As far as the business world is concerned, generative AI has arrived: 80% of companies SAP LeanIX surveyed report employees are already using this new technology, though in most cases only to a limited extent.
According to the survey, a large majority of companies expect artificial intelligence to increase efficiency, and almost half expect a significant improvement in work quality. For those companies that have yet to use generative AI, or to use it extensively, potential data security risks and uncertainty about the legal implications of AI top their list of concerns. Thanks to these and other concerns, three quarters of companies have rules in place to limit the use of artificial intelligence.
Still, despite differences in concerns and expectations, there is widespread agreement among all respondents on one point: 90% consider a comprehensive overview of generative AI‘s use in the company important. Unfortunately, only 14% currently enjoy such an overview. In fact, more than a third of respondents report that data on AI usage is not collected at all or that there is no clear accountability for tracking such data.
Almost all respondents stated they are working with or plan to work with a framework for AI governance. However, these efforts do not yet seem to be adequate or effective, since only 15% of companies surveyed believe they are well prepared to comply with existing and currently emerging legal guidelines for the use of artificial intelligence.
Ultimately, these results from the SAP LeanIX AI Survey 2024 reveal that companies around the world have a long way to go when it comes to effective AI governance.
The advent of generative AI is comparable to the emergence of social media, the launch of the smartphone, or the introduction of the Internet, according to Forrester. They expect annual growth rates of 36% for the technology by 2030.1 CIOs surveyed by Gartner in 2023 are also betting on artificial intelligence. Three quarters plan to increase their investments in this area this year, and 80% report that their companies plan to adopt this technology within the next three years.2 IDC expects global spending on AI solutions to exceed $500 billion in 2027.3
But the experts also point to difficulties that will hinder, in the short term, the spread generative AI and productivity gains it promises. These difficulties include a lack of clarity around regulations, questions concerning intellectual property rights, and broader ethical concerns.
How do companies currently view the opportunities and challenges associated with introducing generative AI? What measures have they already taken, or plan to take, to address potential challenges? These and other questions served as the focus of the SAP LeanIX AI Survey 2024.
Survey of IT experts from around the world on the use of artificial intelligence
From December 2023 through January 2024 SAP LeanIX surveyed customers in relevant IT roles at 226 companies about their use and management of generative AI. In terms of specific IT roles, enterprise architects make up the largest proportion (67%) of the sample. Seventy-one percent of those surveyed work for companies based in Europe, 19% for companies based in the US. Regardless of region, respondents evenly represented companies of different sizes, with 40% working in organizations with over 10,000 employees.
Generative AI already in use at 80% of companies
Whether embedded in applications or as standalone tools, nearly a third of those surveyed report their employees are already using generative AI extensively. The largest segment of respondents (48%) report some generative AI usage, but only to a limited extent. Overall, with 80% of companies using generative AI, one can say that AI has arrived in a business context.
When asked what they expect to achieve with this new technology, 78% of the IT professionals surveyed believe that employees will get work done more quickly. 47% foresee an improvement in the quality of work people do thanks to AI.
Since companies primarily expect AI to increase efficiency, it is no surprise that exploring how to increase efficiency with AI is a top priority. In 42% of companies where generative AI is already being used extensively, this task is the highest priority. For 33%, determining how to integrate AI into the products and services they offer is the top priority. Only 21% see addressing the ethical and security issues associated with AI the top priotiry. Neither evaluating the use of AI by competitors nor using AI for greater sustainability were seen as priorities for the vast majority of companies.
Concerns about data security stand in the way of greater generative AI usage
In almost half of the companies surveyed, generative AI is not currently being used extensively. Twenty percent of those surveyed are not using it at all. Why are so many companies slow or hesitant to adopt generative AI? Data security risks were cited most frequently (78%), followed by uncertainty around the legal consequences of its use (59%). Almost half (48%) of those surveyed also cited lack of employee know-how as an issue. However, ethical concerns or excessive costs were only cited as hurdles by 22% and 16% respectively. Other reasons given were skepticism about the current benefits of the technology and lack of concrete use cases.
The biggest obstacles can best be summed up as: uncertainty and ambiguity. Companies apparently have chosen to address these concerns with a simple remedy: limiting access to or use of AI. 73% of all companies using AI, whether extensively or not, have imposed rules limiting its use.
Limiting AI use at the application and task levels
Companies surveyed mainly limit AI usage in one of two ways. Either they limit usage to AI models from approved vendors (51%), or they limit usage to specific tasks (50%). The third most common restriction (34%) is to limit AI usage to specific user groups.
A comprehensive overview of generative AI in the company is rare
As has already been shown, a large majority of companies are already using generative AI, although three quarters of those companies limit or restrict its usage. Nevertheless, a truly comprehensive overview of where generative AI is used in the company seems to be missing. Although 90% of those surveyed consider such visibility either important or very important, only 14% report having it, a remarkable discrepancy.
Such a wide gap is surprising, all the more so since the majority of the companies surveyed operate in Europe, where the EU’s AI Act, aimed at regulating AI, will soon come into effect. This act requires companies to provide precise information about the risk level of their AI use cases and, if they fail to do so, threatens penalties of up to 7% of their total annual revenue.4 How can companies report on any use cases if they only have a partial view of usage?
Challenges to collecting data on AI usage
Almost half of those surveyed stated that the IT department or enterprise architects collect information on the use of generative AI in the company. Alarmingly, however, in more than a third of companies surveyed, responsibility for tracking such data is not clear or, worse, the data is not collected at all. With no clear responsibility for data collection, it is understandable why some companies lack a comprehensive overview of AI usage.
Unanimous opinion: Companies need a framework for AI governance
The need for a framework to govern AI across the company is affirmed by 97% of the IT professionals surveyed. However, very few (19%) have already implemented such a framework. Interestingly, the percentage increases (32%) among organizations with more than 10,000 employees. 52% of companies are currently in the development phase when it comes to an AI governance framework. Curiously, 26% of those who view an AI governance framework as necessary also report that there are no current plans to develop and implement one.
The findings of the SAP LeanIX AI Survey echo a McKinsey study from 2023, which found that only 21% of companies surveyed had already created comprehensive guidelines for AI.5
Companies do not feel prepared for AI regulations
Gartner predicts that half of all governments worldwide will regulate the use of AI by 2026.6 The European Union is the most advanced in this regard. The European Parliament approved the EU AI Act in March 2024, two years after its official introduction. This summer it is expected to come into full force.7,8
Are companies prepared to comply with existing and emerging regulations? Remarkably nearly 20% of those surveyed could not say. Of those who could, a third say clearly they are not. The majority (52%) feel at least partially prepared. A mere 15% report they are prepared for AI regulations in the regions where the company operates.
One would expect that those companies reporting they have already implemented an AI governance framework would feel well prepared to comply with AI regulations. Indeed, at 36%, the proportion of respondents confidently reporting they are well prepared is significantly higher than in companies still developing their AI guidelines (14%) or that have not yet made any effort at all (2%). However, even in companies with an existing framework, more than half of those surveyed say they are only partially prepared for legal compliance.
Conclusion: Without visibility into AI usage, AI governance is impossible
Three quarters of those companies in which AI is used in some way have established rules for its use. Almost as many (71%) say they have implemented or are working on an AI governance framework. Despite these measures, very few feel well prepared to comply with existing and emerging AI regulations. Why is that?
One reason for this may be the ongoing lack of clarity around these legal requirements, many of which are still being formulated. However, if you consider that, at least in the European Union, the forthcoming regulations have already been developed and that the majority of the companies surveyed operate in Europe, this can’t explain the lag in preparedness.
Does it mean that the measures companies are taking to regulate and govern AI are inadequate or wrong? Not necessarily. In fact, this study points to a more likely explanation: While almost everyone agrees that a comprehensive overview of the use of generative AI in the company is necessary, this information is almost never completely available.
Visibility is the basic requirement when it comes to effective AI governance. Regardless of what form the legal regulations ultimately take companies will definitely need to understand where AI is used in the organization, by whom, and to what end. The fact that this data is not currently being collected systematically or consistently can be partly explained by the fact that responsibility for this systematic collection is, in many cases, unclear.
Almost half of those surveyed report this responsibility falls to the IT department, and especially the enterprise architects (EAs). As specialists in the IT landscape, EAs are well-situated and motivated to collect data and provide information on AI use. With the right tools, they can become a catalyst for the responsible, legally compliant use of AI in the company.
Sources:
- https://www.forrester.com/blogs/spend-on-generative-ai-will-grow-36-annually-to-2030/
- https://venturebeat.com/ai/gartner-generative-ai-will-be-everywhere-so-strategize-now/
- https://aimagazine.com/articles/mckinsey-study-confirms-how-much-generative-ai-has-exploded
- https://www.gartner.com/en/newsroom/press-releases/2024-02-29-ai-regulations-to-drive-responsible-ai-initiatives
- https://www.europarl.europa.eu/news/en/press-room/20240308IPR19015/artificial-intelligence-act-meps-adopt-landmark-law
SAP LeanIX AI Survey 2024: More information about the study
In December 2023 and January 2024, SAP LeanIX conducted an online survey of 226 of its global customers regarding their usage of generative AI. Most of the respondents work as enterprise architects and therefore have unique insights into their organization’s IT landscape. The majority of participants in the study were from Europe (71%), with 19% in the U.S. In terms of company size, the sample was fairly balanced.
Differences in the number of respondents represented in the charts result from the exclusion of respondents who could not/would not provide information on individual topics. For better readability, the results in this report are presented as percentages without decimal places. If the addition of these values does not result in exactly 100%, then this can be attributed to rounding differences.
