More than a year after pulling out of a contract with the Pentagon that relied on technologies based on artificial intelligence to sort through drone videos, Google says it is ready to work with the Defense Department on a wide variety of applications that don’t involve weapons.
Google’s decision to engage with the Pentagon on non-weapons-related technologies stems from the company’s artificial intelligence principles published last year, said Kent Walker, senior vice president for global affairs at Google.
The principles rule out Google’s role in developing technologies that could cause harm or could be used for surveillance in violation of laws.
“It’s right that we decided to press the reset button until we had an opportunity to develop our own set of AI principles, our own internal standards and review processes,” Walker said last week at an event organized by the National Security Commission on Artificial Intelligence.
The decision to stop working with the Pentagon on the drone video contract was a “discrete” one and not indicative of a “broader principle or an unwillingness” to work with the Defense Department, he said.
The commission was created by Congress in the 2019 Pentagon policy bill to figure out how the Pentagon can harness artificial intelligence and related technologies for national security purposes. The panel is chaired by Eric Schmidt, the former chairman of Google’s parent company Alphabet, where he continues to be an outside adviser.
Google is working with the Pentagon on a “number of national mission initiatives,” Walker said, listing cybersecurity, health care, tools to identify deepfake videos, and other AI projects with the Defense Advanced Research Projects Agency, or DARPA, to ensure the “robustness of AI.”
“At the end of the day we are a proud American company and we are committed to the defense of the United States, our allies and the safety and security of the world,” Walker said.
Without a close partnership between the government, academia and industry, the Pentagon could never achieve global superiority in artificial intelligence technologies that are likely to be central to military dominance in the future, said Lt. Gen. John N.T. “Jack” Shanahan, director of the Pentagon’s Joint Artificial Intelligence Center, which is overseeing the military’s pursuit of such technologies.
“Even those for various reasons who view [the Defense Department] with suspicion, and reluctant to accept that we are in a strategic competition with China, would agree that artificial intelligence is key” to the future of national security, Shanahan said.
Google’s withdrawal in June 2018 from Project Maven, a Pentagon program that used the company’s artificial intelligence toolkit called TensorFlow to sift through thousands of hours of drone video, sent shockwaves through the national security world, raising fears that other technology companies could follow suit and handicap the Pentagon’s efforts.
Google’s decision came after thousands of Google’s engineers had protested its work with the Pentagon and wrote a letter to CEO Sundar Pichai saying that it shouldn’t be “in the business of war.”
The Pentagon turned to Google for help in developing AI technologies because it did not want to “reinvent the wheel,” and the company had the best minds working on machine learning algorithms, Shanahan said.
Although the Pentagon was “very pleased” with the results of the project and the cooperation it got from Google, “we lost the narrative very quickly,” Shanahan said.
Although the project was not aimed at developing a weapon, critics of the effort inside and outside Google labeled it as a weapons program, Shanahan said.
When Google withdrew from Project Maven, several top U.S. officials pilloried the company for choosing to withdraw from the Pentagon program while continuing its work in China.
The company’s withdrawal could put Americans at risk, former U.S. Deputy Secretary of Defense Robert Work said at the time. Work is now the vice chairman of the artificial intelligence commission.
Google and China
While Google employees were protesting the company’s engagement with the Pentagon, fearing that they were aiding in the development of potential weapons, the company was working in China on artificial intelligence systems, Work said. Beijing’s military is closely tied with commercial entities, and whatever technologies that are developed for commercial purposes could be used by the People’s Liberation Army, Work said.
The chairman of the Joint Chiefs of Staff, Gen. Joseph Dunford, also echoed fears about Google’s work in China. Any work by U.S. companies in aiding China’s development of artificial intelligence systems would “help an authoritarian government assert control over its own population” and “enable the Chinese military to take advantage” of U.S. technology, Dunford said in March this year. Dunford has since retired.
Google’s Walker said the company’s work in China was limited in its scope.
In 2010, large-scale cyberattacks from China aimed at Google’s operations within the People’s Republic resulted in loss of the company’s intellectual property, Google said at the time. “We learned a lot from that experience,” Walker said. “While many of our peer companies have significant AI operations in China, we have chosen to scope our operations in China very carefully.”
The company’s work in China is limited to “advertising and work supporting an open source platform,” Walker said.
Last year, Google engineers leaked details of the company’s work in China to develop a search engine code-named Dragonfly that would have allowed Beijing to censor and monitor its citizens. In July, Google said it had terminated its work on the project.
Chinese President Xi Jinping has pledged to spend $150 billion on artificial intelligence technologies with the goal of becoming a world leader in the field by 2030.