Over the last few years, and one could even say months, we have experienced a great technological revolution, and the AI They have virtually taken over the majority of conversations people have in the world. It is not difficult for a topic of conversation related to this technology to arise, just as it is not strange for new projects to arise such as the integration of one of the versions of one of the most famous LLMs, such as ChatGPT in a Excel.
The projects carried out by many people never cease to surprise us, although it is clear that those related to Excel
AI comes to Excel
For a long time, we have wondered how it is possible to make a program or application work within Excelthis time the doubt arises again when seeing how a developer was able to integrate the second version of ChatGPT in the program to create Microsoft spreadsheets. Although one thing is clear, it is obviously the power offered by this version within Excel It is not even similar to the one that could be found originally, since it includes 124 million parameters, while its real version used 1.5 billion parameters.
It’s still a real challenge, even if, according to the creator of this project, “if you can understand a spreadsheet, you can understand the artificial intelligence“, because it allows us to know how LLMs work, as well as how to predict how these systems will work. In this way, Ishan Anandcreator of this spreadsheet with ChatGPT encourages us to download this file to try it and deepen our knowledge about the AI which uses this smaller version of ChatGPT.
ChatGPT in Excel, how does it work?
If we follow the video included by the creator of the project, we can understand little by little how this version of ChatGPT works, and that it is a series of spreadsheets that allow us, from several words, to generate the next one through a tokenization process, which assigns a value to words and is then able to predict the next one that would fit perfectly into a sentence. In this way, the spreadsheet is able to predict a series of words that could be implemented in the sentence to give it meaning, as shown in the example shown in the video that explains how it works.
Although there are a series of limitations, since it is ultimately a model recreated in a program that as such is not designed to offer the power of LLM, such as the fact that only words of 10 words can be indicated. less, in addition to being only in English.