Editor's Note: This article is part of a special report on the Top 10 Emerging Technologies of 2016 produced by the World Economic Forum. The list, compiled by the Forum’s Meta-Council on Emerging Technologies, highlights technological advances its members, including Scientific American Editor in Chief Mariette DiChristina, believe have the power to improve lives, transform industries and safeguard the planet. It also provides an opportunity to debate any human, societal, economic or environmental risks and concerns that the technologies may pose prior to widespread adoption.
One of the advantages that CEOs and celebrities have over most people is that they don’t need to spend much time handling the uninteresting, time-consuming aspects of daily life: scheduling appointments, making travel plans, searching for the information they want. They have personal assistants, or PAs, who handle such things. But soon—maybe even this year—most of us will be able to afford this luxury for the price of few lattes a month, thanks to the emergence of an open AI ecosystem.
AI here refers, of course, to artificial intelligence. Apple’s Siri, Microsoft’s Cortana, Google’s OK Google and Amazon’s Echo services are nifty in the way that they extract questions from speech using natural-language processing and then do a limited set of useful things, such as look for a restaurant, get driving directions, find an open slot for a meeting, or run a simple web search. But too often their response to a request for help is “Sorry, I don’t know about that” or “here’s what I found on the web.” You would never confuse these digital assistants for a human PA. Moreover, these systems are proprietary and hard for entrepreneurs to extend with new features.
But over the past several years, several pieces of emerging technology have linked together in ways that make it easier to build far more powerful, human-like digital assistants—that is, into an open AI ecosystem. This ecosystem connects not only to our mobile devices and computers—and through them to our messages, contacts, finances, calendars and work files—but also to the thermostat in the bedroom, the scale in the bathroom, the bracelet on the wrist, even the car in the driveway. The interconnection of the Internet with the Internet of Things and your own personal data, all instantly available almost anywhere via spoken conversations with an AI, could unlock higher productivity and better health and happiness for millions of people within the next few years.
By pooling anonymized health data and providing personalized health advice to individuals, such systems should lead to substantial improvements in health and reductions in the costs of health care. Applications of AI to financial services could reduce unintentional errors, as well as intentional (fraudulent) ones—offering new layers of protection to an aging population.
The secret ingredient in this technology that has been largely lacking to date is context. Up to now, machines have been largely oblivious to the details of our work, our bodies and our lives. A human PA knows when you are interruptible, stressed, bored, tired or hungry. The PA knows who and what is important to you, and also what you would prefer to avoid. AI systems are gaining the ability to acquire and interpret contextual cues so that they can gain these skills as well. Although initially these AI assistants will not outperform the human variety, they will be useful—and roughly a thousand times less expensive.
Various companies have already demonstrated AI systems that have some of these capabilities. Microsoft Research built one that knows when you are too busy to take a call (and which calls should ring through regardless) and that automatically schedules meetings at times you would likely choose yourself. Other companies such as Angel.ai have introduced services that search for flights that suit your preferences and constraints based on simple plain-English requests.
Just as discretion and loyalty are prized among human PAs, digital versions will succeed only to the extent that we trust them with our security and privacy; they will need to act in the best interests of the user, once it figures out what those are. These are interesting challenges for the AI community.