This paper demonstrates how generative artificial intelligence (AI) enhances geoscientific document processing by improving text analysis, table extraction, and figure classification. Traditional workflows struggle with domain-specific terminology, poor-quality inputs, and rare formats. To address these challenges, we employ domain fine-tuned bidirectional encoder representations from transformers (BERT) models to enhance text processing. Additionally, we utilize multimodal large language models for precise table recognition and context-aware image classification. Finally, a domain-optimized retrieval system, GeoRAG, improves the relevance and accuracy of information retrieval. These AI-driven advancements streamline digitalization, enhance data extraction, and enable efficient handling of complex geoscientific documents. While challenges such as hallucinations, interpretability, and output consistency remain, this study highlights the transformative potential of generative AI for geoscience workflows and decision-making processes.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.