Really love these tips! I’ve started using some recently, and they’ve been super helpful. Funny enough, I’ve also noticed I remember things better when I read non fiction or other books. Thanks so much for sharing.
Hey Joshua, love this question! To answer briefly: yes, for sure, you can do both.
But I think it’s better suited for a Custom GPT than a Project, since Projects hold all the context together, which might get messy if you’re using it across different books, and it might mix them up.
Thanks, Michael! I appreciate you taking the time. I’m currently working on a platform that will include all my prompts, workflows, and resources in one place, much easier to access and use.
Love hearing that, Michael. And yes, sometimes you gotta fight through a few cobwebs before it clicks :)) but that’s where the real insight comes. Appreciate you sharing how you’re experimenting with it.
That’s so amazing to hear!!! And such a great reminder for all of us, that repetition is key. Really appreciate you coming back to this and sharing it with me.
Thanks, Neil! For sure. But now, since GPT can read text from images (even handwriting), you can just snap a photo of any part of a physical book you want to dive into. And if it’s a well-known book, chances are it’s already somewhere in its training data.
Plus, the way I structured the prompts, it’s not about GPT spoon-feeding you summaries. You’re the one doing the thinking, GPT just helps you reflect. So it works either way. Uploading the full book is ideal, but not mandatory.
YESSS to all of this! 🔥 AI isn't here to replace our thinking, it's here to amplify it. I've been using similar strategies with my coaching clients and the transformation is insane.
The Feynman technique with AI hits different though. I actually have my clients do this after every automation workshop - having them "teach" ChatGPT what they just learned. The gaps in understanding become SO obvious when you have to explain it simply.
The real game-changer for me has been that connection prompt - linking new concepts to existing knowledge. I use a variation of this when I'm researching new AI tools. Instead of just collecting features, I'm connecting them to actual client problems I'm solving.
Love this, Tiff!! Appreciate you sharing your own grounded, real-world examples.
Totally agree on the Feynman angle. Teaching is hands down one of the best ways to learn.
I felt that big time earlier this year while building a career tool to guide people through their next job move. We used the OpenAI API to help users step-by-step, from discovering career direction (via personality quizzes and past experience analysis) to actually landing the role (CVs, LinkedIn, interview prep, all of it).
I learned way more building that product and coaching early users for free (just to understand their struggles) than I ever did back when I was job hunting myself a few years ago.
There’s something about teaching and doing at the same time. It just locks things in.
Loved this. I actually built a custom GPT a couple days ago to play a curious student/younger sibling (it's called "Curious Companion" and it's here if anyone's interested - https://chatgpt.com/g/g-68407e32d698819182b7344272edf2e8-curious-companion) so I can learn by teaching, and I've been using it a lot since then. It's really fun to use too, and has already sparked some new ideas. One trick I added is that I instructed it to occasionally misunderstand my explanations, in order to keep me on my toes and deepen my understanding by having to correct it too.
This fits with a lot of what's in 'Ultralearning' by Scott Young. The key points he makes are to (1) apply what you're learning ASAP, in as close to the real world application as you can, (2) practice active recall, forcing your mind to memorise it rather than relying on the text or notes in front of you, (3) create drills, breaking down a skill into its component parts to work on them one at a time, (4) explain it to others (the Feynman technique you mentioned), (5) get as much feedback as you can, so you know what's working and what's not.
One interesting tidbit in the book is that it helps to do a test before you've even *begun* studying -- it primes your mind to seek out and retain those answers so you perform better later. Perhaps we could add a "Phase 0"?
What's really great is this is all "swimming upstream". It's the opposite of getting AI to do the thinking for you, it's getting it to prompt *us* to think harder, better.
Learning by teaching is seriously one of the most underrated ways to actually get something. Curious Companion sounds great, love that it adds a bit of friction too.
The Forgetting Curve, as became known, was established by educational researcher Hermann Ebbinghaus in the 1880s. We forget +90% what we read/hear within 24 hours unless it's reinforced somehow. That research was repeated by Scandanvian researchers (Finland? can't remember) in the 2010s and found to be valid. Doesn't that have powerful implications for our educational system?
Which gives me something to write about tomorrow, so thanks. :-)
Looking forward to reading the post and seeing what angle you will approach.
I actually had a section with stats on why we forget what we read & some other things, but cut it last minute.
And you’re right, the implications for education are huge. We’ve got way more tools now to teach, learn, and actually retain. The challenge is shifting the system out of its pre-AI mindset and into where we are now.
In the U.S., our educational system often confuses activity with accomplishment. We continue to emphasize standardized testing (cognitive skills) and almost completely ignore emotions, the primary driver of behavior. Meanwhile, the world is burning down. Is AI useful? Absolutely! Will it turn the tide in education? No.
Same story in developing countries, and I’d bet it’s even worse. Emotion’s always been sidelined, right alongside the life skills people actually need to navigate the world.
Great ideas to retain information better. I recently started using one of those AI Whisper (speech to text) apps that is smart. Seems like a great use for this when reading. I don’t always want to type and I could use these principles to do the same things but through voice
Zach, I run a Linux laptop and would *love* to use Whisper, but it seems hard to install on Linux. How do you do it? DM me if you'd like. And thank you in advance.
You can use basic whisper integrations but the ones I like are MacWhisper (free tier) and wispr flow. I find the iOS app is surprisingly good: https://wisprflow.ai/r/ZACH98
It has one more step above whisper which is an AI model corrects your language before it pastes. That’s my referral code from inside the app. I don’t know them at all or have any affiliation.
I wish there was a free version as good, because I do not like that is is transcribed in the cloud. I would prefer a local one.
This is really turning reading a book with more scientific approach. It opens up to new possibilities and now I’m beaming with ideas haha
1. I can upload the book pdf to notebookLM, extract each chapter and ask Gemini to generate quiz for me using canvas mode.
2. I usually save some quotes on my kindle and export it to google sheets. I’m thinking to ask AI why I’m so resonated with these quotes. What does it tell me? What can I do with it?
I actually played with NotebookLM and uploading the book too, but saved that for another time or maybe a follow-up post :))
The second one sounds really great. Would love to hear what patterns show up when you run that, keep me posted. Been obsessed with reading since forever, so I'm always on the hunt for better, deeper ways to make it stick.
Will update you. Reading N/F? Just like how you said…notes, scribbles, highlighting, underlining….although it never helped in retaining information. But when I reread those same books later, it leaves a mark on what and how my thought process was at the time. 🙂
I know exactly what you mean, and every time you re-read a book, you’re left with new insights and notice different connections. It really is about the time you’re at, you’re right.
To be honest, I’d feel a bit daunted by the “summarize & teach back” approach right from the start. I’ve never been a strong reader myself, finishing a non-fiction book (or even fiction) has always been a challenge.
I tend to learn more through action and referencing back as needed. But your guide really opened up new possibilities.
Curious, what platform or tools are you using to put these exercises into practice?
And how much time does it usually take you to get through a more sophisticated book?
My sense is that while it might not save time, the depth of learning would be much greater.
You’re totally right, this approach doesn’t save time at all 😂 If anything, it slows reading way down.
But for the right books, I actually prefer that. I used to spend a lot of time trying to make ideas stick anyway, so this feels like a more joyful and engaging version of that, as I can bounce off ideas with GPT.
As for platform, I’ve mostly been using GPT to talk through the ideas (any LLM should work anyway). Also NotebookLM is great if your notes or book are already digital.
And one more sophisticated book, as you say, was Antifragile by Taleb. Took me a while because I was actively applying it to my startup as I went. Didn’t time myself tho.
I can totally see how, for the right books, slowing down is worth it, you really want to internalize the ideas. Deep respect for the rigor you’re putting into it!
Antifragile is such a classic. So brilliant that you’re applying it in real time to your startup, that’s the best kind of learning.
That’s such an interesting observation, it's probably because fiction naturally pulls you into a world and makes you feel something, so the ideas land deeper.
Really love these tips! I’ve started using some recently, and they’ve been super helpful. Funny enough, I’ve also noticed I remember things better when I read non fiction or other books. Thanks so much for sharing.
So good to hear you’re already using some of this to better process books! It really is a different experience.
I guess non-fiction might stick because it feels useful or immediately applicable, while fiction hooks you emotionally through the story.
Both land, just in different ways.
I wonder if I can create all of these prompts into a Project feature to go back to or create a GPT on all of these ideas.
Hey Joshua, love this question! To answer briefly: yes, for sure, you can do both.
But I think it’s better suited for a Custom GPT than a Project, since Projects hold all the context together, which might get messy if you’re using it across different books, and it might mix them up.
I invested time to store your prompts where it is easy to insert them into a chat. I think I could have saved the post and my time.
Whatever, I don't think it was a bad use of time, because I want to figure out how to use all this new stuff I keep finding on Substack.
Thank you
Thanks, Michael! I appreciate you taking the time. I’m currently working on a platform that will include all my prompts, workflows, and resources in one place, much easier to access and use.
Glad you found value in it so far!
Hi Daria. I worked with 2 prompts on notebook lm today and used Wyndow’s post on James Clear habits.
I had to clean out a lot of cobwebs but the thinking machine was able to function a bit.
I was able to explain a few things to my wife. Not all mind you.
Thanks again.
Love hearing that, Michael. And yes, sometimes you gotta fight through a few cobwebs before it clicks :)) but that’s where the real insight comes. Appreciate you sharing how you’re experimenting with it.
Super curious - what book are you diving into?
I have not used it for a book but a post by Wyndow
That is a good start for me for now.
https://open.substack.com/pub/aimaker/p/how-i-used-james-clears-atomic-habits-build-ai-systems-run-my-life?r=44d6c&utm_medium=ios
And sorry, I misread when you said it before
oh, got it! That’s a really cool use case, works great for long-form posts like that.
It’s easy to forget about one good idea when I find so many on Substack daily.
I returned to this post and found more prompts to use.
I’m excited to sort out confusion next time I study.
I will be confused for sure for sure.
Thanks Daria
That’s so amazing to hear!!! And such a great reminder for all of us, that repetition is key. Really appreciate you coming back to this and sharing it with me.
I have explored your Substack very little because each post is so important and helpful.
Such a great approach in this fast-paced world where information is consumed so quickly and superficially. Thanks for the reminder, Michael.
Great article. I bet this is easier with digital books.
Thanks, Neil! For sure. But now, since GPT can read text from images (even handwriting), you can just snap a photo of any part of a physical book you want to dive into. And if it’s a well-known book, chances are it’s already somewhere in its training data.
Plus, the way I structured the prompts, it’s not about GPT spoon-feeding you summaries. You’re the one doing the thinking, GPT just helps you reflect. So it works either way. Uploading the full book is ideal, but not mandatory.
Another option is to use a program like BundleIQ to curate all your books,notes, other media, and your own content.
https://www.bundleiq.com/
Great tool, thank you for sharing Steven!
Really? My friend Nick Mohnacky is the brains behind that company. Awesome!!!
YESSS to all of this! 🔥 AI isn't here to replace our thinking, it's here to amplify it. I've been using similar strategies with my coaching clients and the transformation is insane.
The Feynman technique with AI hits different though. I actually have my clients do this after every automation workshop - having them "teach" ChatGPT what they just learned. The gaps in understanding become SO obvious when you have to explain it simply.
The real game-changer for me has been that connection prompt - linking new concepts to existing knowledge. I use a variation of this when I'm researching new AI tools. Instead of just collecting features, I'm connecting them to actual client problems I'm solving.
Love this, Tiff!! Appreciate you sharing your own grounded, real-world examples.
Totally agree on the Feynman angle. Teaching is hands down one of the best ways to learn.
I felt that big time earlier this year while building a career tool to guide people through their next job move. We used the OpenAI API to help users step-by-step, from discovering career direction (via personality quizzes and past experience analysis) to actually landing the role (CVs, LinkedIn, interview prep, all of it).
I learned way more building that product and coaching early users for free (just to understand their struggles) than I ever did back when I was job hunting myself a few years ago.
There’s something about teaching and doing at the same time. It just locks things in.
Loved this. I actually built a custom GPT a couple days ago to play a curious student/younger sibling (it's called "Curious Companion" and it's here if anyone's interested - https://chatgpt.com/g/g-68407e32d698819182b7344272edf2e8-curious-companion) so I can learn by teaching, and I've been using it a lot since then. It's really fun to use too, and has already sparked some new ideas. One trick I added is that I instructed it to occasionally misunderstand my explanations, in order to keep me on my toes and deepen my understanding by having to correct it too.
This fits with a lot of what's in 'Ultralearning' by Scott Young. The key points he makes are to (1) apply what you're learning ASAP, in as close to the real world application as you can, (2) practice active recall, forcing your mind to memorise it rather than relying on the text or notes in front of you, (3) create drills, breaking down a skill into its component parts to work on them one at a time, (4) explain it to others (the Feynman technique you mentioned), (5) get as much feedback as you can, so you know what's working and what's not.
One interesting tidbit in the book is that it helps to do a test before you've even *begun* studying -- it primes your mind to seek out and retain those answers so you perform better later. Perhaps we could add a "Phase 0"?
What's really great is this is all "swimming upstream". It's the opposite of getting AI to do the thinking for you, it's getting it to prompt *us* to think harder, better.
Learning by teaching is seriously one of the most underrated ways to actually get something. Curious Companion sounds great, love that it adds a bit of friction too.
We've gotta exchange the lists of books we read, Daria :-)
Haha, let’s :)
Amazing tips and prompts! This is just what I've been looking for because I have the same problem. Thanks!!
I had a feeling there were more of us struggling with this. So glad it landed! Let me know if you end up trying any of the prompts
The Forgetting Curve, as became known, was established by educational researcher Hermann Ebbinghaus in the 1880s. We forget +90% what we read/hear within 24 hours unless it's reinforced somehow. That research was repeated by Scandanvian researchers (Finland? can't remember) in the 2010s and found to be valid. Doesn't that have powerful implications for our educational system?
Which gives me something to write about tomorrow, so thanks. :-)
Looking forward to reading the post and seeing what angle you will approach.
I actually had a section with stats on why we forget what we read & some other things, but cut it last minute.
And you’re right, the implications for education are huge. We’ve got way more tools now to teach, learn, and actually retain. The challenge is shifting the system out of its pre-AI mindset and into where we are now.
In the U.S., our educational system often confuses activity with accomplishment. We continue to emphasize standardized testing (cognitive skills) and almost completely ignore emotions, the primary driver of behavior. Meanwhile, the world is burning down. Is AI useful? Absolutely! Will it turn the tide in education? No.
Because only we can save us from ourselves.
Same story in developing countries, and I’d bet it’s even worse. Emotion’s always been sidelined, right alongside the life skills people actually need to navigate the world.
Great ideas to retain information better. I recently started using one of those AI Whisper (speech to text) apps that is smart. Seems like a great use for this when reading. I don’t always want to type and I could use these principles to do the same things but through voice
Love that idea, Zach! Would work amazing to use it when teaching back concepts out loud like the Feynman prompt, without typing a word.
Zach, I run a Linux laptop and would *love* to use Whisper, but it seems hard to install on Linux. How do you do it? DM me if you'd like. And thank you in advance.
The one I’m using is only on Mac windows and iOS. I would have to google to see if any good ones are available for Linux
I just asked ChatGPT and I found a cloud-based site that transcribes a recording. Are you doing it live, or via recording?
You can use basic whisper integrations but the ones I like are MacWhisper (free tier) and wispr flow. I find the iOS app is surprisingly good: https://wisprflow.ai/r/ZACH98
It has one more step above whisper which is an AI model corrects your language before it pastes. That’s my referral code from inside the app. I don’t know them at all or have any affiliation.
I wish there was a free version as good, because I do not like that is is transcribed in the cloud. I would prefer a local one.
That was just... Fantastic! Just saved it all to my Evernote and will use it finishing my next book reading! :)
A true compliment when it makes it into the saved collection, thank you William! Can't wait to hear how it works out for your next book.
This is really turning reading a book with more scientific approach. It opens up to new possibilities and now I’m beaming with ideas haha
1. I can upload the book pdf to notebookLM, extract each chapter and ask Gemini to generate quiz for me using canvas mode.
2. I usually save some quotes on my kindle and export it to google sheets. I’m thinking to ask AI why I’m so resonated with these quotes. What does it tell me? What can I do with it?
Gonna be a busy weekend soon :)
Love that, and these ideas!!
I actually played with NotebookLM and uploading the book too, but saved that for another time or maybe a follow-up post :))
The second one sounds really great. Would love to hear what patterns show up when you run that, keep me posted. Been obsessed with reading since forever, so I'm always on the hunt for better, deeper ways to make it stick.
Will start with one prompt😉
You go, Jude! Curious which one you pick, let me know how it works for you.
How do you usually approach reading non-fiction right now?
Will update you. Reading N/F? Just like how you said…notes, scribbles, highlighting, underlining….although it never helped in retaining information. But when I reread those same books later, it leaves a mark on what and how my thought process was at the time. 🙂
I know exactly what you mean, and every time you re-read a book, you’re left with new insights and notice different connections. It really is about the time you’re at, you’re right.
This is a fantastic way to encourage reading!
To be honest, I’d feel a bit daunted by the “summarize & teach back” approach right from the start. I’ve never been a strong reader myself, finishing a non-fiction book (or even fiction) has always been a challenge.
I tend to learn more through action and referencing back as needed. But your guide really opened up new possibilities.
Curious, what platform or tools are you using to put these exercises into practice?
And how much time does it usually take you to get through a more sophisticated book?
My sense is that while it might not save time, the depth of learning would be much greater.
You’re totally right, this approach doesn’t save time at all 😂 If anything, it slows reading way down.
But for the right books, I actually prefer that. I used to spend a lot of time trying to make ideas stick anyway, so this feels like a more joyful and engaging version of that, as I can bounce off ideas with GPT.
As for platform, I’ve mostly been using GPT to talk through the ideas (any LLM should work anyway). Also NotebookLM is great if your notes or book are already digital.
And one more sophisticated book, as you say, was Antifragile by Taleb. Took me a while because I was actively applying it to my startup as I went. Didn’t time myself tho.
I can totally see how, for the right books, slowing down is worth it, you really want to internalize the ideas. Deep respect for the rigor you’re putting into it!
Antifragile is such a classic. So brilliant that you’re applying it in real time to your startup, that’s the best kind of learning.
That’s such an interesting observation, it's probably because fiction naturally pulls you into a world and makes you feel something, so the ideas land deeper.
I haven’t read fiction in a while though.
Glad the tips are helping!