This chapter provides a summary of the key findings from the international benchmarking of Germany’s AI ecosystem and from the analysis of achievements of its national AI strategy. It discusses Germany’s strengths, weaknesses, opportunities and challenges regarding AI development and use. The chapter concludes with recommendations to help steer AI policy in Germany to maximise opportunities and mitigate risks moving forward.
OECD Artificial Intelligence Review of Germany
1. Key findings
Abstract
Context
Germany’s 2018 national AI strategy, updated in 2020, aims for responsible growth and competitiveness in AI
In 2018, Germany was among the first countries to adopt a national strategy for AI. The strategy’s objective is to foster growth and competitiveness and ensure the responsible and trustworthy development of AI. Three federal ministries are responsible for the strategy’s development and implementation: the Federal Ministries of Education and Research (Bundesministerium für Bildung und Forschung, BMBF), of Economic Affairs and Climate Action (Bundesministerium für Wirtschaft und Klimaschutz, BMWK) and of Labour and Social Affairs (Bundesministerium für Arbeit und Soziales, BMAS).
The main goals of Germany’s AI strategy are: i) to secure Germany’s future competitiveness while making Germany and Europe leading locations for the development and application of AI technologies; ii) to ensure that AI use and development are responsible and focused on the common good; and iii) to embed AI in society ethically, legally, culturally, and institutionally through broad societal dialogue and active political efforts.
In November 2019, the German Federal Government published an interim report with results after the first year of implementation. In December 2020, Germany updated the national AI strategy in response to developments. In particular, the COVID-19 pandemic, environmental sustainability, and climate protection were brought to the fore, alongside the importance of European and international collaboration on AI. The EUR 3 billion budget originally allocated to the strategy was later announced to be increased to EUR 5 billion. The three leading ministries evaluated the national AI strategy in 2023-24. In this context, they invited the OECD to support their analysis through an international benchmarking of the German AI ecosystem.
The geopolitical, technological, and economic context changed significantly since 2020
As Germany seemed to be recovering from the COVID-19 pandemic, Russia’s war of aggression against Ukraine began and Germany’s dependence on energy imports turned into an energy-security threat. Energy prices increased substantially, fuelling inflation, and affecting firms’ competitiveness. Several bottlenecks caused first by the pandemic then by the war disrupted global value chains and brought to light Germany’s energy security vulnerability. Domestically, Germany faced the challenge of a demographic change, resulting in high healthcare costs, and contributing to labour shortages.
At the same time, AI technologies continue to evolve rapidly. AI is widely considered the next general-purpose technology, given its potential to transform entire sectors and enable the discovery of innovative new business models and products. The worldwide number of scientific publications on AI nearly quadrupled over the past five years, and venture capital (VC) investment in AI more than doubled over the same period. In November 2022, the release of the conversational AI language model ChatGPT-3 brought generative AI to the attention of the broader public, raising awareness of the technology’s potential and its risks.
Against this background, assessing Germany’s international AI performance and evaluate the achievements to date of its national AI strategy is timely to recognise its strengths, identify weaknesses and ultimately shape its AI vision for the years ahead to both maximise opportunities and mitigate risks.
This report draws on data and on insights from interviews with key German stakeholders in the field of AI to provide an international benchmarking of Germany’s AI ecosystem and progress in implementing its national AI strategy. It discusses Germany’s strengths, weaknesses, opportunities and challenges in AI development and use, and recommends approaches to help steer AI policy in Germany moving forward.
This review is organised according to six pillars: minds, research, transfer and applications, the world of work, policy and regulatory frameworks, and society. The review also includes three sector spotlights on AI in the public sector, healthcare, and environmental sustainability.
Overview of strengths and weaknesses, opportunities, and threats
Figure 1.1 summarises the key findings stemming from the analysis, while the next section provides overall recommendations. Chapters in the report discuss in detail the findings by each area of the analysis.
Strengths
In the unfolding AI race, Germany is equipped with unique competitive assets: research excellence coupled with a priority to develop AI in a human-centred way, as well as Germany’s international clout, create fertile ground for AI development and adoption.
German research pioneers AI development in all key AI sub-disciplines, positioning the country as a leader in global AI trends and standards. A robust network of public and private research institutions characterises Germany's AI research landscape. Institutions like the Max Planck Society, Fraunhofer Society, and leading universities are renowned for their rigorous scientific output, ranking highly in international comparisons. German institutions focus on both fundamental and applied research. This research excellence has become a magnet for talent, contributing to global AI knowledge. German researchers’ academic outputs, as measured in AI research publications, consistently rank among the top worldwide. This prolific output demonstrates Germany's importance in advancing AI knowledge and applications.
Germany’s solid compute capacity underpins the research sector’s ability to research AI models. With the emergence of generative AI and foundation models, AI research is increasingly computationally intensive. Germany has numerous supercomputers that can be used for training AI models and meet growing demand for AI compute from the research and private sector. Though the broad accessibility of compute resources still needs to be assessed, compute capacity represents a major competitive advantage. A majority of other European countries have insufficient AI compute.
Germany has adopted a human-centred approach to AI. The German Federal Government, through its AI strategy and policies, emphasises the development and use of AI for social good. It invests in projects that promise to benefit society, such as healthcare and environmental sustainability. These policy priorities align AI advancements with societal objectives, ensuring that technology serves the public interest broadly. AI developments should prioritise ethical considerations, societal needs, and individual rights. To do so Germany emphasises AI applications that enhance human capabilities, respect privacy, and ensure fairness. In line with the European Union (EU) AI Act, German policies and research initiatives reflect this human-centred ethos and help set a global benchmark for responsible AI development.
Keeping humans at the centre of AI development has translated into a generally positive attitude towards AI in the workplace. Germany has actively consulted with workers on AI adoption in the workplace. By addressing concerns and incorporating feedback, German stakeholders ensure that AI adoption is aligned with employers and workers’ needs and values and enhance the effectiveness and acceptance of AI solutions.
Beyond making AI ethically sound, Germany is uniquely positioned to make it environmentally sustainable. As the world faces a climate crisis, driving AI as part of the green and digital “twin transition” has become crucial. With recent reports highlighting both the significant ecological footprint of developing AI systems and the large potential of leveraging AI for rapid decarbonisation, Germany features AI and sustainability ecosystems that could drive the sustainability of its AI models and applications. Beyond reducing carbon emissions, this approach may spearhead the development of environmentally sustainable AI and prove to be a competitive advantage.
Germany is breaking grounds on AI policy initiatives, both domestically and internationally and it is exporting its vision for the future of AI abroad. Germany is carving out space for policy experimentation. To better support the regulatory management of AI, Germany initiated a series of measures aimed at enhancing regulatory experimentation. These include the Spaces for Learning and Experimentation (Lern-und Experimentierräume) and the creation of regulatory sandboxes, as outlined in its national AI strategy. Germany is expected to enact a federal regulatory sandbox law by 2025. Germany is taking deliberate steps to lead AI regulation and standardisation, to foster the twin-objectives of trustworthiness and competitiveness. Initiatives such as the establishment of a national AI standardisation roadmap, engagement in international standardisation organisations, the introduction of an “AI trust label”, illustrate the country’s proactive approach to ethical AI development. These measures are symbolic of Germany's dedication to shaping a competitive AI industry within the European framework, in line with the EU’s regulatory ambitions.
Weaknesses
While AI hype is peaking globally, Germany's economy shows reserved enthusiasm: although AI is widely considered to be the next general-purpose technology that will provide major competitive advantages to businesses, limited availability of VC and of other AI enablers, alongside weariness for innovation have resulted in only modest adoption.
German companies that fail to adopt AI risk losing global competitiveness and remaining vulnerable to supply chain disruptions. Despite its solid foundation in AI research and development, Germany’s AI adoption across key industries is more fragmented. While public and private research institutes have made significant strides in AI development, the overall pace of industry adoption remains relatively low compared to European frontrunners. In particular, the manufacturing sector, which plays a key role in the German economy, is slower than other sectors in taking up AI applications. This slow adoption can be attributed to various factors that range from industry-specific challenges to broader economic and policy environments. Like many other countries, Germany faces an AI skills bottleneck. Although its educational institutions excel in producing high-quality graduates, the demand for AI expertise far outstrips the supply.
AI roles remain predominantly male dominated. Mirroring a global issue, there is a wide gender gap in AI roles, particularly in leadership positions within Germany. This disparity restricts Germany’s ability to meet labour market demands and to ensure that AI solutions cater to diverse populations and to avoid perpetrating biases.
While employers complain about talent bottlenecks, the scale of the issue is largely unknown. National skills assessment studies lack a specific examination of available AI skills, generating an opaque picture with regards to skills demand and perceived shortages. This highlights a need for a bespoke and precise evaluation of AI skill requirements needed for designing educational policy responses and on-the-job re‑skilling programmes. Germany’s flagship vocational programme (Ausbildungsmodell) needs an AI update. In the absence of a precise AI skills assessment, updating vocational training regulations to incorporate AI content is advancing slowly. Yet, the flexibility of Germany’s vocational programmes and employers’ ability to proactively integrate AI-related skills into their programmes make vocational training a uniquely agile vehicle to address the AI skills gap.
Following a decade of economic growth, Germany may be a victim of its own success as economic actors become increasingly weary towards innovation. Across several sectors, there is a low understanding of AI’s potential applications, coupled with a mindset that past economic success pre-empted a need for innovation. This lack of awareness impedes the adoption of AI, as businesses and public-sector entities fail to recognise how AI can help address critical business challenges or enhance operational efficiency. A limited appetite for risk hampers the growth of German AI champions. Germany's conservative investment mindset, characterised by a reluctance to engage in high-risk ventures, significantly impacts the AI landscape. The limited availability of VC for AI startups stifles innovation and slows down the commercialisation and diffusion of AI technologies. This risk aversion manifests itself both in the world of VC, but also within civil service when procuring AI solutions. Faced with a lack of funding, high potential start-ups may choose to relocate abroad.
Policy and governance mechanisms remain partly fragmented but are crucial to steer beneficial AI adoption. Leadership at a high political level could signal widely that AI is a national priority. In the public sector, there is a lack of clarity on roles and oversight regarding AI implementation. Government ministries each manage initiatives with their own set of AI usage rules, but a centralised approach is missing. This absence of a cohesive strategy and co-ordination function results in lack of AI leadership in government and potential duplication of efforts across different ministries. Germany's federated governance creates bottlenecks, hindering uniform AI adoption across states. The centralisation of AI expertise within national ministries results in expertise bottlenecks at the state level, and this may impact implementation of regulatory sandboxes due to the limited know-how of Germany’s federal states (Länder). Further complicating matters is a fragmentation of authority across states (such as in healthcare or data protection), each developing their own initiatives at varying paces. This disjointed approach not only slows AI integration but also makes it challenging to adopt AI nationwide uniformly.
Civil society needs a bigger seat at the table. While the German Federal Government engages stakeholders in AI policy design, such as during the 2018 national AI strategy formulation, representation is limited. Seats at the policy-making table are typically occupied by social partners (Sozialpartner:innen), leaving other stakeholder groups, including civil society, minority and environmental protection groups on the fringes. This imbalance in participation does not sufficiently capture the full spectrum of perspectives needed to ensure that AI policy is designed inclusively.
But the challenges are technical as much as they are structural. Open data accessibility and digital infrastructure are limited. Developing and training AI applications requires access to large, high-quality, and detailed datasets while ensuring the security of these data. Yet, government datasets, crucial for training and refining AI algorithms, remain predominantly closed. Industrial data are similarly largely underutilised. Similarly efficient and widespread connectivity is a prerequisite for smooth data transfer and availability, while Germany's patchy digital infrastructure, especially in rural areas, limits AI penetration. Within the healthcare sector strict adherence to data protection laws and an overly cautious stance towards innovation limit progress, despite the sector’s desire to enhance access to health data for research. The lack of standardised data protocols and interoperability creates significant barriers to integrating AI, with researchers facing considerable administrative challenges in accessing health data.
Opportunities
A number of opportunities could help navigate these new global realities.
Current hype may be the perfect storm for promoting AI adoption in Germany. The current momentum and surge in AI interest can support driving actionable strategies across key sectors including manufacturing, public services, healthcare, and green initiatives. AI adoption could be driven through incentives, awareness drives, and supportive policies.
This may help re-design industries for the age of AI. Crafting targeted policies for AI to transform critical industries could support AI penetration across sectors. By focusing on sector-specific AI strategies, funding transformative projects, and supporting R&D in key sectors, industries could be made ready for a global economy shaped by AI.
Expanding the circle of consultation could create a more inclusive policy design process. Germany's strong social partnership tradition creates fertile ground for an inclusive AI policy environment. By creating platforms and processes for diverse stakeholder engagement, including citizens, Germany is well placed to ensure wider representation in AI policy development.
Germany is strongly positioned to be a frontrunner in AI policy: Germany can support and fund initiatives that promote standards for trustworthy AI internationally.
From sustainable AI to AI for sustainability: Mapping and measuring the environmental footprint of AI and leveraging Germany’s AI and environmental sustainability ecosystem can help decarbonise energy, transport, industry, and agriculture. Through the formulation of policies that back AI research for sustainability and offer industry incentives for eco-friendly AI adoption.
Threats
As global AI competition heats up, vigilance is needed in the face of rapidly evolving societal risks. In trying to lead a future shaped by AI, governments worldwide are investing heavily in AI capabilities. But as AI development accelerates, it risks sparking disinformation, economic inequality, or bias perpetration, threatening public trust.
The global frontrunners have positioned themselves. German industry may lose competitiveness at the global level if it does not adopt trustworthy AI timely. Widely regarded as the next general-purpose technology, AI has the potential for contributing to large scale economic growth and social innovation and could reduce the impact of labour shortages on the economy. Failure to adopt AI could result in German industry losing its competitiveness edge, as rivals harness AI to improve operations, cut costs, and innovate.
Germany must leverage the EU to keep pace in the global AI race. As the competition for AI dominance intensifies, global superpowers are marking their territories, with the People’s Republic of China (hereafter “China”), the United States and an increasingly ambitious India making a firm stance on the global stage. To remain globally competitive, Germany must harness the collective scale of the EU to bolster its technological capabilities, innovation potential, and avoid trailing behind.
Thus far AI has been viewed favourably. But its rapid advancement and ensuing threats may put this trust to the test, triggering public resistance to wider adoption. In a year filled with high profile elections, deepfakes and AI-generated disinformation will bring emerging threats to the centre stage. In 2024, an unprecedented number of voters will go to the polls across 68 nations. Previous cases of deepfakes and disinformation have dominated news headlines. In what has been coined the “Year of Elections” the risk of AI being used to create and spread misinformation and disinformation is particularly high. This misuse could undermine the integrity of democratic processes and weaken public trust in AI.
Anxiety about automation and economic disparities are increasingly prevalent as topics of public concern. AI’s benefits risk accruing disproportionately to higher-income groups, intensifying economic inequality. Such a scenario could contribute to the erosion of the German middle class, posing threats to societal cohesion and overall prosperity.
AI systems, if not carefully designed and monitored, can perpetuate existing biases. To date, Germany has not faced a case of large-scale bias perpetration compared to other countries, but such a high-profile case could turn the tide against public perception on AI. Securing vigilance and countering these risks will be imperative.
The use of AI systems in the workplace can raise risks for the safety of workers and their rights. Germany needs to implement measures to monitor and manage these risks.
The massive use of computational resources for AI systems raises sustainability concerns. While AI only represents a fraction of overall environmental impacts from digital technologies, the proliferation of AI applications and the exponential dynamic of AI compute requirements call for implementing measurement standards and expanding measurement efforts to assess and mitigate the energy, water, and resource impacts of AI compute infrastructure.
Figure 1.1. Strengths, weaknesses, opportunities, and threats (SWOT) analysis
Key recommendations
Germany is well-positioned overall to keep pace in the AI global competition, but to meet its AI ambitions it could leverage its international clout and economic weight to implement reforms
While vision and strategic co-ordination at the highest level of Government are key, it is crucial to have a robust technological, data and infrastructure foundation, a competitive and technically savvy workforce to diffuse AI across sectors, and society’s trust. To meet existing and upcoming challenges, Germany could focus its attention on:
Keeping sight of the bigger picture. Germany’s national AI strategy could be updated to target sectors where AI is expected to have the strongest impact. This involves identifying and channelling efforts towards specific challenges and creating a roadmap for integrating AI solutions in these sectors.
Securing buy-in at the highest political level and harmonising policy efforts in adjacent areas. Germany lags in key enablers of digital transformation and AI uptake, such as connectivity infrastructure and open-data availability. While the Government issued strategies to advance these fields, they are managed by separate entities lacking co-ordination authority. Given the interdependence of digitalisation, data and AI, commitment at the top political level (i.e. the Federal Chancellery) to orchestrate these policies is required to unlock opportunities, ensure policy effectiveness, and avoid conflicting or duplicating efforts.
Leveraging AI to cut red tape in the public sector, reach healthcare and environmental objectives, and secure industrial competitiveness. AI can transform entire sectors, including manufacturing, which are key for the German economy. AI solutions can also accelerate the green transition, increase public-administration efficiency, and mitigate challenges of an ageing population – key issues for Germany.
Involve and inform citizens and workers. Germany’s approach to AI, including for AI adoption in the workplace, is human-centred and transparent. Continued acceptance of the technology will hinge on stakeholders’ capacity to engage in meaningful dialogue, with a nuanced picture of the technology itself and the impact of its adoption, including in the workplace. Labour-market resilience will need to be cultivated, including enhancing AI expertise and awareness among workers and employers.
Attention should focus on three strategic enablers:
Widening access to data. Data are the raw material for developing AI systems, but in Germany their availability and use are limited, due to cautious interpretation of data protection legislation, and public and industrial data remaining siloed. Data-protection authorities should proactively develop data-sharing protocols for personal data use in sectors, such as healthcare. Furthermore, the Federal Government could mandate government agencies at all levels to publish non-sensitive data in open registries. Frameworks that facilitate the responsible sharing of industry-specific data for AI development should be reinforced. The EU and national policy frameworks recently enacted go in the right direction in fostering data as a key enabler for AI innovation. Implementation of these policies will be key.
Nurture the next generation of AI entrepreneurs. While government financing to start-ups is available in pre-seed and seed rounds, AI start-ups face challenges accessing domestic or European capital to scale up. More risk capital could be made available by revisiting the legal framework for capital-collecting institutions, while targeted government financing could enhance AI start-ups in their growth phase. Finally, the Federal Government should revisit its procurement guidelines to allow AI start-ups to more easily sell to the public sector.
Build and scale a globally competitive computing infrastructure. Germany should assess its current AI compute infrastructure to gauge existing capacity and potential gaps for matching stakeholder demands. This assessment would help ensure that capacity is available to implement Germany’s AI strategy, produce world-class frontier AI research and develop sector-specific solutions. To ensure that national AI infrastructure is inclusive and accessible to all stakeholders, a portion should be made accessible for use by AI start-ups and SMEs.
Methodology
The review draws on qualitative and quantitative data collected in two phases. The first phase (November 2022-February 2023) gathered data on the five pillars of the 2020 national AI strategy (Minds, Research, Transfer and applications, Regulatory Framework, and Society) and an additional pillar: The world of work. Data were sourced from the OECD.AI Policy Observatory and from the OECD’s AI Work Innovation Productivity and Skills (AI-WIPS) project, supported by the BMAS, complemented by data from third parties and ad hoc data collection for specific areas and indicators. Upon completion of the first phase, preliminary results and recommendations were presented to the three federal ministries responsible for implementing the AI strategy.
The second phase of the study (July-October 2023) delved deeper into specific areas: transfer from research to commercial applications, and the use of AI in the public sector, healthcare, and environmental sustainability. The discussions pertaining to these sector “spotlights” (and the ‘Transfer and applications’ pillar) are thus more extensive than other sections. For the second phase of the review, over 90 individual interviewees provided their perspectives on the advancement of AI in several German sectors (Annex B). These discussions took place through a combination of in-person and remote interviews, supplemented by desk research and four site visits to Germany. The review combines results from both phases and discusses the status of AI in each of the pillars, policies, and challenges to address.