16:22Alex Finn · Listicle
7 Tips That Turn ChatGPT 6 Astra Into AGI
One creator's seven-tip playbook for turning ChatGPT 6 Astra's computer-use and reasoning upgrades into real workflow gains.
September 7thThe argument in one line.
Self-education beats formal schooling when you ground an AI tool in curated sources, then layer four habits on top: source curation, learning-style-matched output, teaching it back, and daily habit-stacking.
Read if. Skip if.
READ IF YOU ARE…
- You're trying to learn a new skill on your own and keep starting courses or apps you never finish.
- You already use ChatGPT or Claude for research but want more accurate, source-grounded answers instead of generic web-trained responses.
- You learn better in one specific format (video, audio, hands-on) and want study material generated in that format automatically.
SKIP IF…
- You're looking for a pricing breakdown or full feature list of Gemini Notebook, this is a study-method framework, not a product review.
- You want a guest interview or multiple perspectives, this is a solo talking-head tutorial.
The full version, fast.
Most self-taught success stories (Zuckerberg, Tarantino, Jay-Z) share one habit: they took learning into their own hands instead of trusting formal credentialing. Google's rebranded NotebookLM, now Gemini Notebook, makes that easier because it only answers from sources you explicitly provide, ranking more accurate than Claude Projects, Perplexity, and ChatGPT, which lean more on open-web search and training data. The video lays out a four-step system: curate trusted sources instead of open-web search, generate study material (mind maps, audio overviews, study guides, flashcards) matched to your VARK learning style, teach the material back to cement it, and stack learning into existing daily habits like commuting or working out.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 02:18
01 · Cold open: the dropout pattern
Zuckerberg, Tarantino, and Jay-Z all skipped formal schooling and self-taught their way to the top; Weigand frames self-education with Gemini Notebook as the modern version of that path and previews his four-step framework.
02:18 – 03:50
02 · What is Gemini Notebook?
Introduces the 'Layers of AI Knowledge' pyramid (data, web, training) and ranks Gemini Notebook, Claude Projects, Perplexity, and ChatGPT by how much they lean on user-provided data versus the open web and training data.
03:50 – 07:39
03 · Method 1: Use your sources
Draws a library-vs-Wikipedia and Minecraft fog-of-war analogy for why grounding answers in chosen sources beats open-web search, then walks through creating a notebook at notebook.google.com and adding a YouTube video, a website, and a Google Drive PDF as sources.
07:39 – 11:32
04 · Method 2: Find your optimal way to learn
Introduces the VARK framework and a free skill that identifies a viewer's learning type, then tours Gemini Notebook's Studio panel: mind maps, audio overviews, study guides, and flashcards mapped to each learning style.
11:32 – 13:44
05 · Method 3: Teaching is learning
Cites Seneca and a 2013 UC study showing people who actually deliver a presentation retain material far better long-term than people who just study or prepare one, then shows Gemini Notebook's slide-deck generator.
13:44 – 16:43
06 · Method 4: Learn on the go
Uses a habit-formation study and his own powerlifting/gym routine to argue for stacking learning into existing daily habits, leans on audio-overview podcasts for hands-busy learning, and shows pulling a saved notebook into a regular Gemini chat.
16:43 – 17:28
07 · Bonus: Gemini Spark + CTA
Teases Gemini Spark as the natural next step after source-grounded learning, autonomous AI agents that run tasks in the background, and points to a companion video for setting one up.
Lines worth screenshotting.
- Grounded AI tools rank by source: your own provided data is most trustworthy, live web search is next, and a model's training data is least accurate.
- Gemini Notebook only answers from sources you explicitly add, which makes it more accurate on a specific topic than ChatGPT, Perplexity, or Claude Projects.
- The VARK framework sorts learners into four types, visual, auditory, read/write, and kinesthetic, and each favors a different study format.
- A 2013 University of California study found people who actually delivered a presentation on material retained it a week later far better than people who only studied or only prepared one.
- Short-term recall is nearly identical whether you study, prepare to teach, or actually teach, but long-term retention only holds up for the group that taught.
- Habits form fastest when the same behavior happens in the same context every day, so stacking study time onto an existing routine beats scheduling separate sessions.
- An AI notebook restricted to your chosen sources behaves like exploring a Minecraft map: you only see the terrain you've actually walked, not terrain invented beyond it.
- A saved, source-grounded notebook can be pulled into a regular AI chat app so on-the-go questions still get cited answers instead of generic training-data guesses.
- Turning source material into presenter slides and explaining it out loud forces a deeper level of understanding than passively rereading notes.
- 90% of what the video's creator learned about AI in six months came from source-grounded study inside Gemini Notebook rather than general chatbot conversations.
A four-step system for turning AI into a real teacher
WHAT TO LEARN
Grounding an AI tool in sources you chose, matching its output to how you actually learn, teaching the material back, and stacking study into existing habits beats any single AI trick on its own.
02What is Gemini Notebook?
- Grounded AI answers rank by source: your own uploaded data beats live web search, which beats a model's static training data.
- An AI tool that only answers from the sources you feed it is more accurate for a specific topic than a general chatbot pulling from the open web.
03Method 1: Use your sources
- Curate a small set of trusted sources before you start asking questions, instead of letting a general AI guess from the open internet.
- An AI tool restricted to your chosen sources can't wander into unreliable information, the same way a Minecraft map only shows terrain you've actually explored.
- Citations that point back to the exact page or timestamp in your source let you verify an answer instead of trusting it blind.
04Method 2: Find your optimal way to learn
- Match study format to your actual learning type, visual, auditory, read/write, or kinesthetic, instead of defaulting to whatever format is easiest to produce.
- Mind maps, audio-overview podcasts, study guides, and flashcards are different encodings of the same source material; pick the one that matches how you retain information.
- Knowing your dominant learning style before you build a study plan prevents wasted hours studying in a format that has never worked for you.
05Method 3: Teaching is learning
- In a 2013 UC study, people who actually delivered a presentation on material outperformed people who only studied it or only prepared one, especially a week later.
- Short-term recall barely differs between studying, preparing to teach, and teaching, but long-term retention only holds up for the group that actually taught.
- Turning notes into a slide deck and explaining it out loud, even to no one, forces a level of understanding passive review doesn't reach.
06Method 4: Learn on the go
- Habits form fastest when you repeat the same behavior in the same context daily, so anchoring learning to an existing routine beats scheduling separate study time.
- Audio-overview style summaries let you keep learning during hands-busy time, like training or chores, without needing to read or watch anything.
- Pulling a saved source notebook into your everyday AI chat gets you grounded, cited answers on the go instead of the assistant's generic memory.
Terms worth knowing.
- Gemini Notebook
- Google's rebrand of NotebookLM, an AI research tool that answers questions only from the documents, links, and files you upload rather than the open web or its training data.
- VARK
- A learning-styles framework that sorts learners into four types, visual, auditory, read/write, and kinesthetic, used to match study material to how a person actually retains information.
- Layers of AI Knowledge
- A three-tier model for how AI tools source their answers, from most to least trustworthy: data you provide, the live web, and the model's static training data.
- Gemini Spark
- Google's AI agent tool that can complete multi-step tasks autonomously in the background, referenced as the recommended next step after mastering source-grounded learning.
Things they pointed at.
16:45toolGemini Spark
Lines you could clip.
00:39
“Did you catch the pattern? Every single one of them took learning into their own hands and went from dropouts to top of their game.”
05:35
“But the problem is, what if that river ends up being lava and it just burns your house to the ground?”
11:39
“The ancient Roman philosopher Seneca even had a quote about this saying, while we teach, we learn.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphoranalogy
If you look at some of the most influential people of our generation, what do they all have in common? Mark Zuckerberg taught himself how to code at the age of 10 with nothing but books. Then he dropped out of Harvard to turn that self -taught skill into what is today Facebook and Meta.
Quentin Tarantino dropped out of school at the age of 15 and studied filmmaking by watching thousands of movies from a video store instead of going to film school. And now he's created some of the most famous movies like Pulp Fiction and Django Unchained. And then we have Jay -Z, who left high school and learned business on the streets of the New York City projects.
And he turned those hustle instincts into what is now a billion dollar music production company. Did you catch the pattern? Every single one of them took learning into their own hands and went from dropouts to top of their game.
And with Google turning Notebook LM into their new Gemini Notebook, self -education is now easier than ever. 90 % of what I've learned about AI in the past six months, I've learned this way. And in that time, I went from working at IT at a finance firm to now being an AI engineer at a multi -billion dollar digital media company.
So in this video, I'll show you my four step framework to learning almost anything with Gemini Notebook to the point where you feel almost unstoppable. Now, the problem with traditional education is that they reward you for memorization and not actual learning. I remember in high school, I got straight A's in all of my honors Spanish classes.
But when I graduated and took a trip to Mexico, I couldn't hold a single conversation. Then when you do try to teach yourself instead of just relying on school, it's just binging educational podcasts, trying out different learning apps, and starting dozens of online courses without actually using any of the things that you learn.
Admittedly, I have my own and Duolingo streaks that never actually got me anywhere. If this sounds like you too, you're not the problem.
The habits that you were unknowingly trained on are. Today, the tools to self -educate are better than any classroom. And while everyone is focused on just chat GPT and Claude and just regurgitating their responses, almost nobody is learning Gemini Notebook to actually learn.
And that is your real edge. But , it's the system that you build around it. Before we get into that system, you first need to understand what differentiates Gemini Notebook from all the other AI tools out there.
We're going to start by going over Gemini Notebook versus some of its other competitors for learning. And we're going to map them onto the layers of AI knowledge. And what I mean by those layers is that data is the data that you provide it and you want it to actually be able to tell you answers from.
These are the true sources of information that you want to leverage. Then you have the web, which is performing something similar to a Google search to get an answer for you. This is great for more recent information.
And then lastly, you have the training data. that LLM or large language model was trained off of. This can be a few years old, so this tends to be the least accurate type of information.
So when we look at Gemini Notebook, this is the most accurate because it only leverages the bottom. base layer of that pyramid of the data sources that you provided. Cloud Projects is intermediate because it also leverages that data that you provided, but it also starts scraping the web and it uses its training data as well.
Perplexity is a bit less accurate. It does have a strong emphasis on using web information, but it doesn't prioritize the data that you give it. And then lastly, we have ChatGPT, which is the least accurate.
because it's not using the data super well. And while it does look at the web, it's not as good as the other options at using that web information and mostly just relies on that training data. Like I mentioned, this is a four -step framework to learning anything with Gemini Notebook.
And the first of those steps is knowing how to use your sources. This might date me a bit, but as I was growing up in school, whenever we had to do a research report, we had to go to the library to gather all the information from books. And they always told us, don't use Wikipedia, because they said that anybody can make up anything on the internet and put it on there.
which could really mess up your research findings. The problem is now almost everything you want to learn is on the internet. I mean, if you just look on YouTube alone and try to find out how to prompt, there's so many good videos explaining exactly how to do this.
Many of them coming from the source themselves. Like this one is from Google on how to prompt. So how do we filter out the noise on the internet of all the junk information, like how my teachers thought of Wikipedia?
versus the gold mines that are found in YouTube and other web sources. If you've ever played Minecraft, then you'll know exactly how Gemini Notebook solves this problem. When you first spawn into a world, you're given a map of the world, but you only know the areas that you've seen and not anything that's outside that area.
So logically, as you're planning out your base, you might decide, okay, this open area right here is the perfect place to put my home. And then you need a farm to make food and... Putting that right next to the river is the perfect place that the crops can grow and I can harvest them.
You're only using the information about the area and the sources that you actually have. This is what Gemini Notebook does. If you look at any of the competitors, like looking at ChatGPT, for example, if you ask where to set up your home, they might say that you should put it over here because they think that the river is going to continue down this way.
It's going outside of the known area and venturing into the unknown. which in this case would represent other information from the internet or from the training data of the model itself. But the problem is, what if that river ends up being lava and it just ?
That's why we want to only work with the information that we know and we trust to give us results that are accurate. To do this in Gemini Notebook, go to notebook .google .com. And from here, you can create a new notebook if you haven't made one in the past.
Once you're in your notebook, you can add in any sources that you want. So you go on the left side here and you click add source. Since I'm learning how to prompt, I only want to use specific sources that I've found to be useful about prompting.
So I found this video on YouTube that summarizes nine hour AI prompting course in 20 minutes. So I copied the link, went back to Gemini notebook, pasted it in here. And since it links to YouTube videos, I just clicked.
Enter. And now you can see the course is over here on the left. And it's going to use that as knowledge as I ask it questions.
Then I found this website from Google that explains how to become an AI professional. So again, I copied the URL, went back to my notebook, pasted it in, and pressed enter. And now we had both the YouTube video.
and the website as sources lastly i found this entire pdf book about prompt engineering so i added it to my google drive now in gemini notebook you can add from your google drive here and i just selected the prompt engineering pdf and now we have these three sources as part of my notebook And Gemini will only use that information to answer questions.
So I know the information is accurate based on the sources that I provided. So when I ask it the difference between few shot and chain of thought prompting, like I'm a beginner, it will look through the PDF and the YouTube video and the website and give me an accurate answer citing the source of where it got it from. So this is from the prompt engineering book.
And it brought me to the exact place in that book that it's using. Pretty awesome. Now that we know how to use our sources properly, let's move on to number two, which is finding your optimal way to learn.
There is a framework called VARK, which explains that there are four types of learners. And it's key that you know exactly which type of learner you are so that you're not... putting in all this time and effort and getting back less results.
The four types of learners are visual, in that you like to see diagrams, flowcharts, pictures, and symbols. Auditory, which means that you learn best by listening. Reading and writing, which is as you read information off of a page or you write it down, that's how you retain and learn the best.
And then there's kinesthetic, which is hands -on learning by doing the actions. Growing up, I always thought I was a visual learner because I would see diagrams and just understand what was going on a little bit better. Even the first career that I went into in cloud engineering, these are the kind of diagrams that we would share for system architectures.
And it just made sense to me. But I didn't realize that I was... actually missing out on so much learning once I discovered what my actual learning type was.
I created a skill that helps you learn exactly this, and you can download it totally for free with the link in the description below. Once you download that file, all you have to do is go into Gemini Spark, or if you use Claude, you can do it in Claude as well, and click this upload button. Then you just drag in that skill file and click open.
That will add in this learning style finder skill. If you don't have access to Gemini Spark, you can just copy the instructions from the skill I provided and just paste it in a normal Gemini chat. But if you do have the skill, once it's downloaded, all you do is type slash learning style finder and click go.
This will bring you through a series of six questions to find exactly which of the four types of learners that you are. And after going through this, it became very clear to me that I'm actually a kinesthetic learner. So for me, just reading and listening to lectures don't give the best return as me actually going and solving problems in the real world.
And I found this so true for me when I was learning AI. As I went into the real world and talked to real teams and built out solutions that saved them so much time and money, that's when I learned the most. Now, the cool thing with this skill is that it shows you exactly how you can get the most out of your learning style with Google's Gemini notebook.
I'll show you what you can create for all four types of learners in Gemini notebook. But for me, it's saying that. Study questions, flashcards, and key scenario breakdowns would be the best for me to learn.
So if we hop back into our notebook here, you'll see on the right side that there is a studio. And there's a ton of different things that you can generate. For my visual learners out there, mind maps are going to be key for you.
I created one right here, which breaks down exactly how prompt engineering works. And it goes through like a tree diagram where one section expands into the next and it keeps going so you can see all the relationships. Then you can just click on one of them and it will open up a question regarding that in the chat and get information from your sources.
For my auditory learners out there, the audio overview is insane. I actually still use it even though I'm not an auditory learner. This will create a podcast explaining the concept for you so you can just listen.
I'll play a quick snippet of it for you here. If you set your maximum token output length very high and you misconfigure the balance of temperature and top K, the mathematical probabilities can just flatten out. Pretty insane, right?
For the reading and writing learners, you can click on reports here and then create a study guide. This study guide breaks down all of the higher level concepts into very digestible. parts that you can go through multiple different steps and learn all the key points.
And then for my fellow kinesthetic learners, you can create flashcards. This will ask you questions so you're actively thinking and interacting with the site and then answering if it's true or false and it tells you if it's correct or not. And that brings us to point number three of learning anything with Gemini Notebook, which is that teaching is learning.
The ancient Roman philosopher Seneca even had a quote about this saying, while we teach, we learn. But if you're not a philosopher, there's actually a study on this conducted at the University of California in 2013, where they discovered that the teaching groups significantly outperformed the control groups. The study went like this.
There were three different groups that were assigned. The first one was told to just study like normal for an exam. The next one was told to prepare a presentation for the exam, but they never actually ended up doing the presentation.
And the last group was told to prepare a presentation, and they actually gave the presentation to their peers. And it was pretty fascinating what was learned. In the short term, the people that studied just like normal learned the least, while both groups who prepared a presentation learned more.
But in the long term, one week later when these people were assessed again, Both the people who prepared like normal and the people who prepared a presentation but didn't actually give it performed worse. When you compare it to the people who actually gave the presentation who performed well on the exam.
I could actually speak to this personally as well because me teaching AI to you guys on this YouTube channel. has helped me understand AI concepts and how things work in a much deeper way. If we hop back into our Gemini notebook here, you can create a slide deck presentation.
And the key thing here for us is to create these presenter slides. These are going to be less detailed, more visual with some key talking points that you have to really understand the topic as you're giving a presentation about it. So here's the one that I created about prompt engineering.
We can see it goes through, gives a couple of main points and statistics that we can talk over, give some sliders that we can explain different concepts and how they scale a few frameworks and formulas and just some really good comparisons that. you can go and explain as you're giving this presentation. Gemini Notebook will prepare the whole slide deck presentation for you, and you just need to go in, understand exactly...
all the concepts that are on the slides, and then you can present it to someone else and you will be so much smarter than before you made that presentation. That brings us to the final of the four points, which is that you should be able to learn on the go. I'm going to bring up another study here from the European Journal of Social Psychology, which explains that habits are formed when you are able to carry out the same behavior daily in the same context for an extended period of time.
That relates to learning because if we can build the habit of learning something new in the same context every single day, the likelihood that it sticks with us will be so much higher. I can relate to this one personally because for me, going to the gym every single day has been a habit for years. I was really big into powerlifting actually, and those training sessions can be up to two or three hours long.
Now, I used to just listen to music throughout those sessions, and it was great, but I realized that if I want to be learning something new, this is the perfect way to do it. If I could stack these hour blocks every single day of the week, the amount I'll be able to learn over the course of a few weeks or a few months will be insane.
Not only that, but I was even working at a gym and just vacuuming the floors with nothing else to do. So if I add in the time of my training plus the time I was just sitting and vacuuming floors, That's a crazy amount of learning that I can make in that time.
So for me, podcast was the easiest way to learn throughout this time. So I can keep my hands free to do all the lifting and cleaning machines, but still be in taking new information, which is why I can't say enough about this audio overview feature within Gemini Notebook. But it doesn't even stop there because there's other times where I'm just using my normal Gemini chat and I want to understand things from the sources that I created in my Gemini notebook.
And luckily, there's an integration for this so that even if you're just using your phone or you're just on the website for Gemini, you can always tap into that notebook that you created. This even relates back to the layers of AI knowledge from earlier. Because while Gemini Notebook is so strong on this data section, there might be times where you want to leverage the web searches or information from the training model itself.
And being able to plug that Gemini Notebook into the Gemini Chat, which can leverage the training and the web information, allows you to stack those layers in any way that you want. Here in my Gemini Chat, I was asking it to explain prompt engineering for me from... just a normal chat interface where it can search the web or just understand from its internal training, which was great, right?
It gave me some pretty good responses, but then I was like, oh, I actually don't understand this fully. Can you explain it a little bit differently? So it gave another response and another example.
But I was like, why don't I just use the sources that I know work so well for my notebook? So from here, I click the plus sign. I did more uploads and I chose notebooks in which I can just add in my prompt engineering notebook.
which I did. Then I just asked to explain from these trusted sources in my notebook. And it pulled exactly from the Google prompt engineering white paper and the prompt essentials course that I linked in my notebook and gave me an accurate response.
Congrats. Now you're equipped to go learn. anything using Gemini Notebook.
But if I had to give one recommendation on what to learn next, it would be AI agents. It's pretty clear that AI is the future and having agents be able to work for you 24 -7 is what separates the people that are going to win in life and business in the future versus the people that are stuck clinging to the manual grind work and are pretty clearly going to fall behind.
If you've been exploring Gemini and Gemini Notebook, then you're probably pretty familiar with the Google Suite at this point. And Gemini Spark is a brand new AI agent that can run for you 24 seven completing tasks while you're sleeping.
Luckily, I recorded a video on exactly how you can set this up for yourself. So if you want to learn that next, then go ahead and watch this video right here.
The Hook
The bait, then the rug-pull.
Zuckerberg taught himself to code at 10. Tarantino skipped film school for a video store. Jay-Z learned business on the street. Shawn Weigand names the pattern behind all three, then spends the rest of the video showing how Google's newly rebranded Gemini Notebook turns that same self-taught instinct into a repeatable four-step system.
Named ideas worth stealing.
03:23model
Layers of AI Knowledge
- Data
- Web
- Training
A pyramid ranking how grounded an AI tool's answers are: data you explicitly provide is most trustworthy, live web search is next, and the model's training data is least accurate. Used to rank Gemini Notebook above Claude Projects, Perplexity, and ChatGPT.
Steal forany framework comparing RAG-grounded tools to general-purpose chatbots
07:58acronym
VARK learning styles
- Visual
- Auditory
- Read/Write
- Kinaesthetic
Four learner types used to pick which Gemini Notebook Studio output, mind map, audio overview, study guide, or flashcards, will actually stick for a given person.
Steal foran onboarding flow that recommends content format by learner type
How they asked for the click.
VERBAL ASK
16:45next-video
“if you want to learn that next, then go ahead and watch this video right here”
verbal pointer plus on-screen end card to a companion video about setting up Gemini Spark agents; the description also stacks two free downloadable skills and a roadmap opt-in above the timestamps.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
Visual structure at a glance.
