A Step-by-Step Roadmap to Mastering AI: From Beginner to Confident User

When I first started my AI journey, it felt so overwhelming. GPT4 is the most advanced. later, these AI models is insane and ranked number one on the leaderboard.

It's so exhausting. And after a year of intensively using AI. I can tell you most of these announcements are just noise.

What you really need is a clear roadmap that makes AI works for you. So in this video, I'll show you how to start your AI journey today with a step by step roadmap that anyone can follow from an overwhelming

beginner to a confident AI user. Let's Go First, build your AI learning system. And most importantly, do not start with AI tools.

Forgot those top 100 AI tools type of content. I hate to say this, but most people just jump into AI tools directly when they start AI, because that seems like the easiest route.

And here is the learning system that worked for me. First, start with learning fundamentals and core concepts before tools. Personally, some resources I recommend Andrew Ng's Generative AI

for Everyone course and the chat GPT for everyone from Learn Prompting. And PromEngineering. ai that covered lots of fundamental concepts.

These are all free resources. There are no fancy tools, features explained, but these will teach you the core concepts about generative AI.

What AI can and cannot do, how it thinks, its limitations. Bonus is, also subscribe to 3Blue1Brown's channel.

It has super amazing explainer videos and visualize some of these complex LLM concepts into easy to understand videos for non technicals. When you understand concepts like token limits, RAG, hallucination, you can work

around any tool's limitations instead of being stuck when feature changes, because these core technologies will not change much like the features. And then once you have some basic core concepts, start

accelerating learning using AI. That's what I call "AI-Powered Learning" to deep dive into any specific topics that you are not familiar with.

I mainly use the AI search engines. Perplexity, or you can use any other similar alternatives. For example, I want to learn about AI agents and then I can use perplexity

to prompt to find everything about AI agents for me as a beginner. Including key concepts, how it works and limitations. I can even use the new features deep research to make sure it can

find more in depth resources for me. Immediately, you can quickly get an idea about the topic. Another bonus tip is you can actually prompt Ploxly to create a learning path

for you, which is also my favorite. Just ask what are the key skills, tools, recommended courses, resources, so you can get more ideas

and then you can even take a step further and import all the quality sources you gather to an AI note taking app like Notebook LM and build your own focused learning projects, asking follow up questions.

I'm not kidding. This system has sped up my learning by 10x, and this is a true game changer for me.

That I create a whole separate videos about this. I'll include the link below, the next is to develop good thinking skills.

Yes, you cannot skip this. And this must be part of your AI growth journey because the more we use AI, the more good thinking ability we should develop.

Research from Microsoft has found that when more users are using ChatGPT, the idea becomes more homogeneous, meaning everyone was coming up with similar ideas. So first, learning to ask better questions or prompting when using AI.

The better prompting I mean here is not about knowing the techniques, but mastering your thinking and communication. My tip is do not rely on those prompt generators in your first attempt.

Try to understand the type of your tasks by thinking more deeply on your end goal first and communicating it clearly to include important context that matters,

for example, giving the context that I am new to AI, Explain RAG technology to me like I am a working professional with no technical background will make a huge difference.

Second experiment using different AI models. Sometimes when I'm working on something important, I might test the same prompt on different AI models.

For example, when brainstorming packaging for my YouTube videos, I will use the same prompt on Claude and O3 to see how they approach it differently and help me to think from different angles.

Another big tip is whenever you see the big AI model announcement, don't just ask what are the cool shiny features The real question is not about which model is the best, it's what

specific problems that it can solve. All those benchmark scores and capability claims are just noise if you don't know how to think critically about them.

So try to develop this thinking habit to always ask more questions. Stay curious, not about the tools, the features, but the real problems and solutions and think more deeply beyond just the surface level.

The next, develop your AI news filter and stay in the loop in a strategic way. I know there is lots of AI news out there. It's so easy to be information overload because AI just changes super fast.

But with the right strategy, you can stay ahead without being overwhelmed. First, creating quality sources for AI news. And my tip is just focus on a few quality ones.

You can start with those more well known newsletters first and see if you find any of them helpful and stay subscribed. Again, the key is to be selective.

I have tried and subscribed to many before and these are the ones that I finally stick with. Everyone has their own criteria, but my criteria is don't just look at

who is fastest in reporting news. Instead, evaluate if they have made an extra effort to provide more insights or unique angles that you can't get from elsewhere.

And to make this even more systematic, you can even set up a really simple schedule task on ChatGPT to search for the more broader AI news and send it to your mailbox on a daily basis.

So every day at 10 a. m., I will receive a daily digest. What I really like is how it highlights the most important information in an email subject.

So even if it's. just a scheduled task. You can decide to click or not by just looking at the headline.

Of course, you can add more specific in the prompts, perhaps ask it to summarize the implications for you from this news or focus on specific sources. and you can even change the frequency that fits you.

So this way, you can stay informed without the constant anxiety. I also recommend following Top Voices on LinkedIn or X who constantly share their hands on experience with AI.

So these people not only share the tactical level of using AI, but also on the strategic level and give insights on how to apply AI in a more structured way. So our goal for this stage is to make AI become part of your daily workflow and

to make you become more familiar with AI. Whenever you receive AI updates, you become less anxious because you start tuning yourself to these AI languages, building that daily AI awareness habit.

And eventually able to spot difference between real innovations or just marketing hype. And that comes back to step number two, your good thinking skills.

Next it's time to explore more AI tools and start building your own AI tech stack. And there are two key processes here. First, have a basic concept on the AI tech stack structure, and here's

what I found the most applicable for non technical working professionals. So underneath we have the core AI models or the chatbot layer, GPT4o, o3, Claude, Gemini, Grok.

So these serves the foundation and can be used for a broad category of tasks. But on top, we still have the application layer. So these are different specialized categories for daily tasks like

text and writing, research, visual creation, meeting. My suggestion is start with one major core model or plus one more specialized tool based on your specific task needs and test them thoroughly for a

few weeks and only add new ones after mastering 80 percent of the current ones. For example, I use cloud as my core model for most tasks, but for research, I add Perplexity because it excels at real time information with the search focus.

And here is where most people will start getting overwhelmed. And so we need the second process, a decision framework for AI tool selection. Stage number one, review true needs before features, before pricing, ask yourself.

What specific real problem needs solving, which tasks are taking up most of your time, and most importantly, have you fully explored what your existing tools can already do?

Let me give you a real example. Previously, I need AI to extend images for my thumbnails. My first instinct was to try Runway, a popular AI visual creation tool.

But eventually I found Canvas Magic Expand, which I already had, did exactly what I need. So the truth is your existing AI tools is also improving because

the AI space is very competitive. And stage number two is about exploration and testing. For most of us, we'll start researching the available options,

testing it with the real use cases. But besides the features, I will also study the company behind the tool. What is their AI development philosophy, the data handling policy?

Are they reliable? Are they transparent? So like Claude, Anthropic is transparent about the AI safety principles.

While I found other companies are being vague on that aspect. I also studied the founders interviews that tells you a lot about how they handle your data, their view on the AI development.

And this is also part of the reason why I stick with Claude. And finally, decision time, I use this simple formula to get a sense of the potential ROI.

And it doesn't have to be exact. So you estimate the approximate monthly time cost you save and minus the tool cost and hidden cost.

That is your learning time or integration time to estimate the actual value from a tool. An example is this Descript.

I use it to streamline my video editing process and it saves me at least eight hours per month. So using this formula, I can estimate the monthly value to justify the ROI.

So now you will understand why there are scenarios you might be paying more than you need. Number one, when a tool cost exceeds the actual time saving.

Number two, when hidden costs are too high, like complex integration. Long learning curve, hidden privacy cost. And number three, when the value doesn't scale, meaning you rarely

use it, perhaps only once per month. So that's why personally I just have a simple AI tech stack. Each AI tool in your stack must solve a specific problem that

your existing tools can't handle. Don't add complexity unless it delivers clear values to you. The quality of AI tools matter more than the quantity.

Now we can put all of these into practice because having a framework is great. We need to actually apply it in order to level up our AI skills. There are two core principles.

First, a dedicated time to using AI. This is a must because the more you use AI, the more you will get familiar with it.

You can start with 30 minutes to one hour per week. It's totally up to you. When I first starting out, I spent maybe 15 to 40 minutes per day

playing with AI and eventually now at least a few hours per day because the goal is to transition you as a novice to eventually a power user. And you can see that what differentiate power users from a normal users are not

only the hour spent, but also how they use AI, like research and try new prompts. And experimenting with different AI use cases and developing an AI first mindset by asking the question, can AI help me with this task?

Second, follow the one tool, one task rule. That means using one single AI tool for one specific recurring task before expanding further when starting out.

For example, when clock first launch is project feature, I only focus on using it for automating my YouTube title. Thumbnail and description ideation.

So I set up custom instructions with task code and prompts until now, I'm still using it to automate these tasks for me. And eventually I expand this feature.

And now I'm using to manage my other projects. So each one has its own set of test prom knowledge base and workflows, but it all starts with the one task concept.

So don't try to do everything, but start with a task that get you stuck the most. And as you gain more experience prompting with AI, you can even build your own prompt library.

For myself, I set it up on Notion, and you can pick whatever platform that you like. So perhaps whenever I try a good test prompt or got some inspiration from other experts and tweaked it into my own version, I will save it here.

Like the YouTube comment analyzer, I'm inspired by other experts and I tweaked it to fit my own needs. I don't have many prompt templates here, but only the one I found

really useful as a starting point. Because again, eventually you should develop your own thinking skills. Your own thinking skills.

So these prompts are always just starting point. so using these two principles, you will naturally develop the AI first mindset and transition yourself into a powerful AI user.

Your success with AI is not measured by how many AI tools you know, but by how strategically you can think and adapt.

The key is building confidence through systematic learning and thoughtful implementation, and not chasing every new tool and new release. So focus on developing your AI thinking skills when everyone

is using the same AI tech. This will help you to stand out from the crowd. If you want to learn more about my AI tech stack, detail model reveals, and

how I use them for what tasks, check out my feature content on my community. You can find the link in the description to join. And before you go, also check out these AI Power-Learning System that I mentioned

earlier, and some other videos that will help you to start your AI journey. I will see you next time.