The Text
I chose The Wealth of Nations from Project Gutenberg because I study economics but, embarrassingly enough, had never read Smith beyond the famous lines. I analyzed it two ways: with Voyant and with an AI model, Gemini.
Voyant: Counting
The top five words were great (1,583), price (1,264), country (1,244), greater (1,085), and labour (1,011). Voyant treats “great” and “greater” as separate words, and neither says much about Smith. An “â” also appears in the cloud, which seems to be an encoding glitch from the Gutenberg file. Voyant bluntly counts what it gets, errors included.
The Links tool connects price, country, and great to labour, money, and trade, which traces Smith’s argument that labour sets value and trade builds wealth.
AI: Summarizing
Gemini outlined the book’s five parts and pulled the famous butcher-and-baker quote from Book I. Then it told me it had only read the opening of the file. Its tool cut the text off after a small fraction of the book. It said “invisible hand” appears once, in Book IV, but that came from its training, not the text. I searched Voyant and confirmed the count.
What Worries Me
Voyant read every word and shows its method: I can click any count and see each sentence behind it. The AI read a sliver and filled the rest with what it already “knew.” This time it told me. Many AI tools won’t. In the Age of AI, the danger is fluent answers standing in for reading. The tools work best together, with the AI suggesting and the counts verifying.