AI Isn't Making You Stupid— You're Just Learning Wrong
An exploration into Gemini's Guided Learning
Welcome to The School of Knowledge and this week's free essay. Every Sunday, I share insights to help you understand business and life— not through abstract theory, but through the hard-won wisdom of practitioners with skin in the game. People who've actually tried, failed, learned, and succeeded.
If you want deeper engagement paid members get implementation support, live Q&As, and access to our private chat.
Recent research suggests the phenomenon of 'cognitive offloading' via extensive reliance on outsourcing your thinking to AI is detrimental to deep and fluent understanding. Naturally, this leads to questions such as: “Should we stop using AI to learn?” Maybe. But, in order to answer that question, it's important to understand what the 'best way to learn' is.
Decades of research from cognitive and neuroscience fields has demonstrated that the best ways to learn are the following:
Spaced repetition: Spreading study sessions out over time (today, tomorrow, seven days, four weeks)
Retrieval practice: Actively recalling information from memory (self-quizzing)
Interleaved learning: Mix the study of different but related topics (think Charlie Munger's infamous 25 human tendencies)
Elaboration: Connecting new information to existing knowledge (asking "how," "why" and "what if")
Notice how reading, re-reading or taking highlights aren't on the list? Reading often makes you feel like you're learning— but without integrating any of the above into your learning routine, it'll most likely be fleeting. Especially if you read a lot of material.
So, is the real problem that AI is bad for learning— or is it that you're not learning correctly in the first place? AI is a mirror. If you give it the right principles to begin with, it will act accordingly—and that goes for giving it poor instructions (or none) to begin with.
AI has, and still is, mostly used as an amplified search tool. You ask it a question and not only do you get an answer— you now get a comprehensive report, citations, content creation and a whole host of other DFY (done for you) tools. It's truly mind-blowing. That's great if your sole purpose is to produce AI-generated content for Substackers to deplore— but it's not so great for learning, and the big AI corporations are waking up to this.
The big three (OpenAI, Anthropic and Google) want their AI woven into our educational institutions. This is great for the democratisation of knowledge. Kids from all ages and backgrounds should have access to resources if they help them gain a better education, just like smart devices have been introduced into schools in the past. But AI, in its current state, isn't fit for purpose. It's making people lazy thinkers.
Or is it?
Introducing Gemini's "Guided Learning"
Gemini have introduced to their dashboard "guided learning," and in service of this week's essay, I decided to see how good of a teacher it actually was. For this exercise, I had three requirements when selecting the topic I wanted Gemini to teach me:
It has to be something I am interested in
I have to loosely understand the topic
It has to be something I could elaborate on
Something I've been thinking a lot about recently is capital allocation— potentially, the single most important task of any CEO, founder or business owner. For those who don't know what it is, it's how you spend (invest) the money the business earns. Do you invest in people, equipment, or stock? Or, do you build another factory, acquire another company or stick your cash in a high-interest account? All of these decisions are masked in uncertainty.
How would Gemini help gain clarity?
*For brevity, here is the dialogue between myself and Gemini:
Me: Hi Gemini, give me a few concepts that a business owner should know regarding capital allocation. I will then pick one to learn.
Gemini: Of course. Capital allocation is one of the most important jobs for any business owner. It's the process of deciding how to best use the company's financial resources to create the most value for the long term. To get started, we could explore a few different, but related, key concepts. Would you rather dive into:
Opportunity Cost: The fundamental concept that for every investment you make, you are giving up the potential returns from other investments.
Return on Invested Capital (ROIC): A powerful metric that helps you measure how efficiently you are using your money to generate profits.
Capital Budgeting Basics: The actual process and methods, like Net Present Value (NPV), that businesses use to evaluate and compare potential projects. Which of these sounds like the best place to start for you?
Me: Opportunity cost
Gemini: Excellent choice. Opportunity cost is a powerful concept because it reveals the true cost of any decision, which is not just the money you spend, but also the opportunities you lose. In simple terms, opportunity cost is the value of the next-best alternative that you give up when you make a choice. It's the "opportunity lost." Let's start with a very simple, non-business example to make sure we've got the basic idea down. Imagine you have a free ticket to see either a concert 🎤 or a movie 🎬. Both are happening at the same time and you can't go to both. If you choose to go to the concert, what is the opportunity cost?
Me: The Movie
Gemini: You've got it. The opportunity cost is the experience of watching the movie 🎬. It's the value of the next-best alternative you gave up.
Applying Opportunity Cost to Business. Now, let's use this same logic in a scenario you might face as a business owner. Imagine you have £20,000 of extra cash in the business. After researching, you've narrowed its use down to two main options:
New Equipment: You could buy a new machine that will make your production more efficient. You calculate this will save you £25,000 in costs over the next year.
Inventory Purchase: You could use the cash to buy a large amount of popular inventory at a discount. You expect to sell this inventory for a total profit of £30,000. Assuming you can only choose one, if you decide to buy the new equipment, what is the opportunity cost of that decision?
Me: £5,000
Gemini: That's a very common and logical calculation, but it actually represents something different. The opportunity cost isn't the difference in value, but the total value of the single best alternative you gave up. In this case, by choosing the equipment, you gave up the chance to buy the inventory.
The value of the choice you made (new equipment) is £25,000 in savings.
The value of the choice you gave up (inventory) is £30,000 in profit.
So, the opportunity cost of buying the equipment is the £30,000 profit you forfeited from the inventory deal. The £5,000 you calculated (£30,000 - £25,000) is also a useful number. It tells you the net gain you would have had by choosing the inventory instead. Thinking this way helps you compare options and see which one is financially better. However, the concept of opportunity cost forces you to state clearly, "To get this £25,000 benefit, I am giving up a £30,000 benefit." Framing it this way makes the trade-off stark and clear. To summarise:
Opportunity Cost: The full value of the foregone option (£30,000).
Net Gain/Loss: The difference between the two options (a £5,000 net loss in this case, by choosing the equipment).
This concept helps you evaluate not just whether an investment is good, but whether it's the best use of your capital compared to other available options. Does that distinction between the opportunity cost and the net gain make sense?
Me: Yes, I misread the question. I'll be more careful.
Gemini: No problem at all! It's a subtle distinction, so it's a very easy mistake to make. The important thing is that the concept is clear now. The core takeaway is that opportunity cost forces you to think like a true capital allocator. You're not just asking, "Is this a good use of my money?" but rather, "Is this the absolute best use of my money compared to all my other options?"
The conversation continued with the other two topics before Gemini suggested wrapping things up with a quick summary. But things didn't stop there. Gemini summarised our clunky conversation into a streamlined Word document, produced an infographic with an interactive API, created a 10-question quiz, 25 flashcards and finally a study guide.
The sharp reader will notice that these options can create genuine knowledge because they fall in line with the latest research concerning best learning practices. Let's go through them again and see how Gemini's Guided Learning function stacks up against the research:
Spaced Repetition
Gemini suggested ending the session after we covered the three topics, but I could have continued indefinitely. This revealed an interesting design choice: the system currently relies on user discipline rather than enforcing optimal learning patterns.
Currently, the user would have to be disciplined and treat Gemini's Guided Learning function like a human teacher and 'pencil in' classes with it. Having a cut-off or time cap per day with the Guided Learning function might seem restrictive but would help prevent cramming (a poor method for long-term knowledge retention) and create genuine spaced repetition.
Active Recall
Gemini created a 10-question quiz that I went through the following morning, scoring 9/10, showing the three concepts at least stuck over night. There was also the option for more questions, which I could strategically use to generate a quiz in, say, a week's time and then a month's time. Spaced repetition by design.
It then analysed my performance with 'highlights' for areas I performed well on and a 'focus areas' section where improvement could be made. The learning didn't stop there as 25 flashcards were created with a crucial twist: Gemini didn’t provide me with the answers—meaning I was genuinely forced to make those neural connections that are so critical for building lasting knowledge.
Interleaved Learning
Throughout our conversation, Gemini kept referring back to what we had previously learnt, with all three concepts falling under capital allocation's umbrella. But here's what made it genuinely effective: each concept demanded different thinking approaches. Opportunity cost required abstract reasoning about trade-offs, whilst ROIC and NPV both demanded mathematical calculations to reach meaningful conclusions.
This variety wasn't accidental. When you're forced to switch between conceptual thinking and quantitative analysis—even within the same broad topic—your brain has to actively decide which methodology fits each problem. That cognitive flexibility is precisely what makes interleaved learning so powerful.
Elaboration
Gemini wanted to understand how the concepts were relatable to me and asked for an example of how NPV could be used in my business. We choose a real scenario I was facing: buying versus leasing a van.
Once we'd gone through the exercise, it looked like leasing was the cheaper option but here’s where the elaboration principle kicked in: We hadn't considered the resale value of the purchased van after the 3-year term.
Once I queried this, Gemini took me through the calculation again, but with me selling the van at the end of the 3 years. This missed option now meant that buying the van was the better option of the two. A perfect demonstration of how deeper questioning can reveal hidden variables that completely change the outcome.
Final Thoughts
This week's essay on AI learning was primarily supposed to be focused on ChatGPT's 'Study and Learn' function, but with the introduction of ChatGPT 5 last week, it kept breaking when I was using it. I would be shocked (and delighted) if it's more capable than Gemini's Guided Learning.
Gemini's Guided Learning function has only been available for a few months and was built using the knowledge of industry experts from the cognitive, neuroscience and pedagogy worlds. And it shows. How this stacks up long-term against non-AI learning is, of course, unknown, but if AI is already implementing best practices from the above fields, then that seems like a good place to start. Imagine where it could be in 5 years' time.
Moving forward, this is what I believe to be the best practice for using Gemini's Guided Learning function when learning something new:
Create 'Gems' on the topics you want to learn about
Upon creating a new topical Gem (say capital allocation), attempt to explain your understanding of the topic before it begins
Listen to Gemini's proposals for when to stop and take a break
Use Gemini Pro 2.5 as it's slower, allowing you time to think between questions and answers
Learn multiple topics simultaneously for genuine interleaved learning (capital allocation & basic accounting, for example)
Take quizzes at regular intervals— ideally before you begin the next session
Use Gemini's flashcard feature at irregularly spaced intervals
Ask it to create study guides for filing and regular reviewing (easier to digest than going through chats)
Keep up to date with the latest literature and compare how Gemini is implementing its recommendations
Until next time, Karl (The School of Knowledge).
Whenever you’re ready
The School of Knowledge helps business owners and operators understand the world through practitioners - those who try, fail, and learn through direct experience. Real knowledge comes from having skin in the game.
Join our growing community of like-minded lifelong learners here: