Janell Sihay

How I study

Today in my Marine Toxins class, our professor asked us how we approach reading journal articles. My classmates shared interesting techniques, such as going through the paper and writing down one's understanding before going to Google for clarification. I might try it sometime. As for me, I utilize AI tools such as Google Scholar AI outline and NotebookLM, especially with topics I'm not familiar with. I am aware that using AI, particularly GenAI, is frowned upon in academia. I actually understand why. I tried using ChatGPT to summarize papers before, but I find that I do not understand the paper when I do that. I also think that I'd miss the entire point of getting into the PhD program, which is to develop critical thinking skills necessary to produce new knowledge. So, when reading papers, I make sure that I spend some time "interacting" with the text so I better understand the story.

My current reading process goes like this:

  1. Open the text with the Google Scholar plugin, read the abstract section, the conclusion, then go through the AI Outline (if available). I find that this step builds curiosity and interest, encouraging me to read more. Before, I would plan to read the paper from start to finish, but I would end up feeling overwhelmed, especially when I encountered unfamiliar concepts.
  2. Now that I have a simplified overview of the paper by reading the abstract, conclusion, and the AI outline, I'll know where to focus my attention first. Usually, I try to first check how the objective was achieved. I look at the figures, and when there's something still unclear even after reading the figure caption, I'll look into where it is referenced in the text. This is where NotebookLM has been a huge help. Basically, you can ask prompts for clarification, and it will only reference the specific text you linked. This way, it doesn't make up anything (called "hallucinating"), and it will point you to the specific part of the text where it got the idea. The only feature that kind of scares me is the audio overview feature, which basically turns the text into a podcast. It scares me because it's like I'm listening to real people, the way they talk, the way they construct sentences. But surprisingly, I find it really helpful. Just like the simplification in Step 1, this approach leads to more questions, which makes me want to go over the text one more time. I think that this setup doesn't take away the critical thinking. In fact, I think it reinforces it more, since I'd have to double-check whether the text actually says it.

The only problem with my current reading process is that it is not integrated with my personal knowledge management system. I use Obsidian, but I don't use it to its full potential. I'm trying out different things and see which one works for me. I'll update you soon!

#phdone #reflections