Mastering AI Tools for Research - CM #24
Turning AI automation and AI Agents into trustworthy assistants
Hello, from the jungle of Koh Samui (TH).
Sometimes I forget how privileged I am.
I am in one of the few spots on this island where commercialization and tourism haven’t had their way yet. Around me are trees and palms of all kinds, a couple of brown buffalos, and tens of squirrels that zip and lounge from one palm tree to the next.
When like now, I am sitting on the balcony to write this newsletter, I can see the sea.
I am telling you this, so that I myself pay attention to it.
You know, once you move into any paradise, not as a tourist, but as a resident, you slowly forget how privileged is all this beauty you have around.
It rapidly becomes a given.
But it ain’t.
I am so rich, I need to remind myself.
And I need to be sincerely, heartfelt thankful to you reading these lines now.
Because if it wasn’t for you, I would have never had the courage to explore life as you have allowed me to.
Thanks to you I am no longer surrounded by cement, metal, cars, asphalt and stressed out people, running all day, consumed by their smartphones and with no free time in their lives.
I did not end up here accidentally.
We’ve been on a journey together, and I wanted to thank you for it.
Robin Good
In this issue:
The focus is on what I have discovered by using AI and agents to do my monthly zero-cost thematic tools research and collection.
At the start I was a bit skeptical, but very curious to see what these emerging new automation tools could actually do in this type of situation.
But the results have indeed surprised myself.
AI cannot replace me or do better than me in terms of pure research, data collection and vetting, though she’s coming quite close.
By collaborating with it, AI opens up for me new opportunities to extract valuable insights I could not see before, as it frees me from the more tedious, time-consuming and highly repetitive tasks.
AI by itself can’t go very far without instructions and very specific rules. But if you train it well, by sharing with it your own steps, criteria and rules, it can do a surprisingly good research and verification job.
Mastering AI Tools for Research: Turning Automation into a Trustworthy Assistant
.
Challenging the Notion of Manual-Only Research
I always thought that to do quality research, you had to manually go out and search one by one all of the options, leveraging your sensibility and experience in your field, to identify valuable stuff from the rest.
That is research must be a manual job carried out by a subject matter expert who knows well what he is searching for and has done it many times before.
I felt that AI tools lacked the specificity and trustworthiness needed to carry out quality research that brings back precise and accurate results.
I was wrong.
Why Experts Hesitate to Trust AI Tools
But I was not alone.
If you look at official research studies, you will find that while professionals such as researchers, writers, and experts recognize the potential benefits of AI, many remain cautious about fully adopting AI tools for their research and data collection.
A 2021 study by Springer Nature found that over 60% of researchers still use manual methods like searching through search engines, databases, or browsing journals to discover research articles and resources.
That is most researchers prefer manual approaches, even though there are many powerful automation tools and AI-driven search solutions, mostly because they:
trust more the traditional research methods and
feel that automated tools cannot really be relied upon
A recent survey by Elsevier, published in 2024, revealed that while 96% of researchers believe AI can accelerate knowledge discovery, a significant portion express concerns about its accuracy and potential for bringing back unreliable information.
How to Teach AI to Match Your Expertise in Research
What you don’t hear or read about is the fact, as I have personally discovered, that you can entrust your AI: that is, you can teach it how to become trustable, by simply transferring to it the same steps, actions, criteria and vetting procedures you use yourself to verify any information.
You can indeed instruct your AI / agents on:
1) where exactly to go and search. Which sites, which search engines.
2) what exactly to look for and extract
3) how to evaluate the quality the information they find
4) how to verify whether the info is accurate
5) how to find other similar / equivalent information / resources
6) what to exclude and not take into consideration
7) how much time to invest in each step
8) how many iterations to run on each of these tasks
9) how to organize and present you the final results
The Key to Reliable AI Results
How can you pretend that a non-expert, that does not have your experience or sensitivity can do a research work as good as yours?
Even if you take a highly paid and reputable freelance worker, or even a research professional, no-one will bring back to you better and more relevant results than the ones you can produce on your own.
To achieve that you need to train your collaborator, by sharing with him/her, your preferred search sources, the criteria and steps you use to vet and verify what you find, and the personal filters you use to decide whether some resources provide original unique value to your needs or not.
AI can guess, and if you let it will.
That’s why, the more specific, accurate and detailed your instructions and criteria are, the better the results that AI and agents can bring you back will be.
A Real-World Test: How AI Transformed My Research Process
A couple of weeks ago, as I started for the very first time to test a new AI tool to assist me in researching and putting together a collection of free social media scheduling apps, I crashed head-on into the issues of AI reliability in finding what you need and bringing you back reliable data.
As I started verifying AI agents work, I started noticing that among the many correct and valuable resources and info collected, there were overlooks, errors and plain mistakes.
That’s when I thought: “The AI agent made an error because it did not have enough specific information on what to look for, some of my requests were a little bit vague and could be interpreted in different ways, often agents had no instructions on where to look for what I needed or how to check whether what they had found was really correct.”
And so, I went into the AI system I was using and checked patiently each and every instruction that those AI agents had received before performing their research.
I added more specific actions and steps to take that closely reflected my own way of researching and vetting.
And blam, results started to be all very consistent and reliable, there were no more mistakes and I was surprised by how simple and effective this new approach was.
The final result was to my own surprise one of the best researched and most useful collections of tools I have completed in the last few years.
You can check the collection and the insights I was able to arrive that right here:
Social Media Assistants - Good Tools #23
*The AI research tool I have used and which I would recommend to anyone doing this type of research work, is nothing short of revolutionary and I’d highly recommend you to check it out. Its name is Ottogrid and there’s a free plan available with which you can test it.
The Benefits of Automation: Speeding Up Research While Gaining Deeper Insights
I understand that it can be challenging for some (who have never done it before) to articulate in a detailed manner their research and evaluation process, but by analyzing and codifying its key elements one can drastically improve the research results that AI and agents can bring back.
As a consequence, you can not only drastically reduce the amount of time that manual research normally requires, but you can benefit from the extra time gained to identify interesting patterns and trends as well as to uncover valuable insights into the nature and differences of the resources you’ve found.
If I had done this by hand, I would have likely found and collected the same amount of tools but I would have not been able to extract as much insights and useful info as I did.
The second most beautiful reward of all, when using such systems, is that once you have properly configured and trained your AI agents to do exactly what you yourself would have done, you can update your whole collection at periodical intervals by just pressing a button.
Try to do this (update and verify a collection of resources after some time) by hand and you’ll fully understand how powerful, useful and beneficial these AI agents can be for researchers.
Readers Poll - Improve this Newsletter
Discover How 100 Entrepreneurs Have Monetized their Collections and Directories
Real-World Examples of Online Projects that Generate Profits by Organizing Existing Info
The Curation Monetized *Database Catalog* 2024
Premium database of 100 real-world examples of curation monetized businesses containing:
100+ Real Businesses that make a profit
by curating and organizing existing infoBusiness model(s) used
Publishing/payment platforms used
Direct contacts
+
Top Curated Business Models and Opportunities
Top Key Trends in Curation Monetized
Pricing:
100% free for Curation Monetized Annual Premium subscribers*.
Curation Monetized Premium #24
In this issue:
Data-Verified Curation-Based Business Opportunities (3)
Directory of X companies
Examples of Y
List of Z Platforms
.
Real-World Examples of Curation Monetized (3)
Search Engine Directory for X Apps
Catalog of Z Films
Database Focused on W
If you have enjoyed this issue so far, please support this publication by clicking on the heart/like icon at the end or by sharing / forwarding it to a friend who would appreciate it.
Stop searching for something new.
Organize and add value to what is already there.
from sunny Koh Samui island (TH)
Robin Good
.