Employees who can work well with Excel spreadsheets are highly valued. But Microsoft researchers have developed answerwhich can at least partially replace them. The large language model SpreadsheetLLM is designed to control spreadsheets using natural language commands.
Image source: Rubaitul Azad / unsplash.com
SpreadsheetLLM uses AI to analyze and interpret spreadsheet data, solving most of the problems associated with it by serializing the data, i.e. including addresses, values, and cell formats in the stream. The tool contains a SheetCompressor component that compresses spreadsheets for feeding them to an AI model. It consists of three modules: the first analyzes the table structure and discards non-tabular content; the second transforms the data into a more efficient representation; and the third aggregates the data.
SpreadsheetLLM has some limitations. It ignores cell colors, which can have meaning, and does not perform semantic compression on natural language cell content. But this is enough to reduce token consumption by 96% when querying AI, which means saving computing resources. As a result, users without proper technical training can send natural language queries to SpreadsheetLLM and achieve their goals. But the main goal of the project is not to replace humans, but to help them with finance, accounting, and other areas related to data processing.
The model includes a Chain of Spreadsheet (CoS) framework for analyzing the contents of multiple spreadsheets. SpreadsheetLLM can handle structured and unstructured spreadsheet data, an aspect the researchers say could reduce incidents of “hallucination” in AI responses. The project is still in the research phase and is not yet ready for commercial release.
If you notice an error, select it with your mouse and press CTRL+ENTER.
