 |
International Society of Science and Applied Technologies |
| Improving Software Dependability by AI Agents | ||||
| Author | Grigore Albeanu
|
|||
| Co-Author(s) | Florin Popențiu-Vlădicescu; Henrik Madsen
|
|||
| Abstract | Integrating Software Engineering (SE) and Artificial Intelligence (AI) impacted the current practice in software development and its dependability assurance. In this paper the agentic approach is considered as evolved from classical software agent to AI agents and agentic AI systems. The usage of AI agents separate or belonging to an agentic AI system is analyzed to discover how the software dependability can be improved, and what new challenges appear due to AI integration in SE practices. The coding and testing processes have benefitted from AI, the automation degree being increased. However, many failure modes both from safety and security points of view already have been identified in agentic AI systems increasing the effort of risk mitigation.
|
|||
| Keywords | software agent, artificial intelligence based agent, software testing, LLM, DevSecOps | |||
| Article #: RQD2025-292 | ||||
Proceedings of 30th ISSAT International Conference on Reliability & Quality in Design |