AI-assisted Software Engineering: a tertiary study

The research in Artificial Intelligence (AI) and its applications across the software engineering (SE) domain has progressed significantly in the last decade, evidenced by an increase in systematic literature reviews. Our study aims to provide an overview of existing systematic reviews within this research area and synthesize the findings related to AI subfields (Machine Learning, Search-based, Evolutionary Computation, Vision, Robotics, Expert Systems, Speech Processing, Natural Language Processing, and Planning) supporting the five main SE processes: requirements, design, development, testing, and maintenance. For our tertiary study, we used the methodology based on Kitchenham's guidelines. We selected 11 reviews published between 2000 and 2021, including results from 513 primary studies. The selected reviews cover the AI subfields across the five main SE processes. Studies have covered Machine Learning, Natural Language Processing, and Evolutionary Computations as AI subfields assisting SE processes. We have found that reviews in AI-assisted software testing are the most common, followed by software maintenance and development. Our study can help researchers identify missing reviews on AI-assisted SE topics to help further consolidate this research area.