the focus being on the magic of the machine-to-human negotiation. Many would argue that we can expect most reservations and service enquiries in the future to be made machine-to-machine, not machine-to-human.
Machine-to-human or machine-to-machine?
In a relatively short space of time, restaurants, airlines, stores and theatres will all have virtual assistants of their own. Where it's not possible for your virtual assistant to make your booking directly via an online ecommerce platform, it will be able to do so by communicating directly with other virtual assistants.
Google Duplex itself, which is now in usage in 42 states in the U.S., actually checks a number of third-party online reservation systems first, to see if it can make your booking for you digitally, before attempting a voice call on your behalf. Even then, according to comments made by Google to the New York Times earlier this year, in 25 percent of cases it results in a human-to-human call made from one of its call centres.
However, before too long, your AI won't even have to call an actual person in a salon, it will call the salon's AI. It will do so in less time than it takes you to think about your request, taking into account as many variables as may affect your purchase and it will come back to you with a list of recommended options for you to choose from to move your request forwards. As Elon Musk said last week at the World AI Conference in Shanghai: to a computer, a millisecond is an eternity. In the relentless pursuit of efficiency and commerce, cutting out the middleman (or woman) makes perfect sense. As already demonstrated nicely at last year's Google I/O, human's just get in the way of the AI transaction!
Millions of people around the world who are currently amused by virtual assistants such as Alexa, Cortana, Google Assistant or Siri, are going to find that these assistants and others will become integral to their commercial interactions, whether personal or business. Amazon's Alexa, of course, already has ecommerce capabilities tightly integrated and its possible to use the assistant to purchase a wide range of services, depending on whether they are currently supported in your country. Google, likewise, is rolling out more and more commercial services that will be accessible via Google Assistant. However, just as early desktop computer programmes and data formats were, at first, incompatible with each other, today's virtual assistants aren't yet very good at talking to one another, but this is a temporary state of affairs.
In the future, virtual assistants will begin talking to other systems regardless of their platform or manufacturer, allowing AI to AI conversations to help us with every potential information, education, social or commercial request imaginable.
Amazon and Microsoft announced a small step in this direction last year, with the development of integration that allows Alexa users to open Microsoft's Cortana voice assistant and vice versa.
Creating a world where 'everyone' will have their own virtual assistant, is going to impact social networking, human relations and society as a whole. It's also, naturally, going to have an enormous impact on human behaviour.
From the introduction of Facebook, Twitter and other social networks, our whole society has changed and, naturally, human behaviour with it. Many societies are now reliant on digital technology and services. The access that technology now gives us to information, invariably means that our first point of contact to find out about anything at all - from healthcare to history, and from science to shopping - is the Internet. In terms of communication, consumers have changed behaviours from being phone-first, to web-first, to mobile-first and now, it seems, we are destined to become AI-first.
So, what can we expect from our new generation of AI natives, growing up with an increasingly pervasive use of AI assistants designed to offer advice, communications, information, education, recommendations and shopping?
San Francisco based non-profit Common Sense Media conducted a survey last year of 1,000 teenagers between 13 and 17 years old (i.e. Generation Z). The results found that 61 percent of teens prefer texting, video-chatting, or using social media to talk to their friends, rather than face-to-face. Just 32 percent of those surveyed preferred talk to their friends in-person, compared with 49 percent of teens in 2012.
Our newest generations will grow up using virtual assistants and live in a world where almost everyone habitually uses AI. So, what will be the preferred method of AI natives to connect with other AI natives? It could be simply: 'have your AI call my AI'!
However, it would be a mistake to assume that the use of AI to AI transactions is going to be limited to AI natives. If we've learnt anything from the digital revolution over the last two decades, it is that consumers follow the path of least resistance. Introducing the best technology doesn't always effect a change. More often, it is the easiest to use technology that drives adoption. Virtual assistants are quickly going to become the easiest way for consumers to get things done, in particular when it comes to buying products and services.
A question of data
Despite rapid advances in speech recognition and natural language processing, making virtual assistants more proficient in dealing with human voice interactions, the likelihood is that AI to AI communication will be much more efficient and error free. True, an AI assistant may be able to place computer-to-computer orders today, but only within its own connected partner ecosystem.
From a data point of view, why increase the chances of error by having a human give or receive an order via an intelligent device, when another intelligent device could do it faster, more comprehensively and with no errors. In future, a voice assistant application developer's goal may well be to try to personalise the customer experience, while involving humans as little as possible in the actual transactions, where accuracy, data completeness and verification are the priorities.
Consumers will embrace this new AI to AI communication, because it relieves them of the tedium of answering different sets of questions for every product or service order they make, whilst ensuring that their order is always consistent with their needs and preferences. Vendors will adopt such technologies, since they get to focus on customer experience and the emotional connection with their customers, knowing that all the fine details can be taken care of by their AI.
To use Google's Duplex demonstration as an analogy: in the future, only the customer's virtual assistant and the salon's virtual assistant need to communicate with one another at the same time. Voices will be optional: and, without them, the whole demonstration would take about a second.
This article was contributed by Carrington Malin. Carrington is an entrepreneur, marketing professional and advisor who has worked across almost every sector of technology. He helps companies, startup ventures and public sector organisations develop marketing strategies, digital initiatives and leverage new marketing technologies. He also publishes a daily Asia AI News digest. You can connect with Carrington on Twitter or via Linkedin