Terrifying — or just a giant scam? Big Tech is fueling an AI 'arms race'

Early in the 2020 presidential campaign, Democratic candidates Pete Buttigieg and Andrew Yang tried to build political momentum around the claim that the United States is losing ground in a new arms race with China — not over nuclear missiles or conventional arms but artificial intelligence, or AI. Around the same time, former President Trump launched the American AI Initiative, which sought to marshal AI technologies against "adversarial nations for the security of our economy and our nation," as Trump's top technology adviser put it.

Buttigieg, Yang and Trump may have agreed about little else, but they appeared to go along with the nonpartisan think tanks and public policy organizations –– many of them funded by weapons contractors –– that have worked to promote the supposedly alarming possibility that China and Russia may be "beating" the U.S. in defense applications for AI. Hawkish or "centrist" research organizations like the Center for New American Security (CNAS), the Brookings Institution and the Heritage Foundation, despite their policy and ideological differences in many areas, have argued that America must ratchet up spending on AI research and development, lest it lose its place as No. 1.

Just last week, the National Security Commission on Artificial Intelligence (NSCAI) published a sweeping 756-page report, culminating two years of work following the 2019 National Defense Authorization Act, asking Congress to authorize a $40 billion federal investment in AI research and development, which the NSCAI calls "a modest down payment." The commission also urged President Biden to reject the push for a global ban on AI-enabled autonomous weapons — a ban proposed by thousands of scientists and thought leaders in an open letter written in 2015.

Concerned about the threat of increasing AI sophistication in Russia and China, the commission warned lawmakers that America "will not be able to defend against AI-enabled threats without ubiquitous AI capabilities and new warfighting paradigms." It offered a laundry list of recommendations to put these paradigms into action, including a "Steering Committee on Emerging Technology" within the Defense Department, an accredited university designed to produce and recruit tech talent for the defense sector, and a ramped-up investment in semiconductor manufacturing designed to keep the U.S. "two generations" ahead of China.

One question, however, was not directly answered in the NSCAI's gigantic report or in all the think-tank policy papers that preceded it: Is this science fiction-flavored arms race against largely imaginary Chinese and Russian techno-weapons of the future really necessary? Is it remotely a good idea, or likely to improve the lives of any human beings on the planet? (Excepting, that is, those who stand to profit from it.)

Jim Naureckas, the editor of Fairness and Accuracy in Reporting (FAIR) and a frequent critic of military spending, told Salon in an interview that framing of AI development as an "arms race" is irresponsible, but in the larger sweep of history is also nothing new. "The whole military industry is driven by fear as a motivator," he said. "There's a logic to an arms race that's different from the logic of arms control."

After its release, the NSCAI report was greeted with a deluge of largely uncritical media coverage, most of it echoing concerns about the U.S. losing the "AI arms race" — a term not mentioned in the report itself, but certainly evoked by its framing.

"Unless America acts now," a Washington Post headline read, "China could trounce it in artificial intelligence."

"Which country is emerging as the global leader in AI?" echoed TechHQ.

"America wakes up to the China threat," chimed the Wall Street Journal.

As Naureckas pointed out, the notion that that the U.S. will soon fall behind its global competitors in military technology is a tried-and-true scare tactic, employed at various times in slightly different registers by both Democrats and Republicans. In reality, U.S. military spending remains mind-bogglingly high. For the 2020 fiscal year, the Trump administration approved a military budget of $738 billion, a $21 billion increase from the previous year and it passed with overwhelming bipartisan support, facing only 48 "no" votes in the House and eight in the Senate. In 2019, the militarized budget accounted for 64.5 percent of all federal discretionary spending.

The U.S. has 800 military bases on foreign soil, far more than any other country in the world. According to Military.com, America is the world leader in every significant category of military hardware, and has roughly 1.4 million active-duty military personnel. In 2020, the Stockholm International Peace Research Institute (SIPRI) found that the U.S. allocated more to its military budget than the next 10 nations combined. American military spending is about 2.7 times greater than that of China — which has a much larger population — and more than 10 times higher than Russia's, or that of any other single country.

Meanwhile, bureaucratic and operational waste within the defense budget abound. In 2016, for example, it was discovered that the Pentagon had buried an internal study finding that it had spent some $125 billion in wasteful business operations. More recently, it was discovered that the Pentagon's F-35 fighter jet program — which costed taxpayers somewhere in the neighborhood of $1.5 trillion — has been riddled with software glitches and operational failures since 2006, rendering an untold number of fighter jets (each one costing $100 million) not flight-ready.

In spite of all its administrative bloat and operational dysfunction, the military remains exceptionally well-funded. Why, then, would the NSCAI insist it needs billions more for a hypothetical arms race against badly underfunded opponents? The report's authors may tell a better story than the report itself.

Jack Poulson, a former Google employee who resigned over the company's plan to launch a censored version of its search engine in China, told Salon that profit motives is deeply entrenched in the NSCAI report.

"It should not come as a surprise that a commission packed with tech billionaires would call for increased intellectual property protections, oppose regulation (including on Lethal Autonomous Weapons), propose toothless ethics principles, and call for more federal funding of their industry," Poulson said in a statement.

Indeed, many commission members are past and present tech executives of companies on the fore of AI — companies that have much to gain from future contracting deals with the Pentagon.

The commission's chair, for example, is Eric Schmidt, the former CEO of Google, who remains — as Poulson pointed out — a major shareholder in Alphabet, Google's parent company. Google's head of AI, Andrew Moore, is also a member of the NSCAI.

Google already has an extensive history of working with the Pentagon. According to The Intercept, in a federally-funded $70 million program called Project Maven, Google developed "algorithmic warfare initiative to apply artificial intelligence solutions to drone targeting." The company expecting that revenue would steadily rise from $15 million to $250 million a year for such defense projects.

In April of 2018, however, 3,000 Google employees signed an open letter decrying the company's involvement in defense technology, a move that eventually led to Google's ultimate decision to back out of the deal. Schmidt strongly objected to Google's decision, calling it an "aberration" within the tech industry, which he felt was otherwise inclined to collaborate with the Defense Department. Former Undersecretary of the Navy Robert Work, the vice chairman of NSCAI, called Google's decision "hypocritical," using language that suggested a new cold war is already underway: "Anything that's going on in the AI center in China is going to the Chinese government and then will ultimately end up in the hands of the Chinese military."

Other members of the commission include Oracle CEO Safra Catz, Microsoft chief scientific officer Eric Horvitz, and Andrew Jassy, the future CEO of Amazon Web Services, all of whom received cloud awards as part of the CIA's Commercial Cloud Enterprise (C2E), as Poulson noted.

Oracle, Amazon and Microsoft, in fact, are currently involved in an acrimonious legal battle over a $10 billion cloud-computing contract called the Joint Enterprise Defense Initiative (JEDI). The deal was initially considered to be "gift-wrapped" for Amazon until Oracle butted in, alleging improprieties. In an odd turn of events, the Pentagon awarded the contract to Microsoft, prompting Amazon to sue the federal government for anti-Amazon bias, based on ex-President Trump's overheated rhetoric.

When it comes to securing Big Tech's enormous future contracts with the Pentagon, it appears that Jassy, Catz and Horvitz have set aside their mutual grievances for the time being

Other board members of NSCAI include Gilman Louie and Christopher Darby, who are the founder and vice president (respectively) of a CIA-funded nonprofit called In-Q-Tel, which invests money in private companies who are developing technologies that might be useful to the intelligence community. According to a Wall Street Journal investigation from 2015, half of In-Q-Tel's trustees were financially connected to private companies in which In-Q-Tel had invested.

Another board member, William Mark, a vice president of SRI International, has served on the Defense Advanced Research Projects Agency (DARPA), a government-run program that partners with a variety of private companies and research institutions to "make pivotal investments in breakthrough technologies for national security." DARPA has awarded SRI numerous contracts for the development of speech recognition, translation and, most recently, deep-fake recognition systems.

In other words, nearly everyone involved in preparing or supporting the NSCAI report would seem likely to benefit from the perception that the U.S. is falling behind other nations in vital defense technology.

The Defense Department, Poulson told Salon, "prefers to run the race as if it is losing — which happens to increase military budgets, justify post-government consulting careers and help tech CEOs oppose regulation."

It's only natural that government authorities would seek out industry experts to consult on AI projects — it's a fast-developing field that almost no one outside the tech world understands. Poulson wonders, however, "whether the U.S. will give human rights organizations — such as Human Rights Watch and the Campaign to Stop Killer Robots — as much of a seat at the table as it does tech billionaires."

The very fact that the NSCAI is stacked with panel members with an obvious incentive to weaponize new technologies raises the question whether there needs to be an AI "arms race" at all. That term, of course, harkens back to Cold War hysteria surrounding the threat of nuclear annihilation, which led U.S. lawmakers to grow unduly concerned with the "missile gap," a widely held misconception that the Soviet Union was outpacing the U.S. with superior ballistic missile capabilities. (As intelligence sources knew even at the time, the Soviet nuclear arsenal was in bad shape and much smaller than advertised.)

Arms control strategies, in fact, may be a more effective strategy in the AI realm, just as it was with nuclear missiles, especially given that America already collaborates heavily with China in AI research. As Graham Webster wrote recently in MIT Tech Review:

Unlike the US and USSR, in which science and technology developed on largely independent tracks, the US and China are part of a globally intertwined ecosystem. Even if the US and China cut off trade with each other, both countries would still have to worry about security risks from components, since risks along the supply chain exist everywhere.

For example, Alibaba, a tech giant on the forefront of AI, has multiple offices in the U.S., and Google AI chief Jeff Dean is an adviser at China's Tsinghua University, which opened an Institute for Artificial Intelligence in June 2018. Stanford University's Artificial Intelligence lab has a partnership with one of China's biggest retailers. In other words, an arms race in which the two nations are locked in silos of information, research and development is not just ethically dubious but logistically impossible.

Will China and Russia explore uses of AI in weapons of the future? Almost certainly — both countries have already signaled movement in that direction. But if American politicians and scientists want to maximize the potential of AI, framing its development in terms of an international "arms race" seems like a strategic and philosophical mistake on a huge scale. AI has the potential to revolutionize health care, education, climate science and many other fields — and those things all play a fundamental role in national security. But these new technologies will not make America more secure if they are understood as weapons of international combat.