The business use of AI requires a sustainable and goal-oriented development of AI systems and their integration into operational IT environments. The AI paradigm is called “training instead of programming” and requires the rethinking of existing operational and technological management processes, e.g. in development, procurement and operation of AI systems.

In our research, we develop new management methods, processes and structures that enable companies to use AI in a meaningful and value-creating way and thus generate long-term business value.

Our current research answers the following problems:

• Organizational transformation and operational capabilities for the use of AI (especially maturity levels and development paths)
• Development and operating models for the use of AI
• Scaling of AI systems into productive IT environments
• Leadership and communication for the use of AI in companies

This way, we answer the question of how organisations can industrialise and professionalise the use of AI and what prerequisites must be met for this.

In this field of research, we deal with the implementation of real AI projects. On the one hand, we analyze why a large part of initiated AI projects fail, i.e. are rejected or never “go live”. On the other hand, we are working on the development of management practices in order – starting from ideas – to get from prototypes to successfully operated AI systems in operational environments. In our research we identify relevant factors influencing the success of AI projects, derive appropriate measures and integrate them into an AI project management approach.

Our current research focuses on the following problems:

• Systematic, i.e. method- and tool-supported identification of AI Use Cases
• Barriers in the implementation of AI systems (especially socio-technical barriers)
• Value- and risk-oriented selection and implementation of AI projects
• Integration of AI development in IT and software engineering processes (e.g. agile, DevOps)

Our work aims to address the tension between business opportunity, human desirability and technological innovation with appropriate methods and practices.

In the area of Design Thinking for Artificial Intelligence (DT4AI), we investigate how the innovation and development of AI systems can be more strongly oriented towards the needs of users or end users. Although AI technology is a very important part of the development of AI systems, human needs, emotions and attitudes are often insufficiently considered, especially in early phases of innovation and the subsequent design of AI solutions.

In our research we develop models, methods and tools with different objectives:

• Methods and tools for user-centered need identification and analysis (need finding)
• Methods and tools for the contextualization of AI solutions on the individual, team and organizational levels
• Models for the design of user interfaces or user experience with regard to acceptance and use of AI solutions

Our work aims at developing AI innovations in such a way that their use promotes trust in them and that claims, values or even fears are adequately taken into account.

In the research area democratization of AI we work on problems of the company-wide adoption of AI. Especially for large corporations it is of central importance to understand AI not only as the task of individual specialists, but rather to involve a large part of the workforce in the development of AI.

Our assumption is that AI must be broadly anchored in the company if it is to support products, services and business processes in the future. Finally, all employees must be enabled to contribute to AI innovations, to participate in the corresponding developments and also to ensure the operation of AI applications in the corporate environment.

Accordingly, our work focuses on the following exemplary questions:

• How can companies enable the entire workforce to contribute to AI innovation and to participate in it?
• Which fields of action arise at the level of management, operations and technology operations in the broad use of AI?
• How can companies achieve a balance between freedom/innovation with AI and the economic, sustainable and responsible use of AI?

Our work to date shows that company-wide deployment creates a variety of areas of conflict, which we address with empirically supported research.

In the research area Augmentation of Human Intelligence we deal with questions concerning the supplementation and extension of human abilities by the use of AI systems. To replace humans is neither a principle nor the goal in the design of human AI systems. Quite the contrary. The synergetic interaction of humans and AI leads to a system design in which, for example, routine activities are transferred to the AI, whereas humans can contribute their abilities in the areas of creativity, interpretative skills and situational decision making.

This approach results in the following exemplary questions for our research:

• How can the cooperation between humans and AI be made successful?
• How can the value potential of augmentation in the company be identified and analyzed?
• Which design principles are relevant for an optimal system configuration?
• How does the selection and successful implementation of suitable AI procedures take place in the context of concrete fields of application?

Our work aims at realizing a value-oriented use of the respective strengths of man and machine in the operational environment.

AI projects usually require huge amounts of data to generate high quality outputs. The acceptance of an AI solution depends to a large extent on the first-class and precise operation of the solution. Since training data changes dynamically, the complete life cycle must be considered. The research area Data Bias and Data Pipeline Management deals with the access to training data and distortions in it in order to ensure long-term confidence in the AI solution.

This leads to the following questions:

• How can we ensure first-class, precise operation of our AI models throughout their entire life cycle?
• What types of bias can occur in an AI project?
• How can the different types of bias be identified and prevented?
• How can we ensure smooth access to excellent training data?

We work on procedural models to check your AI project for potential bias and suggest adapted mitigation strategies to ensure that your project is free of bias throughout its life cycle.