A survey on LLMs for spreadsheet intelligence (orbilu.uni.lu)

A recent survey highlights the transformative potential of integrating Large Language Models (LLMs) within spreadsheet environments, addressing the complex and often cumbersome tasks associated with data analysis. Despite the widespread use of spreadsheets, users frequently encounter challenges with formula writing and data cleaning, which can lead to errors and inefficiencies. This survey delineates the "spreadsheet intelligence" problem by outlining a workflow of independent stages, categorizing existing research, and establishing a taxonomy of tasks relevant to LLM applications in spreadsheets. The significance of this work lies in its comprehensive review of methodologies, capabilities, and benchmarks, paving the way for a more intuitive interaction with spreadsheets through natural language processing. By developing an end-to-end pipeline for various tasks and highlighting open challenges, the authors set the stage for future research aimed at creating more reliable and user-friendly LLM systems. This advancement could revolutionize how users interact with data, enabling quicker, more accurate analyses and reducing the dependency on complex formulae, thus making data manipulation accessible to a broader audience.