Voyant-tools
For my blog post, I chose to focus on Beowulf, which is a famous epic poem by an unknown author. I first tried out the link feature on Voyant-Tools. There were many different linked words. By clicking on the different words, it caused additional linked words to pop up.
Gemini
For my second text analysis experiment, I used Google’s Gemini. I followed the steps given to us from Anastasia Salter. Using the same wording as Anastasia Salter, I asked Gemini this prompt: “Can you visualize the top 30 words as a word cloud?” Gemini was able to create a decent word cloud, which was simple and easy to look at.
However, AI struggled with making a clear and cohesive character network chart. The first relationship network chart it made wasn’t very clear. A few of the character names were cut off, and it’s a bit hard to read.
When I asked Gemini the prompt “can you redo the last step, making it clearer and easier to read?” it created the image below. It enhanced several features. It used a deeper blue, which made it easier to read the white text on top. It also made the relationship labels bold, which made them clearer. However, the issue of names being cut off remains the same.
Throughout this process, I reviewed the reasoning that Gemini gave. From my knowledge, I found that most of it made sense. However, I am not confident in my ability to recognize mistakes Gemini would make doing text analysis.
range of approaches & Concerns in the age of ai
There are a wide range of ways to computationally analyze textual data and relationships. Voyant-Tools, for example, is effective and easy to use. There are multiple different methods, such as word clouds, diagrams and more. Although Gemini works to analyze data, it risks skewing the data or creating incorrect results. In the age of AI, I am concerned that using AI to analyze text would be risky if the person using it is not an expert. It may be difficult for them to recognize any mistakes that Gemini makes. Another concern I have is that the use of AI limits cognitive effort. This is because AI focuses on the content produced, rather than the process.