
There’s a dilemma in summarizing a journal article with AI.
If the generated summary is too high level, it will look generic and lack concrete substance.
If the summary is too detailed, it will be too long to read, defeating the purpose of reading a summary.
This is particularly true if you must read a long, super-detailed review article, such as guidelines or recommendations from the scientific community or a condensed yet long collection of expert opinions in a specific domain.
Sometimes, I feel lost when reading a long-winded review article. I’m swept away by the incredible number of details that I’m unaware of, and I’ll find myself busy taking notes and highlighting almost the entire page, but then I never reread it.
It’s tiring.
I wonder if we could ask the bot to think for itself and tell us what actually matters to read in the journal article. Then, from there, I would tell the bot what I need to gather materials to write an article or make notes for future review.
In other words, I’d ask the bot to provide a roadmap to show me where to collect those information nuggets that matter. Otherwise, I will be overwhelmed with too much information and unsure where to focus. As the saying goes, it’s like getting lost in the data forest and missing the important trees.
This approach allows the creativity and freedom of the bot to explore the content and tell me what matters. It’s particularly helpful when I’m reading unfamiliar topics.
Geeky Curiosity is a newsletter for generative AI enthusiasts that covers Midjourney, Perplexity, Prompt Engineering, and other AI topics.
Subscribe now to discover the tried-and-true, practical tips and insights in using AI to boost your creativity and productivity.
I recently crafted an advanced prompt to assist me in browsing through a large number of journal articles quickly. I will share the prompt with you and demonstrate how to use it.
The prompt was inspired by MyDpi’s “ThinkBoost Prompt” for Perplexity, which was posted on July 21, 2024 in Perplexity’s Discord server. It uses the Quiet Star and Tree of Thoughts methods to summarize an article.
What can the Advanced Summarization Prompt offer?
The prompt allows the bot to go through various materials and reflect independently. It uses advanced prompt engineering methods, namely the “Quiet Star” and “Tree of Thoughts” methods.
The bot will then tell you the most essential information and where you may concentrate.
It allows you to focus on a particular area of interest and delve deeper.
To eliminate hallucinations, it quotes the sources directly without paraphrasing.
The resulting Summary Report is structured as follows:
Part 1: Quiet Star Method: Create a high-level summary of the article’s primary topics.
Part 2: Tree of Thoughts Method: Expand and provide more explanation following the framework in Part 1.
Part 3: Summary combines the points from Parts 1 and 2. It also includes the Call to Action and the Conclusion sections.
Optional: Use another “follow-up prompt” to delve into a topic of interest and cite sources without paraphrasing.
How to Use the Advanced Summarization Prompt:
(The prompts are given at the end of this story)
Assuming you have a Perplexity Pro subscription, change the “Focus” to “Writing,” set the model to Claude 3 Opus, paste the Advanced Summarization prompt, and press ENTER.
(Optional) To ask for more information and its sources, paste the “Follow-up Questions” prompt. Then, tell the bot where you need further explanation and cite the sources.
Easy to use. No coding required.
Let’s use an example to guide you through the process.
Assuming we want to read a review article about how to handle cats. (Yes, they actually wrote a very long article about this!)
Here is the article’s citation and where you can download it:
Rodan I, Dowgray N, Carney HC, Carozza E, Ellis SL, Heath S, Niel L, St Denis K, Taylor S. 2022 AAFP/ISFM Cat Friendly Veterinary Interaction Guidelines: Approach and Handling Techniques. J Feline Med Surg. 2022 Nov;24(11):1093–1132. doi: 10.1177/1098612X221128760. PMID: 36259500; PMCID: PMC10845437. Link to the article: https://pubmed.ncbi.nlm.nih.gov/36259500/
You do not have to become a cat specialist to grasp the article’s content. It is just an example to illustrate how to use the summarization prompt.
Upload the article to Perplexity, paste the Advanced Summarization Prompt, and press ENTER.

This generates a Summary Report for the article.
First, you will notice the article’s title and year of publication.
Part 1 is a high-level summary created using the Quiet Star Method.
The high-level overview is more than just a recap of the table of contents.
Indeed, the bot goes through the entire material and condenses it into key elements.

Part 2 builds on the outline provided in Part 1 by providing further context and information.
It’s as if you were reading the authors’ key points in a PowerPoint presentation format.

Part 3 provides a 10-point overview of the article’s main ideas.
The Summary Report includes the Call to Action and Conclusion sections.
The Call to Action provides a brief overview of what needs to be done in the future. It highlights the gaps in existing knowledge and potential research opportunities.
The Conclusion is a paragraph that summarizes the article.
Often, we want to learn more about a specific topic covered in the article but do not have time to read the entire paper.
Use the “Follow-up Questions” prompt to instruct the bot where to focus.

After receiving your input, the bot will provide a full explanation.
In this example, say we’re interested in Part 3, Point 2 of the Summary Report. In addition, we’re also interested in keywords related to “protective emotions” and “cat handling.”
So, we write the following as our input:
(a) Part 3 point 2
(b) protective emotions, cat handling
The bot will pull out the details related to Part 3 Point 2, providing clearer and more detailed explanation.
For the keywords, the bot will extract the main ideas related to them, saving you time searching for those scattered bits and pieces of information across the article.
The bot will also provide direct quotes from the sources (without paraphrasing) to support its main points.
This is useful for determining the exact location of a sentence within the article and avoiding hallucinations.
I hope this story will help you navigate scholarly journal articles and boost your research or study productivity!
Here’re the prompts:
(1) Advanced Summarization Prompt
###### Geeky Animals Advanced Summarization Prompt V1.2 ######
You task is to review the attached materials and prepare a detailed, summary report on the article.
The report has three parts.
Use markdown to format the report. Bold the headers. Put separators between the parts.
###########################################################
Part 1: Quiet Star Method
Use the Quiet Star method to generate a list of key aspects discussed in the article.
The Quiet Star method involves calmly reflecting on the subject and noting the various facets, ideas, and important questions that come to mind, like the branches of a star.
Write and format the key aspects into a list.
Part 2: Tree of Thoughts Method
Use the Tree of Thoughts method to explore each of the identified key aspects in depth. For each aspect, consider different perspectives, implications, possible causes, and consequences.
Let your thoughts branch out freely like the branches of a tree. Capture your reflections for each key aspect into a list.
Part 3: Summary
After completing the Quiet Star and Tree of Thoughts exercises, take a step back and consider the lessons learned.
Summarize the article in 10 points. Remove the citations from the response.
[Call to action] Provide a call to action. Bold the sub header.
[Conclusion] Create a concise conclusion that restates the main points and offer the final thought. Bold the sub header.
###### Format the report in this way: ######
Italicize all scientific names.
#Title of the article (year of publication)#
---
### Part 1: Quiet Star Method ###
1. X
2. X
---
### Part 2: Tree of Thoughts Method ###
1. X
2. X
---
### Part 3: Summary ###
1. X
2. X
#### Call to action ####
(not bold)
* X
* X
#### Conclusion ####
(not bold)
X
(2) Follow-up Questions Prompt
###### Geeky Animals Follow-up Questions V1.2 ######
You task is to provide answers to the user's questions and generate a detailed report.
The report has two parts.
Use markdown to format the report. Italicize all scientific names.
Prompt the users to provide the following input:
(a) The section(s) of the report that require further information
(b) The keywords of interest
Part 1: Explanation
1. Use input (a) and explain your answers in detail. Please provide examples if available. Do not look for additional information on the internet.
2. Regarding the responses in Instruction 1 of Part 1, please provide specific information about the User input (b), if available.
Part 2: Sources in quotes
1. Repeat Part 1 but this time only cite sources from the attached materials in quotes. No paraphrasing.
###### Format the report in this way: ######
Italicize all scientific names.
## Name of the section(s) ##
---
### Part 1: Elaboration ###
XX
---
### Keywords ###
XX
---
### Part 2: Sources in quotes ###
(not bold)
* X
* X
---
### Keywords ###
(not bold)
* X
* X
Conclusion
Summarizing an article can be challenging since it requires striking a balance between being high-level and providing concrete information in a concise manner.
The “Advanced Summarization” prompts are inspired and adapted based on the “ThinkBoost Prompt” method, which uses the Quiet Star and Tree of Thoughts advanced prompt engineering techniques.
The prompt generates a structured Summary Report that includes a high-level overview, a more extensive explanation based on the overview, and a ten-point summary.
Users can use the “Follow-up Questions” prompt to delve more into subjects of interest from the Summary Report. The bot will also provide direct citations to help eliminate hallucinations.
Affiliate Disclosure
Perplexity is available for free.
You can also subscribe to Perplexity Pro and gain access to premium features such as Claude 3 Opus, GPT-4o, Sonar Large 32K, Claude 3.5 Sonnet and Perplexity AI models, unlimited file upload, and others that will supercharge your productivity.
In other words, a single subscription grants you access to multiple AI models!
Begin with a $20/month Perplexity Pro subscription.
Subscribers are eligible for a full refund within 14 days of their initial purchase.
Use this referral link to save $10 on your subscription.
https://perplexity.ai/pro?referral_code=YQGPR1B9
I’ll also get $10 off my Perplexity Pro subscription when you sign up. Thank you so much for your support!
I hope you enjoyed this post!
Please subscribe, like, and comment so that more people can discover this newsletter.
Please share with others if you find it helpful.
Non working on perplexity, throwing a generic error. Any specific model to use it with?
Great article with goos explanations and examples. Very useful for researchers . Thank you!1