I have decided to incorporate in the pages of Salud Everywhere a new option, related to artificial intelligence. It consists of a NotebookLM containing the bibliographic references I have used to write that page, and which I recommend as the main source of information about it.
Table of contents:
What is NotebookLM?
NotebookLM is an artificial intelligence assistant specialized in literature review and research tasks.
It is a Google product based on Gemini that, unlike other more popular tools, has several advantages that seem significant to me:
- It allows you to create openly shared digital notebooks with other people, where the AI only has access to the bibliographic references you add as sources. You can include up to 50 sources per notebook. I never use that many for a single page, so that’s more than enough for me.
- You can ask it anything about the content of these references and, although it sometimes hallucinates and gives odd answers, it usually does a good job. It works especially well when the user selects only the sources they want to use.
- When responding, it does so in an organized way, clearly indicating which bibliographic reference it uses for each statement. In fact, it puts more emphasis on the source than on the answer itself. It not only mentions the document but also clearly shows the specific paragraph it is based on. This allows the user to verify everything they read and know which document and which part of the document they should consult to learn more about the topic. This is especially useful when reviewing long reports.
- NotebookLM does not replace the work of a literature review. Reading a scientific article from the first page to the last or the chapter of a report still seems to me the best way to understand and absorb them. However, it does help identify which sources are most relevant for a specific topic, or which pages of that 300-page report you really should read carefully.
What does it contain and how can I use the NotebookLM on each page?
Each NotebookLM contains bibliographic sources carefully selected by me.
In the right column within each page is the list of all the bibliographic references that I have used for documentation and that I recommend on that topic. In parallel, the NotebookLM of each page contains (almost) all those references. NotebookLM only consults those sources when responding to people who use it.
Aren't all the sources there? No. There are many, but not all. I only post reports that are completely public (for example, from WHO, IASC, ALNAP or other organizations) and open access scientific articles. I don't dare to upload paid scientific articles or copyrighted books, even if they are freely available on the internet, in order not to get into trouble. On pages that contain few references, or where only a few of them are open access, I have simply not done NotebookLM.
If you see something that shouldn’t be shared in NotebookLM, please let me know. I have no intention of making any profit from this initiative, and the last thing I want is to get into legal trouble.
What can I ask NotebookLM?
The best thing is to try it yourself, I think. I find it especially useful to ask “Which documents address topic X and how does each one approach it?”, “Explain this idea X to me” or “How has the understanding of concept X changed over the years?”
Also, sometimes it works to ask “Give me evidence-based arguments to support or reject the following statement”, just as another example. And to learn beyond what these sources contain, you can also ask “Show me areas of interest that require further research beyond these sources”.
In general, I think it can be a great tool for visitors, after reading the article, to cross-check the information on their own, verify its sources, and have a starting point to continue exploring the topic in depth.
What are the "Let's pause and reflect" blocks?
Reading is wonderful, but it is not always enough or the best thing to learn. I, personally, learn much more when I force myself to write about a subject. To do this I have to analyze the information I get from different sources (or consult it again with a clearer intention), order it and understand it well. Only then can I explain it myself in my own words.
This, to a certain extent, is what I am looking for with the «Let's pause and reflect» blocks that you will find in the pages of Salud everywhere. In them I pose an open and complex question that invites you to actively think about the topic you are reading about before continuing.
The idea is very simple: you take a few moments to switch off the autopilot and think, type your ideas, press the button, and a NotebookLM specific to that page opens automatically, with a selection of bibliographic references curated by me. At the same time, your question is copied to the clipboard with a prompt explaining to Notebook how to analyze it.
As soon as you hit paste (Ctrl + V or Command + V) and send it, NotebookLM will start working. It won't tell you if you got it right or not. They are not hit questions. It will give you critical and constructive feedback based only on those bibliographic sources, pointing out aspects you may not have considered and helping you to go deeper into the topic. It is like having expert tutoring on that topic, available to you for free at any time.
Does this mean you are a staunch advocate of artificial intelligence?
No. Or depending on what for. I like writing for humans and reading things written by humans. Just as the great Oatmeal (a.k.a. Matthew Inman) says that AI-generated art seems to lack heart, I also get enormously irritated when reading texts that people with little interest in writing try to pass off as their own. I prefer an imperfect text with a human touch over a cold, sterile one. And mind you, this also applies to the insufferable “scientific language.”
However, I do see value in tools that help me review extremely long reports that I would otherwise never manage to read, or translate what I write into a language I don’t fully master.
Blog
The blog posts in Salud Everywhere expand its content on humanitarian aid and cooperation, health in humanitarian crises and career advice with news, opinion and analysis.
External links
- NotebookLM.
- The oatmeal, 2025. A cartoonist's review of AI art.
- NotebookLM subreddit.