Friday afternoon, Gizmodo reported that Google would not renew Project Maven—its contract with the military to develop image-recognition AI—when it expires next year. A Google executive cited a wave of negative attention the company experienced from pursuing the project as a main reason for not renewing the contract. Nevertheless, top Google executives defended the company’s work on the project and didn’t say whether Google would be leaving the military business entirely. This was a telling omission, considering that Google is actively pursuing a multi-billion dollar cloud computing contract with the Department of Defense.
In other words, Google employees may have won on the AI issue, but what comes next? Is getting cozy with the Department of Defense the inevitable destiny of any company working on cutting-edge technology?
Friday’s announcement came on the heels of a New York Times article published Thursday about how Google was experiencing an “identity crisis” as a result of Project Maven. The contract was awarded to Google last year, but details of its existence only recently came to light. Last month, this prompted about 4,000 Google employees (about 5 percent of the company) to sign a petition demanding that Google leave the “business of war” and pledge to never develop any technologies for military use. The dissenting employees are concerned that the image-recognition technology, which will be used to better identify targets, could one day be used to kill, even though Google CEO Sundar Pichai promised the program was developing the AI for “ non-offensive” purposes.
Shortly thereafter, an email chain leaked to the press in which Fei-Fei Li, a leading Google AI researcher implored her colleagues to “Avoid at ALL COSTS any mention or implication of AI,” when talking to the media about Project Maven.
“Weaponized AI is probably one of the most sensitized topics of AI — if not THE most,” Li, who called the Project Maven contract a “big win” in the emails, wrote. “I don’t know what would happen if the media starts picking up a theme that Google is secretly building AI weapons or AI technologies to enable weapons for the Defense industry.”
Following the revelation of Project Maven’s existence, tensions quickly rose at Google over its pursuit of military contracts. About 12 employees resigned over the issue and DeepMind, Google’s flagship AI program, distanced itself from the project, citing a clause in its acquisition contract that prohibited its research from being used for military purposes. This employee push back led to optimism that Google “could end war,” as an op-ed published Friday in Scientific American put it, even though Pichai said at a company meeting last week that he thought Project Maven is for the best.
According to the Times, other Google employees didn’t see contracting with the military as a problem. After all, everyone else is doing it. In 2016, Microsoft won a $927 million contract to provide the Department of Defense with various products—it was the largest deal in the company’s history. Amazon recently announced it was getting into the policing business by selling facial recognition and tracking software to the United States’ increasingly militarized police departments. Amazon, Google, and Microsoft are all currently competing for a multi-billion dollar contact to provide cloud computing services to the Pentagon. Hell, even Apple got a slice of the military-industrial pie when it acquired PA Semi in 2008, a microchip company that manufactures processors used in missiles, and continued to support the manufacture of these chips for years after the acquisition.
The push back against this trend by a small percentage of Google employees is a laudable reaction. Protesting the militarization of theoretical and applied scientific knowledge is a timeworn tradition among researchers working on cutting-edge technology. The German chemist Clara Immerwahr killed herself after protesting her husband’s advocacy of weaponizing chlorine gas in World War I. The physicist Leo Szilard made a petition signed by 70 Manhattan Project scientists urging President Truman to see an atomic bomb test in action before considering whether to deploy it against people. Today, dozens of prominent researchers are actively working to stop the proliferation of autonomous weapons.
On the other hand, a distressing number of scientists have also been complicit in war. Just ask anyone at Los Alamos around, say, 1943; or Wernher von Braun before or after the war; or many, if not most of the researchers at MIT in the late 60s and early 70s. Even NASA, which was explicitly created as a civilian agency and mandated not to use its research for military purposes, has covertly cooperated in dozens of military intelligence operations.
The point is that if you look at history, many of our species’ greatest scientific and technological leaps forward in the last hundred years or so were in the service of causing death and destruction. Although technology has been important in every war, World War II was a turning point insofar as it sanctified the marriage between US technology companies and the military. It was this burgeoning relationship that President Eisenhower called out in his 1961 farewell address when he warned about the danger of letting the “military-industrial complex” grow unchecked.
America, however, did not heed Eisenhower’s warning. Between 1948 and 1989, the peak of the Cold War, the US government transferred nearly $10 trillion to a few dozen private military contractors. Moreover, in a 1992 study on the profitability of defense contractors from 1970-1989, researchers found that the profit rates of the top 50 defense contractors far outpaced those of comparable companies not contracting with the Pentagon. It seems there’s no business quite like the war business.
At around the same time that study was published, however, a revolution was underway in California. The rise of the World Wide Web resulted in an explosion of wealth creation in Silicon Valley and after the dot com bubble burst in the early 2000s, some of the startups that survived went on to become the most valuable companies in the world.
These companies—particularly Google and Amazon—were of a different breed than the megacorps of yore. They prided themselves on their image and their ability to generate massive amounts of wealth with almost entirely “benign” products. Rather than killing the Earth by extracting oil or killing humans by manufacturing missiles, these companies were developing search engines and selling books. These companies wanted to do the impossible: make boatloads of money while also making the world a better place. This attitude was canonized when Google emblazoned “Don’t Be Evil” on the walls of its headquarters. Yet recently, this motto was literally and figuratively scrubbed from the Googleplex. Amazon, meanwhile, fancies itself as the “world’s most consumer-centric company.”
One of the problems with capitalism is that it demands growth, and in this sense Amazon and Google are paragons of the capitalist mindset. From its humble origins as a search engine run in a garage, Google has grown to be a company conglomerate called Alphabet employing 90,000 people that is a world leader in AI research, open source technologies, self-driving cars, and literal moonshot ideas. Likewise, Amazon has morphed form a book retailer to becoming the world’s “everything store,” a logistics company, a drone research lab, and a cloud computing platform.
There are only so many markets that can be conquered and any company that grows large enough will eventually run up against the highly fortified walls of the military-industrial complex. In the age of the forever war that is fought as much IRL as in cyberspace, the complex is quite welcoming to technology companies, and many try to win Pentagon contracts as soon as they can. In fact, it’s amazing that Google and Amazon could grow as large as they have without the help of military cash.
Yet both companies know that once you’ve dominated civilian markets, the capitalist imperative to grow doesn’t just magically stop. They also know that technology is agnostic and can be just as easily used for good as it can for evil. Maybe Google could be just a little evil and take “only” $9 million in Pentagon money to develop AI for “non-offensive purposes.”
Google may be backing away from Project Maven, but even its short 18 month foray into developing AI for the Defense Department could lead to a whole host of lucrative military contracts. This was insinuated in a leaked email chain among Google executives about the company’s acceptance of the Project Maven contract. The company expected the revenue from its military work to grow from an initial $15 million to over $250 million per year. Indeed, just one month after the initial contract for Project Maven was signed, the Pentagon allocated an additional $100 million to the project.
In the coming weeks, Google CEO Sundar Pichai is expected to unveil a new set of guidelines for the company that will “stand the test of time.” Perhaps it will include a clause that commits Google to never selling any of its technology to the military. This would be a major ethical victory for Google employees who won’t have to worry that they are complicit in the deaths of strangers on the other side of the planet. I hope this is how it turns out, but I am very skeptical that this will be the result.
Google is not a person who must deal with the personal psychic drama of being responsible for the deaths of others, it is a company and a company’s foremost goal is making money. Of course, shareholders could bail on Google if it opts to accept military money as a way to force the company’s hand, but if anything the influx of military money is likely to strongly disincentivize a boycott by ethical investors because the military money will increase the value of the company. Even if Google won’t be building the Pentagon’s AI, some other company will.
This is the catch-22 latent in the ruthless logic of the market, and Google bowing out of one military contract isn’t going to change that.