Google Cloud’s soon-to-be head of AI will have his work cut out for him if he wants to convince disgruntled employees he won’t follow in his predecessors footsteps and develop AI for the military. He’s the co-chair of a defense and security think-tank alongside one of the people responsible for the creation of Project Maven.
Google Cloud announced earlier this week that Carnegie Mellon’s Andrew Moore will replace AI head Fei-Fei Li by the end of the year. Moore is returning to Google, having previously been a director with the company from 2007 through 2014.
In May he was appointed co-chair of the Center for a New American Security’s artificial intelligence task force, a group dedicated to determining how the US government should respond to the threat of AI.
At the time of his appointment Moore said:
Central to all of this is ensuring that such systems work with humans in a way which empowers the human, not replaces the human, and which keeps ultimate decision authority with the human. That is why I am so excited by the mission of the task force.
Proponents of Google’s work on Project Maven were quick to point out the company wasn’t developing weapons systems. According to reports, the company was teaching military developers how to use TensorFlow (an open-source machine learning platform) to analyze drone footage.
Despite such assurances, hundreds of employees protested and at least a dozen quit over the company’s involvement. Google eventually announced it wouldn’t seek to renew its contract to work on Project Maven.
Whether Moore’s commitment to keeping “ultimate decision authority” in humans will be enough to assuage employee concerns over the company’s partnerships with the Pentagon remains to be seen.
In other Google AI news, a handful of employees, including Senior Google Scientist James Poulson, have shown themselves the door, reports The Intercept. Their departure comes in protest of the company’s efforts to build a censorship engine, code-named Dragonfly, to appease the Chinese government.
It’s worth noting that state-owned media outlet China Securities Daily, via Reuters, reported that Google and the Chinese government were not working on Dragonfly, citing “relevant departments” as its source — the Chinese government has long maintained it doesn’t engage in censorship.
Google has a wishy-washy history with Chinese censorship. Back in 2006 it sought inroads into the Chinese economy and agreed to the laws of the land. But, that all changed a few years later when the company had a change of heart. Co-founder Sergey Brin, speaking about Google’s refusal to continue to engage in Chinese censorship in 2010, told Der Speigel:
Having come from a totalitarian country, the Soviet Union, and having seen the hardships that my family endured–both while there and trying to leave—I certainly am particularly sensitive to the stifling of individual liberties.
The company didn’t make it a decade before the allure of Chinese money forced it to reconsider its stance on censorship. When The Intercept broke the story, it indicated Google’s Dragonfly project had already been demonstrated as a near-finished app to the Chinese government.
The demo would seem to show that Google planned to deliver on Dragonfly. But, maybe CEO Sundar Pichai doesn’t mind following projects all the way through to the very point of fruition before stopping to make ethical considerations — the board must love that.
It looks like we’ll just have to trust the company when it says it’s not involved in developing AI for the military or censorship engines for China.
We reached out to Google for comment but didn’t immediately receive a response.