By Steven Hill, Ethics of Algorithms-Bertelsmann Stiftung, November 16, 2018
Why Germany needs to focus on AI that benefits humanity
After a long period of planning and incubation, the German government recently agreed on its new strategy for Artificial Intelligence (AI) development. The plan certainly includes a number of forward-looking steps, but it is also deeply flawed, and for two reasons: First, like the EU’s AI plan, the German plan tries too hard to copy the Silicon Valley model, which neither the EU nor Germany can replicate. Second, it will likely launch a divisive AI race within Europe and further undermine an already divided union. A better plan would involve following a multilateral, pan-European approach that would pave the way for a golden European digital age.
On the positive side, the German blueprint commits €3 billion for research and development and announces the establishment of twelve regional centers which together will form a national AI network. Among other things, the network will be able to focus on assisting small and medium-sized companies – Germany’s successful Mittelstand of “hidden champions” – with implementing AI. Financial resources will also be tapped in an effort to make it easier for Germany to attract and keep both innovative companies and top research talent. In this way, it is thought, Germany’s economy and businesses can remain competitive and successful well into the future. If that kind of plan has worked in Silicon Valley, surely it can work in Germany and the EU?
But Germans and Europeans have a distorted view of Silicon Valley. The architects of the German and EU plans apparently do not realize that the Land of Platform Capitalism is a unique ecosystem that cannot be replicated. That’s because much of Silicon Valley’s technological success has its roots in massive military spending that has subsidized Silicon Valley for decades. Since the 1930s and World War II, the San Francisco Bay area has been a major site of U.S. government research in cutting-edge technology. Responding to the Soviet Union’s first Sputnik space satellite, in 1957, U.S. President Dwight Eisenhower established the National Aeronautics and Space Administration (NASA) and turned to Fairchild Semiconductor from San Jose, California which, at the time, was the only company in the world that could make transistors. In the 1950s and 60s, nearby Stanford University became a research magnet that attracted top tech talent focused on semiconductors and solid-state technology research. The university fast became the heart of an R&D ecosystem involving companies like Bell Telephone Laboratories, Shockley Semiconductor and Xerox PARC that was focused on – and funded by – military priorities.
In 1969, the Stanford Research Institute (today known as SRI International), operated one of the four original nodes that comprised the military-funded ARPANET, the first version of the internet. Other familiar technologies, such as Apple’s voice-recognizing personal assistant Siri, the World Wide Web, Google Maps, internet search and automated vehicles, each began their birth stories as projects of DARPA, the Defense Advanced Research Projects Agency. More recently, DARPA has been funding research in AI and machine learning, subterranean exploration, deep-space satellites, high-performance molecules and improved GPS.
With that stable base of R&D investment in technology companies, venture capital funders have had the luxury of rolling the dice with their private money on new companies and technologies. Seven out of 10 Silicon Valley startups fail and 9 out of 10 never earn a profit, but the ones that make it through the investment casino – such as Google, Amazon, Facebook, Apple – have become hugely profitable and dominant market-shapers.
As President Donald Trump has started trade wars with allies and competitors alike, technology has become an increasingly important component of economic as well as military strategy. China also is now engaged in a similar state-sponsored strategy, having long approached its technological development as an extension of its overall nationalist agenda. Insider talk is increasingly focused on this new kind of “AI Nationalism.” Especially considering AI’s military applications, this kind of scientific competition between nations only exacerbates great power tensions and replicates existing global inequities. Much is at stake in finding an alternative paradigm.
Too little, too late?
Is Germany or the EU prepared to launch such a high-risk, military-subsidized and expensive strategy? It seems unlikely. In fact, the recent announcements about German and EU investment for AI development, while a welcome sign of progress, also reveal a further weakness.
For starters, the amount of funding being proposed is very modest. The EU has pledged some €1.5 billion by 2020 and announced a vague plan for another €20 billion to be tapped from private and public sources. Germany’s announced investment of €3 billion won’t be fully activated until 2025; when combined with planned AI investment from France (€1.3 billion) and the UK (€1.2 billion) this will amount to another €5.5 billion in investment.
But U.S. and Chinese companies in the private sector – not to mention their respective governments – will be investing considerably more. While exact amounts are often top secret information, China’s Alibaba is reportedly investing €13 billion over the next three years in cutting-edge technologies, with a particular emphasis on AI. China’s government has announced plans to build a domestic AI industry worth €133 billion in the next few years and to make the country the global leader for AI by 2030. For starters, the government is investing nearly €2 billion just to build a national AI technology park in Beijing. In the United States, the private sector, led by Google, Amazon, Apple, Microsoft and Facebook, already has invested billions of euros in AI; IBM has invested over $13 billion in Watson, its AI-based supercomputer. Meanwhile, the U.S. military spent in 2017 about €6.5 billion on unclassified projects for AI and related fields (big data and cloud computing), which marks a 32% increase since 2012. But it spent billions more on classified R&D, though the exact figure for this is unknown.
That’s real money. Currently, spending on technology research and development in the EU as a whole averages 2.08 percent of GDP, which is lower than that seen in China and the United States, as well as the OECD average of 2.40 percent. While announcements for more investment and spending are a positive sign, it seems unlikely that Germany alone, or even when combined with other EU members states, will ever match the deep pockets of US or Chinese governments and companies, or embark on a state-sponsored strategy of this nature. Investment and R&D in Germany and the EU therefore need to be more strategic.
The missing AI vision
And that’s a problem, because “strategy” requires “unity,” and when it comes to AI development, Europe is anything but unified. Quite the contrary, rather than launching a pan- European strategy that effectively coordinates the deployment of limited resources, the German plan reinforces a bunkering down mentality among member states in the pursuit of national interests. France also has embarked on a Franco-focused strategy, which means both member states are launching their own separate version of a misunderstood Silicon Valley approach.
Much like Airbus has prevented needless nationalist-based conflict in aircraft production, it makes no sense for Germany to compete with France and the United Kingdom for the title of “AI Leader of Europe.” The inevitable division will result in a rivalry over limited resources and further cement the raw dominance of China and the United States. Yet in Berlin, Paris and London, all eyes seem focused on achieving a national claim to AI dominance.
A better plan would be to establish a new pan-European AI hub, like the one proposed by a group of European scientists from Germany, France, the Netherlands, the United Kingdom and Switzerland. Named the European Lab for Learning and Intelligent Systems, or ELLIS, the proposed AI institute would establish major centers in several member states, each one employing hundreds of computer engineers, mathematicians and other scientists with the expressed aim of keeping Europe at the forefront of AI research.
As with the creation of the Internet itself, government-funded innovation would create the AI basic infrastructure that private companies and university labs could plug into. An adequately-funded EU-wide effort also would help counteract an appalling brain drain in which Silicon Valley companies are poaching top EU talent and buying European startups at an alarming rate. Facebook, Google and Amazon have gobbled up one European company after another, with Google acquiring British company DeepMind Technologies for some €530 million in 2014. Germany’s new AI plan proposes incentives to reduce this flight of human capital, but given natural market limitations due to the limited global use of the German language, it makes more sense to offer incentives on an EU-wide basis. A strategy of this kind would allow a unified approach that maximizes opportunities and fosters successful companies.
A European hub for AI also could help solve another dilemma – the lack of available data for research. AI is fed by huge amounts of data, which is what the algorithms comb through with raw computer power and speed to find patterns that humans could never see. Data is its sustenance – and Europe has no coherent source of it, because it doesn’t have the big, data-intensive, commercial companies like China’s Alibaba and Tencent, or Google, Facebook or Amazon, each of which are vacuuming up all the data they can from consumers and creating “data-opolies.”
A pan-Europe hub could re-conceptualize private information as “social data” that is protected as part of the commons. This would help ensure the availability of open-source datasets and open-source codes that would be available for nations and companies to access for research, as long as they agree to conduct their research on behalf of the public interest and in accordance with guidelines that protect privacy. This approach reflects a more progressive vision for data use than allowing individuals to become “data shareholders” who are paid a pittance for permitting Facebook, Amazon and Google to mine and monetize our personal information.
In short, AI should become a global public good, like GPS, HTTP, airline regulations, and other technology protocols that have been developed. Germany and the EU have an opportunity to become transformative players by leading such a cooperative effort.
Yes, Europe needs to beef up its own AI capacity, but a more crucial strategy would focus on leading the world in a more human-centered direction. That includes forging a research agenda that is less commercially driven, or nationalist- or military-based, and more centered on solving the world’s greatest challenges. The right kind of EU-wide AI development would focus on benefiting humanity rather than exclusively on commercial, for-profit applications and would avoid the facilitation of national silos and an emphasis on high-tech military weaponry. As U.S. history shows, it is easy to fall into a pattern in which military funding becomes a main driver of technological advancement.
The right strategy will make a unique contribution and transform the EU into a global leader. The member states must swim or sink together on this.