Artificial intelligence for the reconstruction of lost documents

The authors examine the AI capabilities in the process of the reconstruction of lost historical sources, e. g. documents, manuscripts, or photo materials. They offer the examples of neural networks (ChatGPT, Perplexity) processing images and texts. The purposes of the study are to analyze literature on AI utilization in restoration and archives; classification of software solutions for restoration of texts and images; analyzing practical experience of photos and documents reconstruction with neural networks. The authors reveal the AI strong and weak points for preservation of documentary heritage. The methods of analytical review, and historical, cultural, and technological methods were applied. The works by A. A. Druzhinina (2024), E. V. Bobrova (20230, etc., make the theoretical foundation for the study. The authors also offer the findings of the study experimental stage that enable to conclude that, despite some drawbacks. AI has substantial potential for cultural heritage preservation and compensation.