How I do Research (right now)
The following is a description of how I do research at the moment, especially how I read papers and keep up with my areas of interest. I am putting it in public so that more experienced researchers (or even peers) can help me improve my strategies.
- I use AI tools heavily in the literature discovery/review process:
- For finding papers, I use Exa and Perplexity heavily. I will also occasionally use Google Scholar or Google Search, but I mostly use Google Scholar to see which papers cited the ones I found using Exa/Perplexity.
- I will usually try to collect these in the location where I will eventually do the work (write the paper or code), per the advice of Cal Newport.
- Occasionally, I will end up just putting a lot of papers into a Google doc, which is nice because I can jot some notes there if I need to.
- This is a major change in the past ~year compared to how I used to find papers using a less efficient combination of tools that would frequently miss key papers for long periods of investigation.
- When reading papers, I take notes on one (1) sheet of paper, trying not to go over that unless the paper is exceedingly long.
- These notes help me later create a PPT slide summarizing the key parts of a paper (well-made figures are a huge help for this).
- Every so often, I will try to summarize a batch of papers/the current state of my knowledge
- After I decide on my favorite papers (usually after reading a batch, but sometimes right away if I think they are really important), I will go to some of the papers they cited most often, and crucially, the papers that cited that good paper. This allows me to get a better feel for very current work and how people have expanded on the good ideas proposed in the original paper.