Cross-posted from the AAAS Sci on the Fly Blog Jan. 20, 2017
Your Facebook feed, last night’s Halo session, or the ads you saw this morning all came from artificial Intelligence (AI). Despite more than sixty years of research in this field, we are just beginning to implement AI in our daily lives which has left many wondering what our world will be like with further adoption of this technology. One example of what we might expect is to look at our relationship with Siri, Apple's personal digital assistant. With a click of a button, we ask Siri to traverse the digital world to find movie times, control the living room thermostat, or predict the weather. Although very few of us would actually ask Siri for help in public, most agree that it represents a large step for AI. So, how did Siri become such an accepted part of everyday life and what can it teach us about how we will live with AI tomorrow?
A play on the Latin word “calonis,” meaning “soldier’s servant,” CALO stands for Cognitive Agent that Learns and Observes. It was developed by SRI International and was part of a group of research projects funded in the early 2000s by the Pentagon’s think tank, the Defense Advanced Research Projects Agency (DARPA), focusing on improving the way computers can support humans. The fundamental research in AI conducted under CALO contributed to the development of a military AI framework designed to help warfighters effectively utilize and navigate the increasing amount of data and electronic systems on the battlefield. More importantly for us, it also led to several commercial applications; one of which was the first smartphone digital assistant, Siri.
So, what do soldiers have to do with Siri? Well, just look at the sheer number of gadgets a soldier must carry: active optics, flashlights, night vision goggles, GPS, smartphone, tablet, communications headset, and multiple radios—not to mention nearly 16 pounds of batteries! Although these capabilities can be very empowering; anyone who has run into a pole while texting can tell you, multi-tasking can be dangerous—especially while being shot at. Data and gadgets alone are not helpful unless we can filter out the signal from the noise and enable easy communication between all the hardware components.
The reality is that humans are notoriously bad at multi-tasking and processing large amounts of information. What we are good at is making tools; and as the complexity of our world increases, so have the number of tools we need to manage—the App Store alone carries over two million apps! Although AI programs like Siri can only communicate with a single app at a time, future digital assistants will be able to understand complex tasks and string together various actions across multiple apps to help us get things done.
As we expand the digital world through larger databases and more networked devices, we will need to rely on newer, more sophisticated AI programs to help us navigate the dense, digital landscape. Future AI programs will take on a more active role; not just gathering information but coordinating and controlling a course of action. For instance, IBM’s futuristic AI system, Watson, can churn through vast amounts of data from medical records and medical journals to reveal new insights in cancer diagnosis and help doctors create a customized treatment path. Even the daily rhythms of a city, such as bus routes and sewage flow, are starting to be monitored and controlled by AI revealing to city planners unforeseen areas of need.
The challenges facing our world are complex and so too are the solutions. Although AI systems can help narrow down our options, none of these systems will be useful if we are not asking the right questions. “What is the quality of life for this course of cancer treatment?” or “How are we ensuring that city services are inclusive to all?” Our inquiries will need to move beyond just simple facts and towards real-world goals. In an increasingly digital world, we need to remember that the future will not be absent of humans. Rather, we will need to collaborate with AI to discover what the right questions are.