A.I. Could Help Humans Break Out of Their Echo Chambers

Yes, according to Ranit Aharonov, global manager of Project Debater at IBM Research in Haifa, Israel.

Speaking at Coins2Day’s Most Powerful Women International Summit in Montreal on Tuesday, Aharonov described her company’s project, an A.I. System that can debate human beings on various topics, addressing everything from how the machine works to the ethical implications of teaching it how to argue with people.

“We all need to understand that A.I. Has a lot of potential to empower people and advance people,” she said, “At the same time, we have to be careful of various biases and so on—but this is true for a lot of technologies.”

Big Blue’s Project Debater was exposed to “dozens of topics” during its training period to help it learn to identify text that contains relevant facts as well as argumentative language in an effort to help it determine a stance on a topic, Aharonov explained. The system eventually will figure out themes surrounding a particular topic, such as morality or effectiveness, so it can better organize its output, which arrives in the form of speech.

Aharonov said the project aims “to show how computers and humans interact” instead of merely existing to beat human beings in competitions. It has both won and lost debates over the years, though “winning a debate… is a very subjective matter,” she said. (Prior to a debate, the audience is polled for its stance; the side that wins is the one that compels more people to change their position.)

So how might Project Debater work in the real world? Aharonov said the technology could help with “decision making.” The system doesn’t just search for relevant documents, but organizes content that presents pros and cons—a useful tool for any organization. (And indeed Internet-using individuals, who find themselves increasingly in their own online “filter bubbles,” isolated from opinions that differ from their own.)

Said Aharonov: “Technology can be used to show the other side of the story.”