Suspending or shaping the AI policy frontier: has Germany become part of the AI strategy fallacy?
by Maximilian Schuessler
When it comes to national AI strategies, governments should harness first-mover advantages wherever possible. Despite continuing calls from the German economy and leading research institutes as well as Germany’s great potential to succeed in AI, the German Federal Government has delayed its plans to publish a national AI strategy until November 2018.
Germany’s aptitude for engineering and IT has long been part of the global technological frontier. German businesses are increasingly mining data-sets, up-skilling for the digital economy and becoming more adaptive to the disruptive effects of new technology. This has also included developments in Artificial Intelligence (AI).
Germany has a vibrant AI and digital economy. German automotive companies are leading in patents for self-driving vehicles, holding 52% of all patents in the field. In research and development (R&D), German initiatives like CyberValley and the German Research Center for Artificial Intelligence (DFKI) specialise in areas of AI such as robotics and driverless cars. Hochreiter and Schmidhuber’s (1997) invention of Long Short Term Memory (LSTM) was a major breakthrough in AI technology. LSTM units can retain information in the long term, and are used for speech recognition, robot control and business prediction tasks by large tech giants such as Microsoft and Google.
AI offers enormous potential for the German economy. The country’s strong manufacturing sector, dominated by chemicals and the automotive industry, can particularly benefit from AI and enhanced digitisation. According to a recent study, Germany’s economy expects increased annual market growth of almost 3% (gross value added) and a 29% increase in labour productivity by 2035 due to AI technologies. Researchers acting as advisors to the German government are working to put AI on the R&D policy agenda.
On the policy side, however, the German government ranks unfavourably compared to similar countries when it comes to setting the rules of the game for AI (see Figure 1). Despite its white papers on the future of work and on cybersecurity (2016), Berlin has been reluctant to set a national AI strategy. National AI strategies are important because market-based responses are insufficient to address the range of ethical, security, educational and regulatory challenges associated with AI. Moreover, the umbrella organisation of the German digital economy, Bitkom, has rightly argued that political leadership in AI is necessary if Germany wants to thrive at the new economic frontier.
Moreover, the uptake of digital government services and progressive approaches to open data in Germany have been slow. Clear AI strategies help governments to harness technology themselves, help others to make investment decisions within the economy, and improve effectiveness and efficiency in delivering digital services. As Oxford Insight’s Government AI Readiness Index illustrates, many forward-looking countries are ahead of Germany in this regard.
To compete with China, the US, France and the UK and to accelerate its innovation, Germany will need to build a comprehensive AI strategy and establish a broader base for AI research that translates research into applications. Interestingly, the German Government has already responded to a burgeoning subfield of Germany's economy: a full report on the ethics and regulation of automated cars is already in place.
Despite its initial plans to provide a first report on its AI strategy before the summer break, the German Government has delayed the publication of its AI strategy. Instead, it has announced some (fairly vague) priorities for its pending AI strategy.
Ensure the long-term attractiveness of Germany as an AI hub and promote German competitiveness;
Develop a comprehensive blockchain strategy to harness its full potential and mitigate risks;
Provide favourable frameworks for startups;
Engage in labour market transformations and continuous education to promote job safety and adaptability.
There are some key insights that can be drawn from Germany’s approach to AI.
Avoid the fallacy ‘the-best-moment-for-an-AI-strategy-is-yet-to-come’
Despite the pervasive effects of AI on all levels and Germany’s legacy of effective labour market policies, the German Government is apparently waiting for a ‘convenient’ moment to develop their AI strategy. However, to provide an effective response, AI policy strategies should be as lean and agile as the technology that they aim to embrace. Given the fast pace in the field and the potential for first-mover advantages, AI strategies must be audacious, prompt and forward-looking. They should provide a vision rather than a mere snapshot of the status quo.
AI requires more than just an economic response
Germany’s dual apprenticeship and vocational training system allows trainees to receive their education partly from companies. This currently accounts for an economy-driven AI response in terms of skills and labour-market transformation. Germany’s unique system affords high adaptability and explains why the country came second on the Economist’s Intelligence Unit Automation Readiness Index despite the absence of a national AI strategy. However, we believe that harnessing AI necessitates a comprehensive policy response, covering policy areas that go beyond the private sector, i.e. ethics, research and development, government and public services, data & digital infrastructure, and capacity, skills and education. Thus, AI policy strategies should complement AI responses from the private sector.
National AI strategies must address implementation on regional and local levels
Addressing AI comprehensively will not only require a federal German AI response, but collaboration and policy differentiation on the level of various states. AI strategies are mainly developed on a national level. However, their implementation frequently applies locally, including, for example, chatbots in hospitals, public transport in smart cities, and digital tax administration. Thus, in addition to a federal strategy, Germany’s various states will also need to outline AI responses. These should prioritise areas where individual states hold primary responsibility, such as education.
National governments should pursue strategic partnerships in AI
As well as considering action at the local level, some action on AI will need to take place through collaboration between multiple states. Franco-German initiatives such as the Joint European Disruptive Initiative J.E.D.I., and the proposal to create a joint AI agency between France and Germany, emulating the American Defense Advanced Research Projects Agency (DARPA), demonstrate that AI is a field ripe for strategic supranational cooperation. The Franco-German partnership seeks to harness AI’s full potential, overcome risk-aversion in investment, and increase the scalability of AI investments.
The EU has also announced a common EU AI framework to empower EU-wide research, narrow the innovation gap with the United States and China and guarantee more sovereignty vis-à-vis international tech companies. An EU-wide approach has the potential to bring AI technologies to market at a larger scale, while also pooling data sets essential for building algorithms and training AI technology. This is why AI policy should include measures on a national level - to account for country-specific contingencies - and internationally in order to harness scale and strengthen sovereignty against AI monopolies.
AI and digitisation will bring considerable benefits to the German economy and society. To keep pace with its international peers, Germany must follow the French example, become bolder, and avoid delaying AI as a priority within its political agenda. Building on Germany’s highly adaptive and vibrant AI economy, policymakers need to provide a comprehensive and agile framework that, harnesses Germany’s full strength and scale, and addresses challenges for which marked-based strategies are insufficient. A visionary AI strategy will also set the tone for the pending EU-wide AI agenda and allow Germany to engage more effectively in the French-German AI partnership.
Oxford Insights advises governments internationally on how to approach AI, including most recently advice to the Government of Mexico on its AI strategy. Earlier this year, we have published an overview of existing and pending AI policy strategies. In the autumn, we will publish our annual Government AI Readiness Index.
Would you like to talk about how to create a comprehensive, innovative and ethical AI strategy? Email firstname.lastname@example.org or email@example.com.