Communications of the ACM
Intelligent Machines Are Asked to Explain How Their Minds Work
Researchers at Xerox PARC are trying to teach intelligent machines to explain, in human terms, how their minds work.
Credit: Philip Toledano
Researchers at Xerox's PARC laboratory are trying to teach intelligent machines to explain, in human terms, how their minds work.
The project, sponsored by the U.S. Defense Advanced Research Projects Agency (DARPA), aims to learn how deep learning systems come up with the answers they do, which will further the development of artificial intelligence (AI) programs.
DARPA's pursuit of "explainable AI" reflects the need for the U.S. military to be able to have full trust in future robotic battlefield systems.
The DARPA program is set to run until the middle of 2021, at which time it could be incorporated into a range of systems, including those for robotic ships, which are being tested by the U.S. Navy.
The PARC project aims to use human teachers to train AI systems in the same way human students are taught, starting with simple concepts before building a deeper knowledge.
From Financial Times
View Full Article
Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA
No entries found