I chose to work with the AI-assisted distant read, as well as the Voyant Tools website, for this choose-your-own-text adventure. For my text, I chose Moby-Dick; or, The Whale by Herman Melville. This was an interesting text because of how old it is. It does not use the same language that we use today, so analyzing it was both interesting and a little difficult. I started by putting the text into Voyant. Voyant offers an in depth analysis of the text, with many different graphs and tools that can be used to gain a deeper understanding of it. My favorite graph that I came across was the TermsBerry. This tool offers an overall sense of what words are used throughout the text. Here is a picture of the TermsBerry:
For my second option, I used the AI guided distant read. I thought this offered a great start to analyzing a text, however I would not trust it to get the entire job done. I copied and pasted the prompts from the worksheet directly into the AI bot, and there were many times when it could not give me the correct output that I asked for. Sometimes the responses were too complicated, and other times they were simply incorrect. What I do like AI for is getting an in depth sense of a text. It is able to run through the text, pick up patterns in word usage, and cross check information to generate a thoughtful response. Here is an example of the AI attempting to identify a theme in the text:
In my opinion, it is better to stick to applications and websites designed for advanced text analysis rather than relying on AI. While AI improves every day, there are still many reasons not to trust it with an advanced project like this one. AI can help you get started in thinking about a text, but it cannot do the entire job. That is where I would prefer a website like Voyant. In this “Age of AI” I feel that we should be most concerned about AI doing complicated jobs and being wrong. Using AI is a great tool but in order to do the job correctly more effort is needed. Here is an example of AI not being able to complete my prompt:
I agree that even now, it is certainly more trustworthy to use tools specifically designed for advanced text analysis (such as Voyant) than general tools such as Gemini. It was surprising seeing that you couldn’t get a word cloud from Gemini! I also agree that is important not to base all your conclusions on AI’s thinking. Rather, have it just influence a part of your thinking.
I’m really surprised to see that an LLM gave you that sort of response to a relatively simple task! I think you’re right that we can’t trust AI enough to do this sort of analysis, but hopefully the technology will improve in the future. However, I see how it can be nice to receive a more synthesized response from a chatbot rather than just seeing all the data from Voyant.
I like the approach you took with this assignment, Austin. I also had a similar experience when asking AI for a word cloud. At first, it gave me a straight up no. However, after repeatedly reframing my question, it finally gave me what I was looking for. I think this really highlights how specific you have to be when working with a program like Gemini or Chat GPT. Without nearly perfect prompts, it can be very challenging to get the output that you want.