Expert opinion: AI puts German economy back on the path to productivity

Contents

Artificial intelligence (AI) is associated with extensive political, social and economic expectations. This also applies to increasing productivity with the help of technology. The Cologne Institute for Economic Research (Institut der deutschen Wirtschaft, IW) has now investigated the question of what could be done in this area in Germany on behalf of the Joint Committee of German Industry and Commerce. Hubertus Bardt, Managing Director of Science at the IW Cologne, summarized the results of the report at the presentation in Berlin: AI is leading the local economy "back on the path to productivity". However, the technology will not bring about a new productivity miracle.

Since 1991, the performance of the economy has declined every decade, particularly due to the bursting of the dotcom bubble and the coronavirus pandemic, explained Bardt. In 2024, it was only 0.4 percent. This applies not only to Germany, but is an international trend. Taking into account the potential effects of AI, the IW projection shows average annual productivity growth in Germany of 0.9 percent for the years 2025 to 2030 and 1.2 percent for the following period up to 2040.

A good two thirds of the expected upward trend over the next 15 years can be attributed to AI, explained Bardt. This is "the biggest technological lever we have". At the same time: "If we don't do it, others will." The increase is a "significant order of magnitude": around 3 percent of gross domestic product could be generated in the next ten years. At the same time, Germany must also make more of other production and location factors. This applies, for example, to the tax and energy system as well as the infrastructure. "We must prevent the number of working hours from slipping too much," emphasized the professor. In view of the ongoing demographic change in working life, "longer working weeks and longer working lives" should also be discussed, despite technology.

Cost and time savings, but not a job killer

According to the study, the use of AI "has the potential to realize productivity gains at the level of the individual workplace". However, the empirical evidence indicates that exploiting this potential is not automatic and that "specific conditions for success" can vary depending on the company, organizational area and workplace. So far, there is little to suggest that "the productivity potential of AI conceals labor-saving technical progress". The researchers do not consider chatbots and highly specialized AI systems in production to be job killers.

According to the experts, there is consensus in the assessment"that AI leads to efficiency gains in the form of cost or time savings". However, the framework conditions, such as the AI Act in the EU, play a decisive role. The implementation of the AI regulation must be "manageable" with the help of concrete implementation aids, checklists and proven practical examples. Furthermore, the digital infrastructure and the availability of skilled workers form the decisive environment for AI in Germany.

The number of professions that are "close to AI" will have risen to a third in Germany by 2023, explained Bardt. Specialists account for 13.5 percent, specialists for 6.7 percent and experts for 11.7 percent. At the same time, only 1.5 percent of the latter are not AI-savvy. Overall, AI is unlikely to be able to "replace a person that we will no longer have".

AI use still restrained

However, the use of the technology is less pronounced than the potential, criticized the economist and business economist. In 2024, around 58% of dependent employees had not yet worked with AI. Only 10.3 percent considered themselves to be experienced with AI. The use of AI is also worthwhile for employees: AI users tend to earn more, which is partly due to their qualifications, but their salaries are also growing faster. According to the IW employment panel, 45% of AI-related employees have also increased their own work performance.

Bardt reported that 80 percent of local companies stated that they did not know the business model for AI. He recommended that they simply get started and have something programmed for a manageable amount of money. The rest would take care of itself and everyone involved could have fun with such solutions. The expert also advised creating the foundations and acceptance for the use of AI, rethinking the culture of innovation and establishing ecosystems, especially between research, start-ups and industry.

Just get started

A German company does not necessarily have to build a large language model, Bardt illustrated. It is not always necessary to think only in terms of large AI projects. Medium-sized companies in Germany generally have their own IT infrastructure, servers and data, which can be used to scan tenders and filter specific requirements with the help of AI. Considering certain risks of misuse, guard rails or stop signs are necessary. However, AI is "not just a high-risk thing".

DIHK Managing Director Helena Melnikov appealed to companies to "get a bit more out of it" by using the technology and gain "that decisive edge". DIHK division manager Rainer Kambeck described the necessary data infrastructure as "challenging for some SMEs". In addition to tax cuts, low energy costs and fundamental deregulation are needed. Polina Khubbeeva from the BDI welcomed the EU approach with AI factories in principle. However, the associated funding sums were not even enough to cover the hourly wage. In general, Brussels should focus more on competitiveness, which could be helped by a new Cloud Development Act, for example.

(olb)