Artificial intelligence and emerging technologies are boosting research efficiency and global competitiveness, transforming industries. They also raise important social and ethical governance issues that require evidence-based policy decisions. This article presents a two-stage method for identifying relevant research topics in a proof-of-concept format. The first stage involves analyzing bibliometric records of preprints to identify key topics described by terms taken from abstracts. In the second stage, examples of relevant peer-reviewed publications are identified based on these key terms. This method provides a balance between a broad search for relevant topics and reliable verification of scientific results. The data sources used are metadata from ArXiv preprints on artificial intelligence (cs.AI 126,363 records) and emerging technologies (cs.ET 4,497 records) for 2021–2025. The study used 46,493 multi-word terms found in the annotations of cs.AI bibliometric records. To identify relevant peer-reviewed publications, it is advisable to use artificial intelligence-based search engines such as Semantic Scholar, Elicit, or ScienceOS to search for publications using the terminology identified in the first stage. The study shows that the use of a controlled lexicon allows for the identification of 4–5 groups of interpretable topics in text of abstracts, emphasizing the importance of using terms consisting of 2–4 words to achieve optimal results. Optimizing tasks using computing systems based on physical modeling principles supplemented by artificial intelligence could be a promising area of research.