Is Machine Translation Essential to U.S. National Security?

Machine translation and, more broadly, natural language processing, have become an important part of the global technological leadership race in artificial intelligence. So much so that this fast evolving field of technology has now been identified as being potentially relevant to US national security.

On November 19, 2018, the US Bureau of Industry and Security, Commerce issued an advance notice of proposed rulemaking (ANPRM) and has given the public one month to comment on whether to include a number of emerging technologies on a so-called Commerce Control List (CCL).

The CCL already contains “many sensitive technologies” that are perceived as important to national security, for example because of their potential application in “potential conventional weapons, intelligence collection, weapons of mass destruction, or terrorist applications” if they fall into the wrong hands. Or because keeping them under lock and key “could provide the United States with a qualitative military or intelligence advantage.”

The Bureau has now issued a shortlist of what it refers to as Emerging Technologies, which “have not yet been evaluated for their national security impacts,” to be considered for inclusion on the CCL. At a later date, another ANPRM will be issued for so called Foundational Technologies.

Smart Dust and Machine Translation

Under Emerging Technologies, the Bureau lists a total of 14 general categories containing nearly 100 individual areas of technology. The lists runs the gamut from genetic engineering, quantum computing, adaptive camouflage, smart dust, or brain-computer interfaces.

And there, under the category “Artificial intelligence (AI) and machine learning technology,” machine translation makes an appearance as an example of natural language processing.

The public is asked to provide input on “the status of development of these technologies in the United States and other countries; the impact specific emerging technology controls would have on U.S. technological leadership,” and other considerations including “potential end-uses and end-users of the technology.” The deadline for comment is December 19, 2018.

The level of restrictions has yet to be determined, but any technologies included on the CCL will at least require a license for export “to countries subject to a U.S. embargo, including those subject to an arms embargo.”

(vii) Natural language processing (e.g., machine translation)

The proposal has received wide media coverage from major outlets including the Financial Times, the Washington Post, Bloomberg and Yahoo News, and has also been picked up by legal newsfeed Lexology and technology blog Apple Insider.

Many commentators are making the link between the proposed control of exports and “growing trade friction with Beijing” (Yahoo News) while the Washington Post reported that tech experts have been left “fearful that it could result in greater market barriers for companies doing business in China,” which is one of the countries subject to a US arms embargo.

Indeed, natural language processing research is somewhat of a microcosm for the US-China tech battle. The likes of China’s Alibaba, Baidu, iFlytek, and Sogou are going head to head with the Silicon Valley tech giants in neural machine translation research. Baidu and Sogou, along with Google, Microsoft, Amazon, and Facebook all put their names to this year’s EMNLP, the world’s largest conference of natural language processing, while European sponsors were in noticeably short supply.

Comments on the proposal, of which eleven have been received but none published at press time, can be submitted via email or to the Federal eRulemaking Portal.

Download the Slator 2019 Neural Machine Translation Report for the latest insights on the state-of-the art in neural machine translation and its deployment.