Infographic – ChatGPT

How is ChatGPT being used? Thanks to the WildChat project by the Allen Institute for AI, one million conversations were made publicly available. In collaboration with Basel-based data journalist Felix Michel, we analyzed this dataset to identify the most frequently used three-word sequences (3-grams) in both French and English.

The results are visualized using a treemap, where the surface area of each of the 8000 rectangles corresponds to the frequency of one of the 4000 most common 3-grams in each language. A selection of rectangles—highlighted either arbitrarily or randomly—also display their absolute frequency values.

ClientEPFL
01 DIMENSIONS CHATGPT
02 DIMENSIONS CHATGPT
03 DIMENSIONS CHATGPT
04 DIMENSIONS CHATGPT
05 DIMENSIONS CHATGPT
06 DIMENSIONS CHATGPT
Other projects