Chatty Machines? Future Computers Could Communicate Like Humans

Binary code.
What if humans could speak to computers the same way they speak to other humans? (Image credit: DARPA)

In the future, you might be able to talk to computers and robots the same way you talk to your friends.

Researchers are trying to break down the language barrier between humans and computers, as part of a new program from the Defense Advanced Projects Agency (DARPA), which is responsible for developing new technologies for the U.S. military. The program — dubbed Communicating with Computers (CwC) — aims to get computers to express themselves more like humans by enabling them to use spoken language, facial expressions and gestures to communicate.

"[T]oday we view computers as tools to be activated by a few clicks or keywords, in large part because we are separated by a language barrier," Paul Cohen, DARPA's CwC program manager, said in a statement. "The goal of CwC is to bridge that barrier, and in the process encourage the development of new problem-solving technologies." [Humanoid Robots to Flying Cars: 10 Coolest DARPA Projects]

One of the problem-solving technologies that CwC could help further is the computer-based modeling used in cancer research. Computers previously developed by DARPA are already tasked with creating models of the complicated molecular processes that cause cells to become cancerous. But while these computers can churn out models quickly, they're not so adept at judging if the models are actually plausible and worthy of further research. If the computers could somehow seek the opinions of flesh-and-blood biologists, the work they do would likely be more useful for cancer researchers.

“Because humans and machines have different abilities, collaborations between them might be very productive," Cohen said.

Of course, getting a computer to collaborate with a person is easier said than done. Putting ideas into words is something that humans do naturally, but communicating is actually more complicated than it may seem, according to DARPA.

"Human communication feels so natural that we don't notice how much mental work it requires," Cohen said. "But try to communicate while you're doing something else — the high accident rate among people who text while driving says it all — and you'll quickly realize how demanding it is."

To get computers up to the task of communicating with people, CwC researchers have devised several tasks that require computers and humans to work together toward a common goal. One of the tasks, known as "collaborative composition," involves storytelling. In this exercise, humans and computers take turns contributing sentences until they've composed a short story.

"This is a parlor game for humans, but a tremendous challenge for computers," Cohen said. "To do it well, the machine must keep track of the ideas in the story, then generate an idea about how to extend the story and express this idea in language."

Another assignment that the CwC is planning is known as "block world," which would require humans and computers to communicate to build structures out of toy blocks. There's a tricky part, though: neither humans nor computers will be told what to build. Instead, they'll have to work together to make a structure that can stand up of its own accord.

In the future, DARPA researchers hope that computers will be able to do more than play with blocks, of course. If it's successful, CwC could help advance the fields of roboticsand semi-autonomous systems. The programming and preconfigured interfaces currently used in these fields don't allow for easy communication between machines and humans. Better communications technologies could help robot operators use natural language to describe missions and give directions to the machines they operate both before and during operations. And in addition to making life easier for human operators, CwC could make it possible for robots to request advice or information from humans when they get into sticky situations.

Follow Elizabeth Palermo @techEpalermo. Follow Live Science @livescience, Facebook & Google+. Original article on Live Science.

Elizabeth Peterson
Contributor

Elizabeth is a former Live Science associate editor and current director of audience development at the Chamber of Commerce. She graduated with a bachelor of arts degree from George Washington University. Elizabeth has traveled throughout the Americas, studying political systems and indigenous cultures and teaching English to students of all ages.