By BU Today
July 13, 2017
Boston University and University of California, Berkeley researchers have received a grant from the U.S. Defense Advanced Research Projects Agency to find create a translation tool that explains the decision-making process of artificial intelligence to human users.
Researchers at Boston University and the University of California, Berkeley received a four-year, $7.55-million grant from the U.S. Defense Advanced Research Projects Agency to find new ways of getting inside the “mind” of artificial intelligence (AI), creating a translation tool that explains its decision-making process to human users.
Getting feedback on why an AI device makes a particular decision could help improve its accuracy by giving opportunities for humans to offer small corrections.
In addition, it could increase the trust that humans put into a machine, making it a better collaborator on complex jobs.
All of the primary neural network’s processing is devoted to solving its task. “That’s why we want to use a second neural network that has access to that machinery and input data, and can learn to translate all that into a textual version that humans can understand,” says BU professor Kate Saenko.
From BU Today
View Full Article
Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA
No entries found
from Communications of the ACM: Artificial Intelligence http://ift.tt/2tj0GDu