Will German SMEs follow suit?

As present as the terms "deep learning", "process mining" or "natural language processing" may be in the working world by now, German SMEs do not attach a particularly high strategic relevance to artificial intelligence (AI). For its current study, Deloitte surveyed more than 300 executives of medium-sized companies on the challenges, opportunities and risks of AI in SMEs.

64 percent attribute only medium to low relevance to AI for SMEs as a whole. In relation to their own business model, 58 percent also assign it a low level of importance. If the respondents look to the future, the opinion picture changes: 59 percent are convinced that the importance of AI will increase.

AI technologies are available in rows - but which ones are most important for SMEs? 45 percent of respondents see rule-based systems in first place in terms of relevance, followed by machine learning (41 percent) and deep learning (36 percent). This prioritization coincides with the potential that companies are most likely to expect from the use of AI, namely the automation of processes (77 percent), the efficient use of data (72 percent) and the acceleration of processes (66 processes).

Only 43 percent expect AI to lead to the development of new business models. The proportion of companies that develop applications themselves is correspondingly low at 26 percent - 34 percent of respondents use AI as a product, 30 percent as a service.

When it comes to the ideal areas of AI use in the company, a majority of 65 percent name the IT area as the predestined department. The participants also identify high potential for use in logistics and in materials and production management.

Surprisingly, it is precisely those areas in which established AI technologies can achieve comparatively rapid and uncomplicated efficiency improvements - such as in controlling or accounting - that most respondents rate as significantly less suitable for AI.

The ambivalent attitude of SMEs to AI is also reflected in their assessments of possible application risks: A full 47 percent of companies fear the wrong framework conditions for its use - such as insufficient data quality or simply a lack of data. Forty-five percent are concerned that effort and return do not match. Another 44 percent see the risk of non-transparent decisions or decision-making processes.

AI is a matter for the boss

Who wears the AI hat in the company? For just over half of the participants, organizational responsibility currently lies with departments below management level, for example IT or the operational areas that directly apply AI (16 percent in each case). Only one percent of respondents have their own AI department. By contrast, 48 percent already place responsibility with top management.

A similar dichotomy can also be seen when it comes to budget: It is true that for the entire sample, the median for AI investments is around 250,000 euros per year. However, 41 percent invest 500,000 euros or more, while 48 percent of the companies remain below the median. For eleven percent, the AI financial volume is even less than 50,000 euros per year. Nevertheless, 38 percent of respondents expect a strong or very strong increase in AI budgets in the future.