A Review of the Practices and Risks of Implementing Artificial Intelligence in Business Process Management Systems

Despite the growing popularity of generative artificial intelligence, statistics show low levels of successful implementations in company processes. At the same time, there is a gap in systematic studies of the causes of such failures.
The purpose of the work is to analyze the practices of using generative artificial intelligence as a tool for automating processes (both successful and failed), as well as to identify typical risks and their causes. The study was carried out on the basis of statistical data, scientific papers and information about projects from open sources. Based on the collected risk data in each subject area, common application risks were identified and a matrix of causes, sources and consequences of each of the risks for the company was compiled. In addition, the gradual degradation of artificial intelligence tools was considered to compile a complete picture of the situation on the generative neural network market.
Based on the study, it was concluded that a complete uncontrolled model of process automation by generative artificial intelligence is impossible due to its instability in the long term. The proposed risk register serves as the basis for developing a more rational and cautious approach to the introduction of generative artificial intelligence, shifting the focus from total automation to a controlled hybrid model.

1. Han M. S., Gorbunov V. L. A face recognition method based on CNN // Innovative approaches to solving technical and economic problems: Collection of works of the International Conference, Moscow, May 29, 2023. Moscow: National Research University «Moscow Institute of Electronic Engineering,», 2023. P. 119–128. (In Russ.). EDN: OHUEUN

2. Kozlova N. Sh., Dovgal V. A. Analysis of the use of artificial intelligence and machine learning in cybersecurity // Bulletin of Adygea State University. Series: Science, Mathematics and Technology. 2023;(3):65–72. (In Russ.) https://doi.org/10.53598/2410-3225-2023-3-326-65-72

3. Lin Kyi, Abraham Mhaidli, Cristiana Santos, Franziska Roesner, Asia Biega. “It doesn’t tell me anything about how my data is used”: User Perceptions of Data Collection Purposes. CHI ‘24: Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 2024. https://doi.org/10.1145/3613904.3642260

4. Ho J.Q.H., Hartanto A., Koh A., Majeed N.M. Gender biases within artificial intelligence and ChatGPT: evidence, sources of biases and solutions // Computers in Human Behavior: Artificial Humans. 2025;4:100145. https://doi.org/10.1016/j.chbah.2025.100145