Week 3 Lab

For this lab assignment, the text I downloaded from Project Gutenberg was The Odyssey by Homer. After uploading this file to Voyant Tools, it was interesting to see the most frequently used words. First, the most used word is Ulysses, which turns out to be the Latin name for the main character Odysseus. It was also interesting to see the other words that were common, such as house, sea, and ship. These are heavily tied to the story, as it is about Odysseus’s long journey home after the Trojan War.

Next, I wanted to compare the Cirrus word cloud with a word cloud generated by Gemini. The prompt I fed it was simple: “generate me a word cloud using the top 30 most commonly used words in the attached txt file.” I’m not going to lie, I was a little disappointed in the result. First, Gemini automatically excluded common “stop words” from the word cloud. The word cloud Gemini gave me just did not look clean at all, with words overlapping each other all over the place. Gemini then told me that, compared to the word cloud generated by Voyant Tools, Gemini’s word cloud was better because it looked at the context of the words, not just the word count, and was therefore more meaningful. While Gemini may be right in that the words it provided me are more “influential”, it kind of strayed away from my original prompt which surprised me.

In this “Age of AI”, we must be careful when using AI to help come up with conclusions about texts. There may be algorithmic bias, which is when an AI model is pre-trained on literary summaries and may use prior knowledge to come up with conclusions based on previous conclusions from other texts, instead of looking at and analyzing only the text you provided. Also, it has been shown that AI can “hallucinate” from time to time. This means that you must check what AI is telling you, as there is a good chance what it sees might not even be there at all.