As the European Union’s Artificial Intelligence (AI) Act comes into force and two years after ChatGPT was first published, bidt is conducting a representative online survey of 1,500 internet users to analyse the current use of generative AI in private and professional life as well as the attitudes towards the regulation of generative AI among the population in Germany.
November 2024 marks two years since the first publication of the ChatGPT AI system. Since then, ChatGPT has become the fastest-growing internet application to date (Der Standard 2023), making generative AI a technology that has been adopted by the general public within a very short space of time (Oliver Wyman 2024). Concurrently, the technology has sparked some controversy, triggering numerous social and political debates. Many of these debates focus specifically on the professional application of generative AI, addressing its impact on tomorrow’s working world and exploring ways to maximise the technology’s potential and mitigate its risks (IMF 2024). In this context, hastily published guidebooks often promise a “revolution of everyday work” along with maximised productivity and efficiency. The launch of ChatGPT and similar models has raised questions of how such basic AI systems should be handled within the context of the European Union’s regulatory framework to regulate AI – commonly referred to as the AI Act (netzpolitik.org 2023). This resulted in a compromise that places basic AI models in a separate risk class within the regulatory structure. Consequently, these models face less extensive conformity and transparency obligations than in other risk categories. Effective from August 2024, the AI Act aims to provide a comprehensive regulatory framework for AI use and development in Europe while taking into account the potential risks of the technology (European Commission 2024).
In light of the new regulation and the high expectations for the technology, many questions arise: What is the current extent of the prevalence and acceptance of generative AI in private and professional life? What effects of generative AI do people already perceive today and what do they expect or fear looking ahead? What guidelines do companies establish for the use of the technology by employees? How much importance do people place on the regulation of generative AI and what is their view of the new AI Act?
To answer these questions, the Bavarian Research Institute for Digital Transformation (bidt) commissioned the market research institute Reppublika Research & Analytics to conduct a quantitative online survey. For bidt’s second survey on the prevalence and acceptance of generative AI, a total of 1,500 internet users aged 18 and over, including 768 employees, were surveyed in Germany from 30 September to 15 October 2024. The data was weighted to be representative of internet users in Germany according to age, gender, education and federal state (in accordance with b4p 2023 specifications).
Majority of the population has not yet heard of the EU AI Act
According to a Eurobarometer survey conducted in 2023, the vast majority of people in Germany state that EU measures have an impact on their daily lives (European Union 2023). Such an effect is particularly likely in the case of the EU AI Act, which is expected to have a lasting influence on the use and development of AI technology in Europe. However, the results of this survey show that a majority of 64% of internet users in Germany have never heard of the EU AI Act. Only 36% are aware of this legal act. Awareness of the regulation is higher among people with a higher level of education and those who have already used generative AI themselves compared to those with a lower level of education and individuals who have no personal experience of using generative AI.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
Nonetheless, 56 % of respondents agree that the use of generative AI should be more strictly regulated. Among respondents who have already heard of the AI Act, only slightly fewer respondents agree with this statement (52 %) than among respondents who are not familiar with the Act (59%). These results indicate that there is a general lack of awareness of the AI Act and therefore insufficient communication about the regulatory measures. They also suggest that the regulation’s desired impact is only partly recognized as effective.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
This is also reflected in the assessments of the AI Act among those who have already heard of it. For example, 53% of this group agree or fully agree with the statement that it will be difficult to implement the AI Act in practice. Only 14% hold the opposite view. Furthermore, 47% believe that the AI Act will not be able to keep pace with the rapid development of AI technology. Among those aware of the law, 44% agree or fully agree that the AI Act imposes significant bureaucratic burdens on many companies.
However, alongside these critical assessments, there are also optimistic expectations: 32% of respondents believe that the AI Act will strengthen the competitiveness of the German economy, while only 18% expect the opposite. 44% predict that the protection of personal rights will be strengthened by the AI Act. Opinion is divided on whether the AI Act will negatively affect the innovative strength of German companies. While 27% agree or fully agree with this statement, 29% disagree.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
Overall, our results indicate a rather reserved assessment of the AI Act. In particular, the possible additional bureaucratic burden combined with the uncertain effectiveness of the regulation is viewed critically.
Use of generative AI at a similar level to the previous year
Almost three quarters (73%) of all German internet users state that they have already heard of generative AI. 35% have already used it.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
Overall, the use of generative AI is roughly the same as in the previous year, although the share of respondents who have already used the technology multiple times has increased slightly (Schlude et al. 2023).
A closer look at the current use of generative AI reveals that the proportion of users decreases with age. Almost two thirds (64%) of 18 to 29-year-olds state that they have already used generative AI, while this proportion is only just under 15% among respondents aged 65 and above. The results also suggest a link between the use of generative AI and the level of formal education: the proportion of users among people with a low level of formal education is 25%, while the share among highly educated people is twice as high at 50%.
Among all users of generative AI, 43% had used the technology at least once a week in the four weeks prior to the survey. As many as 17% used generative AI at least almost daily during this period.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
ChatGPT is by far the most widely used system among users of generative AI. Overall, 81% of the users of generative AI state that they have already used this chatbot. Google Gemini, used by 30% of generative AI users, ranks second but trails far behind ChatGPT. Microsoft Copilot, which is also based on ChatGPT, ranks third with a share of 26%.
Workplace adoption of generative AI remains low
In the medium to long term, the use of generative AI is expected to lead to major changes in professional environments, as ChatGPT and co. have the potential to fulfil tasks that previously could only be performed by humans. The following section therefore takes a closer look at the prevalence of generative AI among employees and their experiences with the technology.
As many as 79% of those in employment have heard of generative AI – just over a fifth have not. 25% of all employees have already used generative AI in a professional context.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
Slightly less than half (46%) of employees who had already used generative AI at work reported using the technology at least once a week in the four weeks prior to the survey. Meanwhile, more than a fifth (22%) used generative AI almost daily or more frequently.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
Similarly to our results regarding all internet users, there is also a link between frequency of use and both age and level of education among employed respondents. Employees who are younger and have a higher level of formal education use generative AI more often for work.
Compared to the previous year, there is also no significant change in use of generative AI at work. In particular, generative AI has not become more established in the professional environment since the last survey – contrary to the expectations of some (Schlude et al. 2023).
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
Professional users of generative AI primarily assess their use of the technology as positive. In each case, over 60% state that they were able to use the results in a meaningful way, saved time at work and received helpful support with difficult tasks. A similarly high share agrees with the statement that the use of generative AI had inspired new ideas. 20 to 25% have had mixed experiences with using the technology. Only around one in ten users disagreed or strongly disagreed with these statements. However, the fact that the respondents had mostly positive experiences with the use of generative AI in these contexts does not mean that professional users do not take a critical view of the results of the technology. In fact, 70% of them report that they have critically reviewed the generated content.
A third of employees do not perceive much attention for generative AI in their own company
Employees who have heard of generative AI were asked to assess how much attention the topic receives within their company. There are noticeable differences in how employees perceive this attention. More than a third (35%) state that the topic receives little or no attention in their own company, while a good quarter (26%) consider the level of attention to be high or very high.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
The perceived level of attention varies greatly according to the respondents' level of education. Among those with a low level of formal education, 40% state that generative AI receives little or no attention in their company. In contrast, only 28% of those with a higher level of formal education share this view. There is an even clearer difference in the proportion of people who state that their company pays high or very high attention to the use of generative AI. While only 19% of people with a low level of formal education responded accordingly, the proportion among the comparison group of highly educated people is 36% – and therefore almost twice as high. It is also striking that uncertainty about how much attention one's own company pays to generative AI decreases with the level of education. For example, 22% of respondents with a low level of formal education answered "don't know" to this question, compared to a share of only 8% among those with a high level of formal education. One explanation for these results may be that occupations and activities are associated with different educational requirements (bpb 2023) and that the topic of generative AI systematically affects the occupations and activities of people with higher levels of education more strongly than those of individuals with a lower level of education (IMF 2024). This also seems to go hand in hand with a greater awareness of generative AI in companies, which especially require employees with a higher level of formal education.
The rather heterogeneously assessed attention in companies is also reflected in the implementation of guidelines for the use of generative AI. More than half of employees (53%) who have already heard of generative AI state that there are no guidelines for the use of this technology in their organisation. In contrast, only about one fifth state that they know of such rules in their own company. At the same time, there is also a relatively high level of uncertainty among respondents as to whether their own employer has defined any guidelines at all. A quarter of all employed respondents state that they do not know.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
There is an interesting association between the existence of guidelines in the company and the perceived attention companies pay to generative AI. As many as 29% of employees who perceive a high or very high level of attention toward generative AI in their company state that their employer has not defined any guidelines for the professional use of generative AI. In contrast, among the employees who perceive little or no attention toward generative AI in their company, 83% give the same response.
Among professional users of generative AI, 44% state that their employer has provided guidelines for the use of generative AI. However, an almost equal proportion (42%) state that no such rules exist in their company. In contrast, among employees who have heard of generative AI but have not used it professionally, only a comparatively small proportion of 14% state that their own company has defined guidelines for the use of the technology. On the one hand, these results show that although significantly more companies are dealing with guidelines in which generative AI also appears to play a role, on the other hand, even in such companies there are often no specifications at all, even two years after the publication of ChatGPT and Co.
While in private everyday life, generally available applications such as ChatGPT are most commonly and even almost exclusively used, specially developed or at least customised versions of generally available generative AI systems are becoming increasingly relevant in companies. For example, these systems provide better protection for sensitive data and can be adapted to specific business requirements (IW 2024). Currently, just under a fifth (18%) of employees who have already heard of generative AI state that their employer provides them with a generative AI system that has been developed or adapted specifically for their own company. In cases where such a specific system is available, it also appears to be widely used.
Majority do not see generative AI as threat to own professional activities
Only 4% of employees fear that their own professional activities will become completely redundant in the next ten years due to the use of generative AI systems. This represents a significant decline compared to the previous year's survey in September/October 2023 (Schlude et al. 2023). Back then, 11% of employees expected generative AI to make their work activities completely superfluous in the next ten years. In the current survey, 39% – around three percentage points more than in 2023 – expect generative AI to render parts of their own activities redundant. However, at 48%, almost half of the employees assume that generative AI will have no impact on their own professional activities in the next ten years – an increase of 5 percentage points compared with the previous year. There was also a slight decrease of 2 percentage points among those who are unsure of the developments in the next ten years – resulting in a share of 9% of employees who state that they “don’t know”.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
However, the expected change in one’s professional activities varies significantly depending on whether an employee has already used generative AI professionally. For example, 58% of professional users of generative AI state that parts of their job will disappear in the next 10 years. In contrast, only 39% of the group who do not use it responded in the same way. Among professional users of generative AI, 11% expect all their activities to become redundant. Meanwhile, only 1% of employed non-users state the same. No impact on work activities is expected by 29% of professional users and by 49% of employees who have not yet used generative AI professionally. The results show that where generative AI is already being used by employees, it is often perceived as increasing the risk for the automation of work activities.
Overall, more than half of employed respondents assume that the use of generative AI will generally lead to job losses in the future. Only 11% disagree or strongly disagree with this prediction.
Klicken Sie auf den unteren Button, um den Inhalt von Datawrapper zu laden.
Conclusion
Two years after the publication of ChatGPT and around a year since the bidt's last survey on the subject, it is clear that the use of generative AI in both private and professional contexts has hardly changed in the last year. Only around one in ten employees use generative AI for professional purposes at least once a week. However, when employees use generative AI, they do so quite successfully, as a majority of users report useful results and time reductions in their professional lives. Users of generative AI, in particular, also expect that at least some of their professional activities will be automated by generative AI in the future. Overall, more than half of all employees believe that the use of generative AI will lead to job losses. There is also a relatively widespread call for regulation of the technology. However, only few respondents are familiar with the European Union's AI Act, which recently came into force. Those who are aware of it are at least partially critical of it.
The results thus show that greater efforts are needed in terms of information policy. Politicians in particular need to do a better job of informing the public, but also businesses, about the AI Act. Also, when implementing the AI Act at a national level, it will be essential to fully utilise the scope of existing possibilities to keep pace with technological developments while minimising the additional bureaucratic burden for companies so that the regulation does not become an obstacle to innovation.
In the medium-term, generative AI will lead to major upheavals in the working environment. It is likely that some job profiles will become less important, others will change and new ones will emerge. The current wave of automation caused by generative AI will primarily affect higher-paid professions held by people with a higher levels of formal education (IMF 2024). The described survey data already shows that corresponding population groups are significantly more affected by the influence of generative AI. In order to take this change into account and maintain the employability of broad sections of the population in the future, investing in the development of skills is more important than ever. In the future, specific skills for programming and maintaining AI systems, but above all general skills in dealing with AI as well as cross-sectional skills in areas where AI does not yet deliver satisfactory results, will become increasingly important (OECD 2023). Ultimately, only a well-informed and competent population will be able to benefit from the advantages of generative AI and avoid the dangers of using this technology.
