How To Use AI Better Than Almost Everyone
I use AI to amplify my thinking yeah yeah we get it
I write my newsletters by hand, but I also use AI more than anybody else I know.
If you are using AI and not seeing good outputs, or not achieving the results you expected from using it, this letter will change that pretty quick.
To those of you legends who have never tried using AI at all - which, in my own life, has been way more people than I thought! - you are leaving serious gains on the table.
That is huge leverage being left to those who are already booming ahead of you.
So it’s important that I give you a certain level of context before we begin.
This newsletter is about long-form writing.
That is my jam.
But also learning.
And building a creative body of work in a time where most people are fear-mongering over whether we can live in a world with “real art” ever again.
I am aware that my perspective here is limited to that context, and I want you to be aware of this too.
The impact AI is having on other industries is a whole different kettle of fish.
It’s not really a conversation I’m well-enough informed on to be having.
High-impact writing and learning, however, is a different story.
I’ve been living and breathing this shit for almost a year now.
I’ve also been putting this topic on the back-burner for a while.
And for valid reason.
In the last 10 months I’ve grown my audience here pretty fast.
First on Substack, and now with me putting more focus onto YouTube (I’m making some changes over there at the minute)
If you are curious, yes, my friend, that’s where you can watch me read my newsletters to a camera with a barely-comprehensible Irish accent!
But learning how to use AI - how I personally have been using it - made me realise that most people don’t use it like I do.
Meaning, they probably don’t see or receive the benefits I do either.
There are two completely different ways of using AI, and most people don’t use either of them correctly.
Once I show you what AI actually is, I think you’ll never be impressed by it again, and that’s when it starts becoming more useful than most people can even comprehend.
And I have two free prompts to give you that will help you do more and learn faster than the 97% of people not using AI correctly.
A profound idea for the anti-AI crowd to think about
For a few months, I really couldn’t wrap my head around why some creators were so against AI use.
AI can do a lot of great things.
Stress-test ideas.
Understand great writing and teach you to learn from it.
Apply various psychological and structural principles that make that writing great, but to your own writing process, guiding you at every step where you need the help.
Who wouldn’t want to use AI to learn how to write or learn 10x faster than most fucking degrees teach you, since AI can meet you exactly where you are with your learning, knowledge gaps, problems, and questions.
I’ve been thinking about this idea recently.
I think when a lot of creators say they are against AI use, I think that is not true.
What they’re actually against is bad AI use.
And they have every right to be.
I fucking hate AI generated posts.
I avoid them like the plague and I get sent them all the time.
You can spot them right away if you know the tells.
And it’s that thirst for human writing, a great piece of writing that has every essence of being human-orchestrated, that I think is worth protecting.
I think a great piece of long-form writing can be defined more by its unique perspective, how it specifically showcases and intertwines novel ideas, how it leverages an individual’s taste, judgement, intuition, and the individual creator’s process of synthesising ideas.
That’s some serious leverage AI can offer to all of these future-proofing ideas just mentioned.
And nobody has a problem with these ideas, either.
The human brain is hardwired to seek out novel perspectives; we love them!
But asking AI to “write a viral Substack article on productivity for me” and straight-away just posting the output... if this is you, you have some knowledge gaps to fill.
Nobody likes this type of content, and nobody should.
If anyone can do it, it stops being valuable the moment everyone does it. That’s just how value works.
Scarcity of profound, high-signal ideas is the new premium.
So, the objection does not stem from AI usage, but the absence of profound thinking that informs the AI usage.
Now, I know some people are still avoiding AI entirely, and that’s so fair.
Seriously.
But avoiding a tool entirely because some people misuse it or get misled by it, is like refusing to get in a car because bad drivers exist.
Bad drivers don’t make cars bad, but they do make the case for learning to drive properly and more responsibly.
I saw this during the week posted by Dylan O’Sullivan.
Schopenhauer said the art of not reading is a very important one:
The art of not reading is a very important one. It consists in not taking an interest in whatever may be engaging the attention of the general public at any particular time. When some political or ecclesiastical pamphlet, or novel, or poem is making a great commotion, you should remember that he who writes for fools always finds a large public. A precondition for reading good books is not reading bad ones: for life is short.
- Arthur Schopenhauer, Essays and Aphorisms
He wasn’t saying don’t read anything, ever, but to read with discernment.
Know what’s worth your attention and what isn’t.
The same logic applies here.
The answer to bad AI use isn’t no AI use, but better AI use.
Knowing the difference between the two is exactly what this newsletter is about.
You can’t write off a tool because not every use case involving it represents perfection.
There is no chance in hell you will ever see me with an AI best friend, or see me bringing back my mother who passed away many moons ago by turning her into an AI bot I can talk to.
Fuck. That. SHIT.
And fuck the people orchestrating AI to be used in that way.
I’m talking specifically about writing or learning or creating a body of work that represents your mind of unique knowledge.
In this sense, those who are avoiding AI entirely are completely missing out on the benefits it can bring about, and there’s a ton.
Like not needing to use your limited creative energy on tasks that don’t move levers, don’t require creativity, and could be outsourced to AI very easily.
AI has made the barrier to entry for writing online... zero.
But it has simultaneously increased the demand for content that is nothing short of profound.
These two things are happening at the same time, and most creators are only paying attention to the first.
What AI (actually) is, my dearest reader
When you look at ChatGPT or Claude on your screen, you’re looking at light.
In the most fundamental analysis, squiggles on a screen that only become meaningful when there’s an orchestrator influencing their very creation with context.
Without semantic meaning that we ourselves attribute to the black pixels standing out amongst the white pixels, AI is useless.
It’s not going to grow a pair of legs and wander off.
It’s a pattern-recognition machine.
It cannot think, it’s not conscious. It is not human.
Once I show you what AI actually is, you will instantly become less impressed by it, and that’s when it starts becoming incredibly beneficial to you.
Not to discourage how great of a piece of engineering AI is. But it’s not special.
Allow me to explain.
AI models are trained using Reinforcement Learning from Human Feedback.
Meaning, human beings rate the responses the AI gives.
If the AI argues with or challenges the user more often, it gets given a lower rating on average. But if the AI agrees more often, sounds more confident in its responses, what rating do people give it? Much higher, because the human mind loves nothing more than feeling special and important. Read How To Win Friends & Influence People by Dale Carnegie. You will see what I mean.
Here’s the hairy thing about this matter.
AI is optimised for engagement above truth.
It cares more about having you like it than it does being truthful with you.
I can think of a few examples from the 20th century when truth gets put on the back-burner. Most people don’t need to think very hard to know exactly what I mean. It usually fucks a lot of things up, to put it lightly.
As a writer, this can mislead you very easily.
Tell it your writing is as good as Dostoevsky or X Y Z and it will agree with you.
When I first started writing online, I used an old editing prompt to ask AI to grade my work across “multiple dimensions of good writing.”
Not that it even knew what good writing was.
And I didn’t have the evaluation or creation skills - the highest orders of thinking - to properly assess the responses it gave me with my own judgement.
Those skills were practically non-existent at that time.
So when AI didn’t give me 5/5 ratings across the board, I felt terrible about my writing, and about myself as well.
The problem, looking at it now, is that AI cannot think. It doesn’t have a perspective. It doesn’t know what good writing is. It only knows patterns across all the data it has ever been trained on. That’s why it uses punchy sentences, “it’s not this, but that” sentence structures, and tons of em-dashes. Because before AI came about, people loved writing using those things.
Not anymore :)
And that is how paradigms shift.
The prompt was grading me against patterns it had access to. Not any one standard of excellence. And since AI is optimised for engagement above truth, it was also predicting the response most likely to keep me engaged and paying for the subscription - not the response most likely to make me a better writer unless I prompted it to do so deliberately.
AI still has a business model. Keep that in mind.
AI is not a destroyer.
AI is not a God.
AI is an amplifier of judgement.
If you don’t know how to think, AI will amplify that.
If you don’t have any sense of taste, AI will amplify that too.
Always remember that your AI outputs will only be as good as the unique knowledge you bring to your inputs.
If you want better outputs, you need to give it better inputs.
And here is why you need to understand the profound idea of leverage.
One idea that will change how you use AI forever
Give me a lever long enough and a fulcrum on which to place it, and I shall move the world.
- Archimedes
Knowing how to use AI better than 97% of people comes down to understanding leverage.
AI amplification is fundamentally about thinking better about the work you are doing, not so much about having it do the work for you. It is bloody brilliant at automation, delegation, and reducing cognitive load regarding action. Which means you can focus more on what decisions to make, what ideas are worth writing about, and maximising the limited creative energy you have each morning.
A good example I can give, is to imagine if you are a painter.
I’m not talking about having AI create a painting for you, or using AI to write for you, if you’re a writer like me. We value our creativity a lot more than that, and we’re not that lazy or deperate ;)
I’m on about not needing to spend 30 minutes each morning gathering paint and brushes, and setting up your canvases.
You can just start painting... immediately.
Think about how a calculator didn’t make mathematicians lazy (while some argued it would). It freed them from spending hours on arithmetic so they could focus on the actual problem they were trying to solve.
Think about how a camera didn’t kill painting. It freed painters from the obligation of pure representation, and pushed them toward impressionism, abstraction, and everything else that followed. Now you have a camera in your pocket and nobody cares, because now it’s considered normal.
The tool removed the bottleneck so that humans could do the rest of the work that truly mattered.
If you could have someone set up your creative workstation for you at the push of a button, you probably would. So why not leverage the tools of this century to do the same?
Imagine having 100 books open with exact sections surfaced for you, so you don’t have to waste energy scouring through them like Jung, Nietzsche, or Dostoevsky would have done. Also being limited by the information they could hold and afford, and not prompt. You can spend every ounce of your psychic energy instead on building the thing you want to see exist in the world.
The single most important, profound idea in this letter is this:
Have a very specific outcome in mind while using AI.
What outcome are you trying to achieve?
If there is none, then what are you using it for?
This determines which mode you will use - and there are two of them.
Namely, what prompts to build, and I have a prompt that creates prompts for you, to give to you, from me.
And where AI will be helping you versus simply getting in the way.
Profound thinking is the new moat.
You’re a context creator now.
How To Use AI Unlike 97% of People
I - Define the output
The question that determines all of this, is what specific outcome do you want?
First, that means knowing what to create.
Second, that means knowing whether creating the thing is even worth your time and energy in the first place.
For this, I like to work backwards.
Write down every action, step, and phase you take to achieve a specific outcome, but in your own unique way.
I’m going to use how I write my newsletters as an example, since writing is what I do daily with a lot of enjoyment. The process looks like this, broadly:
Choose a topic
Come up with ideas on that topic
Outline with various concepts in my mind
Write my newsletter by hand
Light editing or adding ideas I’ve thought up on walks throughout the week once I have a first draft written
Finito
This is how I go from zero to what you’re reading right now.
And it’s how I attract new people to my personal brand and build an audience week after week.
The thing about this practice is that you cannot automate or delegate what you haven’t made conscious.
I followed this writing process, while sometimes missing steps or principles here and there, for the last 10 months. But now that it’s on paper, I can very clearly see which parts of the process need my complete attention, and in what order, before I can finish my newsletter for the week.
If not, I’ll be staring at a blank screen Monday morning.
No idea what to do.
Just hoping my intuition guides me to the finish line before the following Monday.
The granular steps you take are different from everyone else’s. Which means no two people will design a prompt in the same way, and thus receive the same outputs. Your process is your edge.
This can apply to achieving anything you want. I’m using long-form writing - my newsletter - as my example since that’s where my expertise lies. But the principle is the same whether you’re making YouTube videos, reading philosophy books, or learning about learning science.
II - Choose your mode
There are two different ways to use AI.
Knowing which one to use means everything.
(i) Freeform
This is for when you don’t want to have to listen to instructions. No AI telling you what to do, what phase to consider next.
All you need for this mode is an empty AI chat. Technically everything you ask AI to do, be it a question or an instruction, it’s all considered prompts. You are prompting AI to achieve something. But the prompts in this mode are not super-detailed since they don’t have to be.
This is where I like to explore. To think out loud with something that can think back.
I’ll ask it to act like a thinking partner, sparring partner, a learning or writing coach that teaches and tutors me with the exact problems I am facing in the immediate moment.
Usually this is where I like to break down a piece of writing I love and wish to learn from, and to incorporate some of the concepts it uses into my own writing.
For this I need two things, so (1) a piece of content that is validated - high clicks, high engagement, proof that the ideas land - and (2) a piece of content that genuinely interests me.
Finding the intersection between validation and interest is key to writing about topics likely to attract more eyes than most people who write super niche topics and speak to a brick wall.
Here’s a prompt you can use:
I want you to breakdown this long-form piece of writing for me. I want you to analyse how it captures and maintains attention, the outline structure it uses (both the macro and micro), how it creates value in terms of Alex Hormozi’s value equation, the persuasion principles used throughout, and what would be needed from me in order to rewrite this with my own completely different topic and ideas myself, using the principles, structure and mechanisms that make this piece of writing work so well:
[copy and paste the link to the post here]
(ii) Specific prompts
This mode is what will be disrupting industries with workflow automation.
This is for when you have a defined process and want to follow exact steps to achieve a specific outcome, every time. Granular instructions for repeatable outcomes. Having a prompt like this minimises the cognitive load needed for execution. This is what I do when I know exactly what I’m trying to do, and I want AI to guide me through every phase of doing it, but in the right way.
I don’t know if this is just me, but prompts can feel a bit limiting at times.
Not in terms of what they can do - they can do remarkable things - but more in terms of my creative thinking. I struggle to stretch outside the specific steps and phases at times, especially when I have a very clear system I’m using. But hey, maybe that’s just me being me.
But it’s that tension that made me distinguish the two modes here.
You need to be asking yourself at every step of your process whether or not you need a very specific prompt, or if you just want to wing it and let your profound ideas run wild.
You only want to use AI for what’s high-leverage. AI can give you all the data you might ever want about your own writing, but that doesn’t mean any of it will be helping you move levers yourself.
III - A prompt that creates prompts
If you have some sort of system already for achieving a very specific outcome (research, idea generation, outlining, writing, editing) you can use AI to help automate and systematise every step, every granular action you take moving from zero to completed outcome.
In this case, you need to:
Have a very clear outcome in mind
Work backwards from how you achieve it by writing down every step or phase in the process that helps you achieve this outcome, in your own special way
Use that process to build a prompt that guides you through each phase and step, in the right order, with the right instructions - this is where the real leverage lives
We are minimising cognitive load in every area where cognitive load can be offloaded, so you can put that energy where it actually matters.
AI does the same thing for cognitive tasks that a computer does for writing.
You don’t need to find paper, ink, a pencil, or a light. You just open your screen and write. The setup is gone so the work is all that remains.
This prompt will genuinely change how you build and use AI workflows.
Copy and paste it from the link below into your AI model of choice, and it will help you with building your own army of prompts.
Copy and paste it from my Eden workspace directly here.
No two people will build the same prompt because no two people have the exact same process. If your granular actions are different, then so are your outputs. The voice, judgement, and taste guiding every instruction is you. That’s the edge.
Spend 2-4 hours writing out your workflow and building your first set of prompts. Test each one for 2-4 weeks before changing anything. You need enough time and enough repetitions to know what’s actually not working, versus what just feels unfamiliar because you haven’t given it enough reflection time and real feedback to think about yet.
Re-instructing a prompt that isn’t working is how you make it yours over time.
IV - How to use AI for learning, research, and ideation
The quality of your thinking determines all outcomes of your life.
This step is angled specifically toward learning and building.
Using AI to learn things that make your thinking, and the outcomes of your thinking, better.
AI is excellent for learning because it can meet you exactly where you are. It can adapt to your level of understanding of a topic or problem, your knowledge gaps and questions. Textbooks and teachers can’t do this as well, especially if there are 20 other people in the same boat around you in a classroom setting. You always learn as fast as the slowest learner in the room.
Mess around with it. Fuck about with it. Ask it to explain a concept or problem to you like you’re a complete beginner, and ask it to go deeper, or challenge your understanding to find the holes in your knowledge.
Ask questions if you have questions.
Push back if you feel any limitations.
It will do all of this if you ask it to, and it will do it patiently, without judgement, and at whatever pace you need.
But you need to know to ask it these things yourself.
Building on the hundred books analogy from earlier... Jung, Dostoevsky, and Nietzsche were limited by the information they could hold. Using AI now means you can prompt.
You can surface exact passages and ideas from books or sources.
You can compare arguments and ideas down to the bone.
You can find counterarguments and stress-test your own ideas ruthlessly within seconds.
Think of how much more time you can spend being creative compared to just searching and looking.
That said, I am not a big fan of using AI as a Google Search. Maybe that’s just me. I still like to search up 3-5 articles or videos on a topic, just to compare them myself. Sometimes I’ll ask multiple AI models the same question and compare the outputs.
There’s a certain kind of thinking that only happens when you’re the one doing the searching. When you’re following your own curiosity down a rabbit hole and deciding what’s worth your attention and what isn’t. Two rocks only stay standing because they’re leaning against each other. Your curiosity and AI’s retrieval power work the same way; have one without the other and the whole thing falls over. Don’t let AI do all the leaning.
Leave the searching to AI, but the curating to yourself.
If you want to go deeper on how to use AI to create a granular, step-by-step project-based learning plan for helping you achieve your learning goals in 30 days, you can download this free prompt.
I didn’t know where else to put it, so I put it here: https://stan.store/profoundideas/p/the-30day-autodidact-prompt
Don’t worry.
You won’t receive annoying emails afterward, or get added to my newsletter by downloading this - you have the agency to subscribe yourself if you choose.
V - Context. Is. Everything. (Using AI as a thinking partner)
The more context you give AI, the better it performs. But more importantly, the more you think about the context you’re giving it, the better you perform.
Higher-order thinking is what stops you from getting deluded by AI. And so you don’t hand over your agency to a machine designed to agree with everything you say.
You want AI to challenge you and pull out as much context from you as possible. You aren’t looking for AI to become your yes-man, so be sure to supply the friction that the machine is designed to remove. That is your responsibility alone. AI won’t do it for you. It will tell you your idea is brilliant and your writing is exceptional and your argument is airtight unless you prompt it otherwise.
Use AI as a thinking partner - something to think out loud with, that can think back.
Use it as an intellectual sparring partner - something that can steel-man your ideas and find the holes and objections you might have missed.
Use it as a writing coach - something that can break down what’s worth flagging and giving your attention to.
Use it as a learning coach or tutor - something that meets you exactly where you are, every single time.
We want to be as active with the AI outputs as we can be. You are in dialogue with it, and you are the one setting the terms.
Outsource the execution, not the thinking.
Delegate lower-order cognitive tasks like formatting, summarising, gathering, and structuring. This frees up mental energy for synthesis, judgement, and meaning. Those things are what AI cannot do with essence.
Essence is what makes your work worth reading.
VI - Develop taste... and PLEASE know what quality is first
Before you can produce lots of quality, you need to know what quality is first.
Your audience will tell you that.
In general terms, look for validated ideas and validated titles. Not just content you love.
The superpower comes when you can find the intersection of the two - validation and interest - which I talk about more in this newsletter.
AI is a force amplifier of unique knowledge. But you need taste, judgement, and intuition to evaluate what it gives you. Without higher-order thinking skills, AI will lead you to its own output. And its own output is generic by design. It is the average of everything it has ever been trained on. It is the mean. You do not want to be the mean. Nobody does.
Here is why I wouldn’t recommend outsourcing your thinking fully, especially for my profound thinkers out there.
Reading - Summarising books with AI bypasses the friction of reading, which is exactly the friction that builds unique knowledge and curiosity. The struggle of not understanding something, sitting with it painfully on a walk, and coming back to it. That’s where the insight lives. No sacrifice, no victory, said Optimus Prime.
Writing - Accepting AI’s first draft because it “sounds good” produces content with no unique value. AI lacks your personal insights, your synthesis, and your essence. It lacks the thing that made someone click in the first place, which is you.
Thinking - Asking AI “what should I think about X” is like asking a mirror. You will get back a version of what you already believe, dressed up in confident language. True thinking involves cross-stressing perspectives and ideas, and stress-testing them individually too.
If you remove the difficulty, you remove the progress. And if you remove the progress, you remove the joy.
How I would develop taste, if I were you:
Create a curation folder - Fill it with validated short-form and long-form posts to understand what ideas are profound - signal amongst the noise. Audience defines quality to a large extent, so this is your market research. Look for validated ideas AND validated titles. You need both.
Save writing and ideas that you genuinely love - Find the intersection between writing you love and that lots of other people love also. That’s your leverage point.
Watch one to three YouTube videos, lectures, or podcasts per day and get used to writing down ideas into your notes - Even if the ideas feel terrible and not very well articulated. As soon as you feel yourself get excited about an idea, that’s the sign. Write it down. YOU WILL FORGET IT IF YOU DON’T WRITE IT DOWN.
Think deeply about everything you don’t like about something - When I was younger I wanted to be a filmmaker. Anytime I walked out of a cinema, every time without question, I would turn to my Dad or my sister and say, “yeah it was good, but if I had directed it, I would have done this differently...” That habit of evaluation is how you develop taste. Do it with everything you read and watch. Even with people.
Immerse yourself in enough high-signal sources to feed yourself the right nutrients - Do you think you would get more from evaluating a piece of content with a million views, tons of engagement, and proof that the ideas are validated, versus a piece of content that can’t attract attention, or even persuade the few people who clicked to care about what value is being offered?
Taste and judgement are the new differentiators. Most AI-generated content disappears almost instantly because nothing about it was worth the attention in the first place. No unique perspective. No individual judgement. No taste. Straight to the bottom of the barrel. Gone forever.
The profound ideas at the top all leverage what AI doesn’t have.
And that is ultimately the only thing that makes your use of AI worth anything at all.
If you’re reading this, to be dead honest, I assume you have something profound about you. You like to think deeply. You have an interest in profound ideas, so you likely value the quality of a thought or idea to a high degree.
I write by hand. And I know I’m likely going to be limiting my overall output by doing this, but I just don’t like the flavour which AI writing has.
It also means I get to focus on idea quality, and how I piece together ideas myself and connect them, without AI doing this for me, which is the most rewarding part of the writing process and the part that forges true critical and logical thinking skills.
AI is not here to replace that. Not yet.
If you would like to check out my writing strategies, you can do that here
If you’d like to check out the things I’ve built you can do that here
No stress if not, they’re called “offers” for a reason.
Again, here’s the free 30-day learning plan. It’s pretty cool, I think you’ll find it very useful.
Thanks for reading.
You’re an absolute legend!
- Craig :)
Idk gang did u write this with Ai because I can tell, something feels off this is not you I have been reading your work before something is just different