Defense Department funds massive speech recognition and translation program describes U.S. government research spookily like the The Total Information Awareness System. TIA was a research program bent on developing a wide range of technologies that would be put together to vastly upgrade the U.S. Intelligence gathering abilities. The TIA system was noticed by the public, a hue and cry was raised, some Congress people harrumphed and caused DARPA to make some changes that appeared like they shut TIA down, but most of the projects that TIA was made of continued on.
Global Autonomous Language Exploitation (GALE) is simply the continuation of that research, under a different name. I am surprised arsTechnica missed the connection.
They describe it as automatic speech recognition of any spoken language, and automatic language translation of written text into English. The third component is a "distillation" engine which can answer questions and summarize information.
The stated goal is to capture all the public communications like TV broadcasts, newspaper articles, blog postings, etc, and sift them through a giant computerized mixmaster. The software would automatically sift through the massive quantities of information and they supposedly hope to find terrorists or other threats more easily. All communications have to be translated to English because of the lack of linguists to manually translate these communications.
It's a three step process. Stage one is to capture all capturable communications, and translate it into written text. Stage two is to translate the written text to English. Stage three is to sift through the massive piles of data and make sense of it.
If they only apply this program to public communications there isn't much to quibble about. Public communications are, well, public. And there is a lot of valuable intelligence that can be gathered by reading foreign newspapers and the like.
But are they likely to stop at public communications? This is the government that launched illegal wiretapping of U.S. Citizens in direct violation of federal law, and then lied to Congress.
It's well known the U.S. government has been tapping e-mail and other forms of "signals" intelligence for a long time. Supposedly these taps have to be associated with court orders and have to be targeted at specific individuals. But those legalisms assume the government agents are going to act with integrity and do only what the law allows them to do. Again, our current government administration has proven over and over they have no qualms about ignoring and violating U.S. law.