Artificial Intelligence in Controlling—A Look Ahead

AI was first mentioned in 1956 at a conference at Dartmouth College. While expectations remained modest in the years that followed—and many research projects were even shut down entirely during the so-called “AI winters”—the field has gained tremendous momentum in recent years and has become a megatrend. Many applications have already been playing an important role in people’s lives for some time. For example, search engines, social media feeds, and spam filters in email services are based on AI. Active use has also become relatively simple, and in many cases, specialized training is not necessary to achieve useful results. Keyword: ChatGPT. AI can already improve and accelerate business processes today. Consequently, numerous companies are embracing the technology. According to a study by the ifo Institute in Germany, the proportion of these companies currently stands at 13.3 percent. 

While the processes underlying how AI works are not new, there have been several developments that have paved the way for the progress made in recent years. For one thing, companies today have access to large amounts of data (big data). For another, there are optimized AI algorithms that allow this data to be put to better use. Furthermore, the computing power and capacity of computers have increased significantly in recent years. In accordance with Moore’s Law, the number of transistors on a chip has roughly doubled every two years since 1965. 

What is AI?

But what exactly is AI? The term describes self-learning computer systems that, by utilizing large amounts of data and algorithms, perform human-like intellectual tasks such as problem-solving or learning, assist humans in decision-making, and may ultimately even replace them. There are various levels of development, ranging from algorithms that handle repetitive tasks to big data analytics and self-learning systems. 

Even the purely mathematical processing of data sets using basic arithmetic operations is classified as AI in some definitions. However, such methods belong more to the field of data analysis and statistics than to AI, and do not fall within the specific scope of AI functionalities. 

The way AI works can be described in the following steps: 1. the collection of structured and unstructured information and data, which resembles the process of human sensory perception; 2. the analysis and meaningful processing of the collected information and data; and 3. the execution of actions based on the collected information. A fourth step is autonomous learning through training and feedback based on the data collected beforehand. The last step, in particular, is crucial to defining AI and distinguishes it from other applications, some of which are already quite complex. 

Applications of AI

The main areas of application for AI in businesses that are already relevant today can be divided into different categories. 

First, there are human-to-machine dialogue processes. These allow users to communicate with a machine or computer in natural language, thereby avoiding complex screen-based or keyboard interactions. This can take place verbally or in writing. An example of such dialogue processes is voice control in navigation systems or virtual voice assistants like Amazon Alexa. 

On the other hand, there are machine-to-machine processes based on the interconnection of technical devices with one another and with a central control system. The data exchange this enables is essential, for example, for applications in the Internet of Things (IoT). One specific area of application is the prediction of upcoming maintenance tasks using machine learning methods that draw on sensor data. This can be done, for example, with air conditioning systems as part of building automation. 

Intelligent Automation (IA) refers to the combination of process optimization and AI. It helps companies improve their internal processes. One of its many applications is in the automotive industry. Here, IA can be used to more effectively forecast and adjust production in order to better respond to supply and demand. Furthermore, robots and automated systems can take on tasks such as assembly, welding, painting, and quality control, thereby minimizing employee injuries and delivering higher-quality products at lower costs. 

The final category, intelligent decision support, refers to the analysis of data using AI algorithms to enable more effective decision-making. Examples of applications include assistive systems in medicine, where AI-based diagnostics can support humans. For AI to effectively support human users in their decision-making, high-quality data is required. 

To what extent is AI currently used in controlling?

The use of AI in controlling can take various forms, though not all of them are already a reality in companies today. These forms can be categorized into the four levels listed below. 

The first level is semi-intelligent data analysis, which is used in many companies to complement human intelligence. Specific AI techniques used at this level of AI deployment include automated data cleaning and preparation, pattern recognition and forecasting, as well as automated financial reporting and analysis. 

The next level of development involves proactive support from AI based on an even more comprehensive data set. A corresponding intelligent assistance feature can help controllers perform their duties in various work situations. This can occur, for example, when interpreting complex financial data, generating real-time reports, or assisting with budget preparation. 

The use of AI with enhanced content capabilities represents the third level of AI application in controlling. Here, AI can not only interpret data but also provide context-specific recommendations for actions in controlling, based on company-specific goals and guidelines. In this form, AI has expanded decision-making autonomy and can independently perform certain repetitive tasks or make minor decisions based on recurring patterns in the data. 

The fourth stage involves the comprehensive strategic use of AI. At this stage, AI is capable of autonomously analyzing data and acting strategically based on the results, or independently proposing strategic measures. All approaches from the previous stages are combined here. The most important source of learning for the AI at this stage is the behavioral patterns of controllers, which are observed by the AI system. In this way, it can observe all relevant cause-and-effect relationships within the controlling context of the respective company and learn from them. 

Use cases from the first two levels can already be found in companies today. However, AI is not yet advanced enough to independently identify opportunities for optimization and determine the most efficient approach. As a result, the technology is still a long way from the fourth level mentioned. 

Examples of AI in Controlling That Are Already Relevant Today

At the first two of the aforementioned levels, there are various fields of application for AI in controlling, each offering different levels of benefit to stakeholders. 

One such area of application is forecasting. By analyzing past data—such as supply, demand, sales figures, or production costs—AI tools can integrate various historical data streams. When this data is modeled to project future trends, reliable forecasts can be generated for numerous areas of a business. 

In addition to forecasting , AI systems can also make the planning process significantly more efficient. By using these systems , patterns and anomalies in transactions, financial data, and business reports can be identified, and drivers—for example, for revenue planning—can be pinpointed. The planning process in particular lends itself to the use of external data, such as economic indicators, to better assess developments in the market, among competitors, or related to risks. 

Furthermore, AI can assist with risk management in controlling. By processing data from various sources and recognizing patterns, AI algorithms can identify risks at an early stage. For example, modern tools are capable of reviewing contracts and ensuring compliance with applicable regulations. 

AI can also assist with data analysis . Thanks to AI, this process can be largely automated without requiring significant staff resources to prepare the information. For example, data from various functional areas—such as sales, production, or logistics—can be quickly analyzed using AI and immediately presented in graphical form. 

What to Keep in Mind When Using AI in Controlling

For the integration of AI into a controlling department to succeed, certain foundational requirements must be met within the company. If the company has problems generating, processing, and completing data, or if it is struggling to set up a seamless digital tool landscape, the successful implementation of AI will be difficult. Implementation should therefore only be considered if a comprehensive database and relevant tools are in place and the company’s processes are prepared for the change. Only when, for example, the use of an ERP system has established standardized processes and a complete data foundation as a basis does the use of an AI-based system in controlling make sense. 

It is also important to note that when working with AI, there is a human component involved in formulating data models and selecting algorithms. This is because—at least at the current stage of development—human input is necessary to ensure that data queries arise logically from day-to-day business operations and that correlations are not mistakenly interpreted as causal relationships. At the same time, humans can ensure that the data is comparable and consistent. The symbiotic interaction between AI and humans therefore currently offers the greatest opportunities to increase effectiveness within the company and gain a competitive advantage. 

Before using AI in controlling, a clear objective should also be defined. Only when the intended benefits of using AI in controlling have been precisely defined can the technology be successfully integrated into the various stages of the value chain. Possible objectives may include reducing the likelihood of errors, improving the quality of decisions, or providing information more quickly. 

To achieve these goals, a roadmap should also be developed. Questions that controlling managers should ask themselves include: How can I properly prepare my data for use with AI? How can I extract information from the data? How can I draw conclusions from the data? And how can I use these conclusions and external data to create models? 

Conclusion

The use of AI is bringing about lasting changes to business processes. This also applies to management accounting. AI offers numerous potential applications in this field, some of which are already in use today. It not only simplifies day-to-day work but also provides new insights. In many cases, this does not require a data scientist or an AI specialist. Consequently, for most controlling departments, the question regarding the use of AI is not one of “if,” but of “when.” Accordingly, executives should begin early on to familiarize themselves with the technology, identify useful areas of application within their company, and build relevant expertise within their team. 

The far-reaching impact of AI will also affect the job requirements for employees in controlling. Qualifications such as a strong aptitude for numbers, experience with Excel spreadsheets, and industry knowledge will become less relevant in the future, while expertise in using BI tools and an understanding of IT systems will become increasingly important.  

Despite the major upheaval that AI brings, it will not render controllers obsolete in the foreseeable future. Rather, controllers will gain the freedom to focus on activities that are critical to value creation within the company, as they will spend less time collecting and processing data. In addition, AI will enable them to conduct more in-depth data analyses than before. This will give rise to new areas of responsibility and lead to a shift in the respective roles of financial controllers and machines.