Publikation

Explainable Artificial Intelligence (XAI) Supporting Public Administration Processes – On the Potential of XAI in Tax Audit Processes

Nijat Mehdiyev, Constantin Houy, Oliver Gutermuth, Lea Mayer, Peter Fettke

In: Proceedings of the 16. International Conference on Wirtschaftsinformatik. Internationale Tagung Wirtschaftsinformatik (WI-2021) March 9-11 Duisburg-Essen Germany Seiten 1-17 Universität Duisburg-Essen 3/2021.

Abstrakt

Artificial Intelligence (AI) can offer significant potential for public administrations which – in Germany – are likely to face considerable skills shortages in the next years. AI systems can, e.g., support the automation of work processes and thus disburden administrative staff. As transparency, traceability and fairness play a major role in administrative processes, explainable AI (XAI) approaches supporting a better understandability and traceability of the results of AI systems enable a proper usage of AI in public administration. In this article, we investigate the potential of XAI for the support of tax authority processes, especially the selection of tax audit target organizations. We illustrate relevant tax audit scenarios and present the potential of different XAI techniques which we currently develop in these scenarios. It shows that XAI can significantly support tax audit preparations resulting in more efficient processes and a better performance of tax authorities concerning their main responsibilities. A further contribution of this article lies in the exemplary application of XAI usage guidelines in the public administration context.

Projekte

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence