The field of data management and extraction is witnessing a significant shift towards automation, driven by the increasing importance of data in various industries. Recent developments have focused on leveraging large language models (LLMs) and vision language models (VLMs) to generate high-quality metadata and extract complex information from various data sources. These models have shown promising results in automating tasks such as metadata creation, data extraction, and cataloging, with some studies demonstrating that LLMs can generate metadata comparable to human-created content. However, the successful application of these models requires careful consideration of task-specific criteria, domain context, and the need for human verification to ensure accuracy and quality. Noteworthy papers in this area include:

• A study on using LLMs to extract DCAT-compatible metadata, which achieved high-quality results with fine-tuned models and few-shot prompting.
• A paper on using a VLM to generate catalogue descriptions for photographic prints, which highlighted the importance of human review and trust in AI tools.
• A methodological study on expediting data extraction using LLMs, which demonstrated the potential of review-protocol-based methods but also emphasized the need for more robust performance evaluation.