The use of artificial intelligence (AI) in firms in the German economy is developing very dynamically. In 2021, around one in ten firms in Germany used AI technology, almost twice as many as in 2019. Almost all firms collaborate with third parties, although the nature of the collaboration varies greatly between them. As a rule, the higher a firm’s AI maturity level, the less it needs to rely on external competences in this area. These are the results of a study by ZEW Mannheim on behalf of the Federal Ministry for Economic Affairs and Climate Action. The study looks at the role of collaborations with third parties and the necessary internal prerequisites to be able to use AI more effectively in firms.
Among larger firms, the proportion of those using AI is higher than among smaller ones, with the highest proportion found in the service industry. The majority of firms using AI have their own competences in the technical handling of data and the development of software solutions. The study shows that 64 per cent of firms aim to optimise their AI competences through employee training. At the same time, half of the firms are investing in improving the technical prerequisites for AI use.
Firms using AI increasingly work with cooperation partners
However, many firms lack the technological and organisational capabilities to use AI effectively. That is why almost all firms in Germany (90 per cent) rely on cooperation with other companies or institutions when it comes to AI. Only 26 per cent of firms using AI technology develop their AI applications themselves, while about a third (32 percent) of firms work with a cooperation partner and the majority (41 percent) mainly leave the development to third parties. Especially older and larger firms and those with a medium AI maturity level work with cooperation partners. In particular, they seek collaboration in the areas of machine learning and automation. In contrast, firms with high AI staff intensity or high AI maturity predominantly do their own development. “Firms mostly work with a cooperation partner when they have their own AI competences but are not among the ‘top’ AI users in terms of maturity and intensity,” explains study author Dr. Christian Rammer. Collaboration takes place not only in the area of R&D and the development of AI applications, but also in the area of data access and analysis, IT infrastructure and the integration of AI into internal processes. The aim of cooperation with third parties is to speed up the implementation of AI projects and to gain access to complementary knowledge and technologies.
Technical interfaces a key challenge
Looking at the different areas the cooperation partners come from, the study shows that IT firms and software developers make up the largest group with 71 per cent, while 55 per cent of firms cooperate with universities or research institutions. Other cooperation partners include customers and other companies in the industry, as well as specialised AI start-ups. Cooperation partners are more often located outside their own region in Germany (73 per cent) than in the same region (57 per cent). However, firms also enter into international collaborations, especially those with a high level of AI maturity.
The question of technical interfaces is a key challenge for the cooperating AI firms, as is the lack of compatibility of software solutions. “Establishing industry- and user-specific solutions and developing standards can counteract these problems,” says Christian Rammer. Another challenge is the lack of a common understanding of AI. This problem can be solved by introducing awareness-raising measures and actively disseminating knowledge about AI. The nationwide expansion of a strong IT infrastructure as well as legal regulations to improve access to data and the possibilities of using them can support the rapid diffusion of AI in firms. A sufficient supply of skilled labour is also an important element in creating appropriate framework conditions to promote AI: “Especially when it comes to non-technical capabilities, many firms still do not see themselves as sufficiently prepared. For example, integrating AI-related content in degree programmes can help in this regard, as can a larger offer of advisory services,” says Christian Rammer.