Using artificial intelligence technologies to optimize methodology and organization of scientific research

• Datum: 2025
• 1. Verfasser: Popereshnyak, S.V.
• Format: Artikel
• Sprache: Ukrainian
• Veröffentlicht: PROBLEMS IN PROGRAMMING 2025
• Schlagworte: artificial intelligence; scientific research methodology; research organization; reproducibility; ex plainability; Bayesian methods; optimization; POMDP; data bias; academic integrity; UDC 004.8: 001.89
• Online Zugang: https://pp.isofts.kiev.ua/index.php/ojs1/article/view/861
• Назва журналу: Problems in programming

The rapid digitalization of science puts forward new requirements for the methodology and organization of research. Artificial intelligence (AI) tools are able to accelerate data analysis, support hypothesis generation, and automate routine procedures, but at the same time exacerbate issues of reproducibility, transparency, and academic integrity. The purpose of the work is to substantiate and test approaches to integrating AI technol ogies to optimize the methodology and organization of scientific research in the field of digital development. The work applied a systematic analysis of modern practices, mathematical modeling of processes (optimiza tion of the choice of modes "human/AI/hybrid", Bayesian assessment of reliability, regularization of bias and interpretability, POMDP-planning of scientific cycles), as well as the design of a modular architecture for supporting research. Empirical testing was carried out on prototypes of workflows: automated literature re view, intelligent processing of experimental data, preparation of publication materials. An integrated model of scientific process management was proposed, combining: (i) formal selection of human and AI roles under resource and quality constraints; (ii) aggregation of evidence from human and machine channels through a Bayesian scheme; (iii) simultaneous limitation of algorithmic bias and increase of explainability; (iv) strategic POMDP-planning of experiments taking into account risks and costs. The paper showed that the use of the proposed model reduces the time for preparing reviews and analyzing data, increases the reproducibility of conclusions and transparency of decisions, as well as reduces the risks associated with the “black box” and data bias. Practical recommendations for implementing AI in research units are formulated: regulations for disclosing the use of AI, data quality control, requirements for the explainability of models, and the role of the researcher as a responsible interpreter. AI should be considered as a tool for strengthening scientific work. The integration of the proposed models into the methodology and organization of research increases the effi ciency, reproducibility, and ethical reliability of scientific results, opening the way to scalable, resource-effi cient research practices. Problems in programming 2025; 3: 91-101