A systems analysis of patent automation is presented. Fragmented solutions are identified: search, generation, and blockchain operate in isolation. Problems are identified: fragmentation, lack of adaptation to Russian legislation, and vulnerability of document flow. The concept of a system integrating AI assistants for applicants and examiners with a single blockchain registry is substantiated. This will ensure security and support at all stages of patenting. The need for a system for the digitalization of patenting in the Russian Federation is confirmed.
This paper examines the evolution of physical storage media—from magnetic tape and optical discs to HDDs, SSDs, and beyond—along with the parallel development of operating systems and digital data formats. Special attention is given to long‑term preservation formats such as XML and PDF/A, as well as to the emerging need to structure information for AI‑based systems, ensuring resilience against rapid technological progress. Using a retrospective analysis, the study proposes strategies for guaranteeing data accessibility over extended time horizons (10, 25, and 50 years). A novel periodization is introduced, not merely for storage media, but for the interconnected triad of format–media–software, tailored specifically to the operational requirements of scientific libraries. One of the central findings is the necessity of shifting from passive archival storage to proactive management of the entire technological environment. The results underscore the importance of moving away from monolithic solutions towards hybrid hardware‑software ecosystems that simultaneously ensure both the physical preservation and the semantic accessibility of scientific heritage in the long term.
This paper addresses the problem of the hidden use of generative artificial intelligence in the academic environment and proposes a potential approach for the automated detection of generated scientific texts. The goal of the study is a comparative analysis of the information entropy indicator, specifically the Shannon equitability, in the abstracts of original scientific papers written by human authors and those generated by the Sber GigaChat large language model. A sample of 41 papers from the Tonkie Khimicheskie Tekhnologii (Fine Chemical Technologies) journal was used as the data corpus. The texts automatically preprocessed and then their lexical diversity and predictability were evaluated using Welch’s t-test and the Mann–Whitney U-test. The results showed that the generated abstracts are characterized by metric homogeneity and a significantly smaller length compared to those written by humans. The average Shannon equitability in machine-generated texts was statistically significantly higher (0.9826) than in human-written texts (0.9708). This phenomenon is explained by the fact that neural networks form concise constructs without the repeated duplication of highly specialized terminology characteristic of humans, which mathematically increases the uniformity of word distribution. It is concluded that the Shannon equitability can serve as a reliable indicator of the algorithmic origin of a text; however, to improve the accuracy of the examination and protect against prompt engineering methods, it is advisable to combine it with deep semantic analysis within complex authorship check systems.
This study examines the potential of artificial intelligence (AI) to improve the effectiveness of the World Food Program (WFP). Focusing on complex humanitarian situations, it explores the ability of AI to address operational inefficiencies by streamlining logistics, reducing costs, and improving decision-making in crises. It begins by developing a conceptual framework based on case studies of AI use in such areas as earthquake relief in Nepal (International Red Cross), COVID-19 monitoring using BlueDot algorithms (World Health Organization), and the use of AI in UNICEF education initiatives, to demonstrate the effectiveness of AI technologies such as predictive analytics and machine learning. These examples highlight challenges in scalability and ethical governance of AI tools when applied to humanitarian projects. The study focuses on analyzing WFP AI-based projects and identifying challenges such as skills shortages, data privacy issues, resistance to technological change, expected high implementation costs, and weak digital infrastructure in crisis regions. While WFP projects such as SCOUT, SHAPES, and SKAI ensure more efficient and sustainable delivery of humanitarian aid, the limited adoption of AI by WFP highlights the need for further innovation.
Existing methods for evaluating the performance of information and analytical centers are analyzed. A system of key criteria is developed to objectively measure the economic impact and the qualitative return on the functioning of these organizations. Particular attention is paid to the practical aspects of the implementation of evaluation tools.
