German industry pushes the pace in artificial intelligence  

• Share of companies using AI rises from 9 to 15 percent within one year 
• Two-thirds see AI as the most important technology of the future 
• However, a large majority still ignore ChatGPT and generative AI

Berlin, 14 September 2023 - Artificial intelligence is experiencing a noticeable boost in the German economy. Meanwhile, 15 percent of companies use AI, compared to only 9 percent a year ago. At the same time, the proportion of those for whom the use of AI in their own company is not an issue has declined significantly: from 64 to 52 percent. A good two-thirds (68 percent) consider AI to be the most important technology of the future. This contrasts with 29 percent who see AI as a hype that is massively overestimated. The vast majority of 68 percent see AI as an opportunity for their own company, but one fifth (20 percent) see it as a risk. And one in nine (11 per cent) think that AI will have no influence on their own company. These are the results of a study commissioned by the digital association Bitkom, for which 605 companies with 20 or more employees from all sectors in Germany were surveyed. "It is true that only every fifth company that sees AI as the most important technology of the future also uses AI itself. But since this year, the German economy has been stepping up the pace on the topic of artificial intelligence. We have to work permanently on the discrepancy between knowledge and action," says Bitkom President Ralf Wintergerst. "In order for Germany to gain momentum in AI, companies must further increase their efforts and investments. At the same time, however, politics is also called upon to ensure that it does not remain with declarations of intent and strategies for AI promotion, but that the use and development of AI in Germany is significantly facilitated. In particular, the use of less sensitive data must be made easier.

Companies trust AI more with language - and don't use ChatGPT & Co.

Compared to the previous year, companies have more confidence in AI, especially in the areas of text and language. Thus, 84 percent believe AI has very great or rather great potential for text analysis and text comprehension in their own company, compared to only 74 percent a year ago. Speech recognition has great potential for 74 percent (2022: 66 percent), and 70 percent say so for generative AI for creating texts, images or music. Attitudes towards this new form of AI were surveyed for the first time. The majority of companies also see great potential for the other types of AI, such as facial recognition (68 percent, 2022: 61 percent), forecasting (67 percent, 2022: 71 percent), pattern recognition (67 percent, 2022: 66 percent) and image recognition (60 percent, 2022: 57 percent). "ChatGPT was an eye opener for many people and has also triggered intense discussions in companies. This gives the topic of artificial intelligence the broad attention it deserves," says Wintergerst. 

However, the insight into the importance of generative AI is not yet fully reflected in its use in companies. Only 2 percent are currently using generative AI centrally in the company, while a further 13 percent are planning to do so. Around a quarter (23 per cent) have no plans to use generative AI yet, but can imagine doing so in principle. However, just as many (23 per cent) cannot imagine doing so. And more than one in three companies (37 per cent) have not even considered it yet. "Generative AI offers exciting application possibilities in the most diverse areas, and at the same time the technology is already widely available today and easy to try out at low cost," says Wintergerst. "Really no company should put off discussing the use of generative AI. Those who wait today will soon have to work all the harder to catch up with the others."

Even if a company does not have a strategy for the use of generative AI, it may well be that the technology is used by employees in an uncontrolled and self-initiated manner. Half of the companies (51 per cent) firmly assume that this is not the case and that there is no one who uses generative AI for their own work. Another 23 per cent do not know exactly, but assume that no one from their workforce uses ChatGPT & Co. professionally. In contrast, this practice is widespread in 1 per cent of the companies, 7 per cent know of individual cases and 9 per cent are not sure, but assume that individual employees use generative AI professionally. So far, there are hardly any rules. Only one in a hundred companies (1 per cent) has established rules for the use of generative AI by individual employees, 16 per cent plan to do so in the future and 28 per cent want to do without it in the future. Around half (48 per cent) have not yet dealt with the topic at all.

The greatest potential of generative AI is seen in support for reports, translations or other texts (82 percent). This is followed at a considerable distance by tasks in marketing and communication (59 percent), for example in image creation, in the IT department (58 percent), for example for code generation, and in research and development (50 percent), for example for the evaluation of data. This is followed by use in production (44 percent), for example as an assistance system for machine control, in customer contact (39 percent), for example for processing enquiries, in the human resources department (26 percent), for example for communicating with applicants, in supporting internal knowledge management (23 percent), for example as a chatbot with access to company information, and in management (19 percent), for example in strategy development. Bringing up the rear is the legal or tax department (12 per cent), for example for drafting contracts. "Generative AI can already do much more than most companies give it credit for," Wintergerst comments on the assessment. "It is surprising that the potential in contact with customers, in internal knowledge management or in contract drafting and reviewing is still hardly seen. There are already practical applications here."

Further results of the survey show how divided German business is on the topic of generative AI. Thus, 42 percent of all companies say that those who use generative AI have a competitive advantage. 19 percent even expect generative AI to change their business model. At the same time, however, 51 percent think that generative AI looks spectacular but brings little benefit to companies. There are also very different assumptions regarding the impact of generative AI on employment. 30 per cent expect that the use of generative AI will lead to redundancies. Conversely, a similar number (29 per cent) think that generative AI will help to overcome the shortage of skilled workers, and 29 per cent also expect generative AI to make their own employees more productive. 

If you ask the companies that are actively involved in the use of generative AI what inhibits the use of generative AI in the company, regulation plays a decisive role. The biggest obstacle is data protection requirements (85 percent), but concerns about future legal restrictions (81 percent) and uncertainty due to legal ambiguities (76 percent) are also frequently cited. In addition, two internal reasons play an important role for many companies: lack of technical know-how (84 percent) and lack of human resources (78 percent). A lack of data (69 per cent) or time (68 per cent) are also frequently mentioned. In contrast, a lack of acceptance by employees is an obstacle in only one in two companies (50 per cent), and concerns about data falling into the wrong hands (43 per cent), a lack of use cases or a general lack of trust (42 per cent each) and costs (40 per cent) play a role even less frequently. 34 percent of the companies state that they are focusing on other future technologies. "When it comes to the use of AI, companies are not interested in subsidies, they need planning security and stable framework conditions, for example with regard to data protection," says Wintergerst.

A third believe they have missed the boat on AI - a quarter are investing

If you ask the companies where they see themselves in the use of AI in general - i.e. also beyond generative AI - compared to their competitors, only 2 percent say "at the top" and 13 percent "among the forerunners". At the same time, 43 percent see themselves among the laggards and 38 percent even believe that they have missed the boat. Wintergerst: "We are currently experiencing huge leaps in AI development. Even those who have hesitated in the past should at least consider using AI now." In the current year, every fourth company (25 per cent) wants to invest in AI; in 2022 or earlier, 41 per cent have already invested. In the coming year or thereafter, however, as many as three quarters of all companies (74 percent) plan to invest in AI.

Those who use AI assess the benefits and risks of the technology differently

Companies that already use AI see the greatest advantages in the fact that AI generally strengthens competitiveness (71 percent), while in the overall economy this is only mentioned by 44 percent. 52 percent of companies using AI emphasise the avoidance of human errors (33 percent in the overall economy) and 51 percent see that processes are accelerated (39 percent in the overall economy). The assessments are less clearly different when it comes to faster and more precise problem analyses (49 percent AI users, 48 percent overall economy) or expert knowledge through AI that would otherwise not be available in the company (41 percent AI users, 38 percent overall economy). The concrete effects on the business model are assessed significantly differently. 37 percent of AI users say that AI makes it possible to improve products and services, compared to 24 percent of all companies. And around one in four companies that use AI (24 percent) see completely new products and services emerging thanks to artificial intelligence - in the overall economy, the figure is only 16 percent. 

When asked about the greatest risk of AI, there is no difference between AI users and the overall economy. Both mention breaches of data protection regulations most frequently (80 percent AI users, 70 percent overall economy). However, there are clear differences in the other risks. For AI users, for example, errors in programming (70 percent, overall economy 57 percent) and liability obligations in the event of damage (69 percent, overall economy 47 percent) rank high. In the overall economy, after data breaches, companies are most concerned about new IT security risks (69 per cent, AI users 58 per cent) and application errors in AI use (67 per cent, AI users 34 per cent) . "It can be worthwhile to follow the forerunners in AI and benefit from their experiences. We should therefore strongly promote the exchange among companies," says Wintergerst.

Businesses want regulation that works in practice 

Companies want regulation from politicians - but with a sense of proportion. Almost three quarters (73 percent) are of the opinion that clear AI regulations can give European companies a competitive advantage. At the same time, around two-thirds (68 percent) are reluctant to use AI because they are worried about violating regulations. And around half of all companies (48 per cent) complain that excessive regulation is the reason why products like ChatGPT are not developed in the EU. "Companies don't want a Wild West in AI, but simple, understandable and above all practical regulation," says Wintergerst. "At EU level, work is currently underway on the AI Act. The AI Act must not become a technology killer that leads to similar massive legal uncertainties and a patchwork of different interpretations as is the case with the General Data Protection Regulation." At the federal level, it is also important that the announced AI action plan from the Federal Ministry of Research is closely interlinked with the other government projects. Wintergerst: "The federal government's new data strategy is of crucial importance. The data strategy must initiate the urgently needed paradigm shift in dealing with data, so that we can really use the potential of data for solving societal challenges, but also for the use of new technologies such as artificial intelligence."