Meet China’s AI Competition: Germany’s Drive toward AI Innovation Needs Sound Policy and Partnership with U.S.
Last week, German Federal Minister for Economic Affairs and Energy Peter Altmaier suggested the creation of a large private company for Europe to pursue marketable artificial intelligence (AI) applications. This is yet another in a string of exercises in Germany and across Europe to reap the economic benefits that AI promises to bring. Despite Berlin’s emergence as a vibrant hub for AI development in recent years, AI research and use in Germany has been slow. German politicians openly acknowledge that the country has lagged behind AI powerhouses the United States and China, which are home to 40 percent and 11 percent of all AI companies, respectively. Germany trails with a distant 3 percent.
In November 2018, the German government published its national AI strategy. Dubbed “AI Made in Germany,” the strategy aims to boost AI funding, expand data pools, and promote AI research. That is accompanied by other objectives including anticipating AI development’s impact on the labor market and setting ethics standards for data access. Germany’s ambitious AI strategy not only encompasses key elements that are also parts of U.S. and Chinese AI strategies, but also goes a step further to call for technical standard setting and international cooperation. However, as the country lacks a vibrant AI ecosystem propelled by large amounts of venture capital or tech giants, policymakers in Berlin are seeking effective government policy intervention to stimulate AI growth.
China as Model for Sound AI Policy
Enter China. The Chinese government has successfully created a number of large tech companies and fostered a competitive AI ecosystem through its policy directives. Easy access to data pools of over one billion Chinese internet users; enormous government research funding; and wide-spread adoption of native tech applications, such as WeChat, for public services enabled Chinese tech companies to scale up rapidly. In the process, companies like DJI and SenseTime have become industry leaders in providing market-competitive products such as consumer drones and computer vision AI. Those policy efforts have only intensified. Together, China’s Made in China 2025 strategy and 2017 New Generation Artificial Intelligence Development Plan (AIDP) form the core of Beijing’s vision to dominate the global AI competition, anchoring AI as the key to China’s economic and security interests. Furthermore, the central government’s support forms a floor, not ceiling, for AI development. Local governments often promise additional funding and policy incentives for companies across industries. Last May, the city of Tianjing pledged $16 billion to set up an AI fund, far surpassing Germany’s €3 billion pledge for AI innovation before 2025.
To learn from its competition, the German government needs policies that help to create an AI ecosystem that is driven by exponential growth, increased accessibility, and AI integration.
To learn from its competition, the German government needs policies that help to create an AI ecosystem that is driven by exponential growth, increased accessibility, and AI integration. Robust government funding for AI research at universities and institutions is a necessity, and it must be accompanied by sufficient venture capital for AI startups and financial incentives for companies looking to invest in AI applications. Enhancing data availability, both by offering access to government data and encouraging the formation of shared industry-specific data pools through institutional cooperation, is fundamental for machine learning. Helping company executives to understand the potential of AI and imagine possible uses of AI applications for their businesses will then connect AI innovation with Germany’s upcoming industrial transformation. These and many other policy remedies have been adopted by the Chinese government and they can all be found in Germany’s AI strategy.
But if the German government hopes that simply following these guidelines will propel it to the status of global AI leader and usher in a new era of growth and prosperity, it will be disappointed, not least because the country’s comparatively limited resources for AI innovation or the shared risk-averse approaches by German companies and among EU countries could still hinder progress. China’s vision for global AI dominance will both systemically challenge Germany’s economic interests and rewrite the rules of future AI competition.
China as Disruptor in the Global AI Race
China’s AIDP holds AI as a strategic technology that will enhance national competitiveness and protect national security. Together with Made in China 2025, the document lays out the Chinese government’s plan to reduce its dependence on foreign technologies through AI development. Large amounts of resources and attention have been directed to this national strategic priority and as a result, China is now leading in AI research publication, patents, and venture capital investment. Moreover, Beijing plans to further strengthen its competitive position by addressing weaknesses in China’s AI ecosystem, which would drastically tilt the global economic and geopolitical order in its favor. That process could also diminish the available tools for Germany to defend its interests. And in any case, Germany or Europe will not be able to mount a challenge without the U.S.
Beijing plans to further strengthen its competitive position by addressing weaknesses in China’s AI ecosystem, which would drastically tilt the global economic and geopolitical order in its favor.
At the top of Beijing’s agenda to achieve technology independence is the semiconductor industry. Processing massive data sets for machine learning requires exceptional computing power generated by accelerator chips. Presently, China relies heavily on chips that are mostly designed in the U.S. and manufactured in Taiwan, which restricts rapid scaling up of computational power for cutting-edge AI research and exposes entire industries in China to potential foreign sanctions, as observed in the U.S. chip sales ban to ZTE. The Made in China 2025 strategy prescribes a quota of 80 percent for domestic semiconductor manufacturing as a share of domestic consumption by 2030.
One the one hand, Chinese companies and research labs have made significant progress in high performance computing infrastructure, with companies such as SenseTime boasting some of the largest aggregate computing power in the world. On the other hand, Chinese manufacturers foresee low-cost AI chips combined with innovative software and some of the world’s largest data sets expanding their market share globally.
That in turn helps to solve what Chinese policymakers and industry experts see as a second weakness, the lack of ability to set technical standards. Despite the fact that Chinese companies are extremely good at adapting breakthroughs in AI technologies with genuinely innovative product ideas and short development cycles, they have not produced many Standard Essential Patents (SEPs). These core intellectual properties enable the adoption of international standards and would further solidify Chinese companies’ global market positions and make resisting Chinese products much more costly and difficult, as European governments already found out in the case of Huawei’s 5G infrastructure equipment. Dominant market position and ownership of key AI SEPs would also greatly diminish the effectiveness of foreign embargos on components.
Another trend in China’s AI development is the emergence of dual use AI technologies. Chinese military leaders are convinced that AI systems will be fundamental in future intelligent warfare and the Chinese Ministry of National Defense has set up new AI research bodies in recent years, aiming to build AI-enabled weaponry that would allow China to leapfrog military technologies and give Beijing a more effective way of power projection. Chinese companies have been selling advanced surveillance AI technologies and increasingly autonomous robotic weapons to other countries. China’s DJI manufactures civilian-purposed drones as well as stealth combat and surveillance drones. Similar to the case of Huawei’s dominant role in global 5G infrastructure, these developments should be taken seriously by Germany’s security community.
The U.S. as the Key Partner in Germany’s AI Strategy
There have already been attempts from the German government to cooperate with EU partners to push for AI innovation. In July 2018, Angela Merkel and Emmanuel Macron announced plans for a new Franco-German artificial intelligence institute. Nine months later, not much has happened. The EU Commission’s AI plan from December 2018 calls for EU-wide collaboration, but ignores that most member states do not have strong AI ecosystems and that the ongoing drive to further regulate the tech industry at both the EU level and the national level could seriously hinder private investment and commercial adaptation in AI technologies.
The U.S. is a different kind of partner in the global AI race. An abundance of venture capital investment in AI startups and expensive AI research projects founded by large tech companies in Silicon Valley have not only created a vibrant AI ecosystem, but have also consistently stood at the forefront of AI innovation. U.S. tech giants such as Google, Microsoft, and Facebook have created the world’s most popular machine learning frameworks, whose code libraries and AI software development tools are used world-wide, including by Chinese AI researchers. Their SEPs help to set industry standards for AI applications. The U.S. is also the leader in global semiconductor design.
Instead of copying the playbook of already established AI superpowers and starting from scratch, Germany should make good use of strong U.S. advantages in AI innovation through a close partnership.
Instead of copying the playbook of already established AI superpowers and starting from scratch, Germany should make good use of strong U.S. advantages in AI innovation through a close partnership. Attracting U.S. AI venture capital and talent would increase the resources for AI development in Germany. Acquisition of AI startups by larger companies in both directions enhances knowledge sharing and accelerates AI adaptation. Technical interoperability for both commercial and security purposes helps to set global industry and ethical standards, preserve global AI market share from competition, and maintain policy tools for effective government interventions.