4 minute read

I have spent about seven working days learning about five different topics that came out top in the decision matrix that resulted from my process for choosing a focus for my research going forward. You can read my notes about each topic here:

  1. AI for translating unseen languages
  2. Extraterrestrial technosignatures
  3. Recognising AI sentience
  4. Exoplanet biosignatures
  5. AI misuse: pathogenic DNA

This time has given me a decent overview of the state of each field before I choose one to dive into in more depth (doing a proper literature review is the next step of my research process). I planned to spend one day on each topic, and in most cases that is fine for scratching the surface, but I found that two days was better for a proper overview.

Here’s my process for a two day review of a research topic.

Day 1. Gather learning resources and get an introduction

Spend the first day gathering and ranking resources to learn from, filling holes in your basic knowledge and getting an introduction to the field.

Type a few different search terms into Google. If a field is completely new to me I will search for popular terms first (popular science articles will contain these both for ease of reading and for search engine optimisation) in order to identify the more technical terminology used in a field and then search again for those. For example, when searching for information on recognising AI sentience, I found that a common term in use was ‘artificial sentience’ and so used that as my next search term.

Open the top hits and check the source is reputable and trustworthy (i.e. is it a well-known science institution, website, blog, magazine or textbook - if not check the about page for credentials).

Scan the webpages and close any that are not relevant or too shallow.

2. Collect recent review articles

A recently published review article is one of the best ways to get an in depth overview of a topic.

Follow the trail of citations (check the cited by and the reference list) to gather more papers. To start with, I will use only open access and later, if necessary, I will try to get the PDFs of the ones behind a paywall (many are posted as pre-prints on arxiv, biorxiv or on one of the author’s webpages or research gate profiles).

I use Mendeley to organise and read papers online - it has a useful chrome plug-in that lets you import (and find more) papers.

3. Make a playlist of educational videos

YouTube is a great resource: most topics will have some educational videos on there. A lot of conferences post talks on YouTube which can be invaluable for learning about the latest research and opinions in a field that have not made it into publications yet. You can also find seminars and lectures that are great for an introduction. Again, check the sources.

Make a playlist of any relevant videos you find.

4. Start broad to assess and refresh current knowledge

I usually start with a popular science article or Wikipedia, even if I think I know the field. It often puts the topic into a broader perspective and they are usually quick to read.

Identify any gaps in your basic knowledge and fill them as you go along with the help of Google and/or Wikipedia.

Watch a seminar or two on the topic. Make notes of any key ideas, terms, questions, people and institutions.

Look up any relevant institutions, experiments, facilities, existing data or source code repositories.

Day 2. Deepen and summarise knowledge

Dedicate the second day to diving deeper into a topic using review papers and writing down what you have learnt.

1. Read papers

Scan the latest review article and read the introduction and any sections that will deepen your knowledge without overwhelming you with technical details. Highlight any key ideas or references. If there is a future perspectives section also read that. If the review builds on an older review or a complements any other review articles (sometimes these will be together in a special issue of a journal) add these to your library and read them in the same way. Look up any unfamiliar terms.

Don’t waste time reading very technical papers or detailed studies at this stage - save that for the in depth literature review if you decide to pursue the topic. The introductions of these papers can, however, be valuable to get additional background and/or to find other review articles.

2. Write a summary

Summarise all your knowledge on the topic by writing it down.

Make a list of key references, resources and further reading. This will be useful if you pursue the topic further or come back to it later. Consider both what is known and what is currently unknown.

List any challenges (e.g. lack of data) that might affect whether or not you can make a contribution to the field.

You will notice as you write that you have missed something so go back and search or read a bit more to fill any gaps.

Remember that two days is not enough time to cover everything and don’t get side-tracked by off-topic or overly technical papers.