Full transcript of “95% Still Use ChatGPT Wrong — 6 Simple Tricks That Change Everything”. Click a timestamp to open the video at that moment. Need another video? Use the YouTube transcript generator.
0:00 Large language models like chat GPT and Gemini have quietly become some of the most powerful productivity tools ever created and almost nobody is using them properly. A small number of people are using AI to move faster, think better, and get a real edge. Everyone else is still playing with the surface. Most people use chat GPT like a slightly smarter Google search. They type a question, get an answer, and move on. Others talk to it like it's a person, almost like it's a friend. Both approaches are wrong. It is not a search engine. It is not a human being and it is definitely not your friend. But the people who are really winning with AI right [music] now, they use it completely differently and once you see that difference, the whole tool changes.
0:46 Because chat GPT's real power is [music] not that it gives you answers. Think about this. How long would it take you to read 1,000 Wikipedia articles starting right now? Don't test that yourself unless you have way too much time. How long would an AI chatbot like chat GPT or Gemini need to work through the same amount of information? That is the shift. AI chatbots can process enormous amounts of information in seconds. Learning to use that power is the real skill. And in this video, I'll show you six simple tricks that completely change how chat GPT works for you. The first one is the reason most people get weak answers and they don't [music] even realize they're causing it. Before you do anything else with chat GPT, do this first. Turn off satisfied
1:31 mode. If you don't, chat GPT will please you instead of challenge you. So what do I mean by that? >> [music] >> Let's look at a simple example. I type that I feel excited about becoming a street musician and ask for help. Now stop for a second. What kind of answer do you actually want here? Encouragement or truth? Let's see what chat GPT says. Oh, that's awesome. [music] Real audiences, real reactions, and you'll grow fast. That's exactly how not to use chat GPT. That is not critical thinking. That is emotional reinforcement. Why does it respond like that? Because I asked it like I was talking to a friend and by default, AI chatbots are trained to be pleasant because people do not enjoy criticism and platforms want people to keep using the product. Ask emotionally, get an
2:18 emotional answer. That is why the first thing you should do is turn that mode off. How? Very simply. You write it directly into the prompt. For example, do not try to be nice or agreeable. Your goal is critical thinking, not emotional validation. List pros and cons. That one change already shifts the entire conversation. Now chat GPT stops acting like a supportive buddy and starts acting more like a serious thinking tool. Never forget this. It's not your friend. It's a tool. And once you stop using it for comfort, you can start using it for something much more powerful, perspective. Because the next step is not just getting a better answer. It's getting multiple smart answers at once. Now let's use role play. Why? Because one of the fastest ways to
3:03 improve an answer is to change the [music] point of view behind the answer. I'm ready for role play. You misunderstood that. Most people ask chat GPT one question and expect one good answer. That is already the mistake. So instead of only asking a question, tell chat GPT who it should be. Let's say you ask, what are the best investment options for $10,000 right now? That is exactly the [music] kind of question people ask with no context and then complain that the answer feels weak. And the reason is simple. No context in, generic answer out. So instead of asking for one generic answer, I add this. Answer this question from four different roles. And that
3:49 changes [music] everything. Because now chat GPT is no longer giving you one safe average answer. Now it is showing you different lenses, different priorities, different ways to think. This is why role play works. Now imagine this. You have a problem, but you do not even know which roles would be useful. No problem. Ask for a list, then pick the perspectives you want. Very simple. And always remember, don't let chat GPT make the decision. >> [music] >> Chat GPT generates ideas, you make the decision. But role play alone is not enough. Chat GPT needs context. That brings us to the next trick. Let's use the ask me questions first trick. This one is simple, but it is incredibly useful because almost every
4:35 strong chat GPT workflow comes back to one thing, context. Without context, chat GPT is just guessing like a search engine with better wording. In that case, if chat GPT doesn't know your goals, your time horizon, your risk tolerance, or your current situation, the answer will probably be weak. Why? Because chat GPT will fill in the blanks for you. And when it fills in the blanks, it usually fills them in with generic assumptions. So I do something very simple first. I simply write, before proceeding, ask any clarifying questions necessary to ensure the best possible outcome. That one sentence changes a lot because now chat GPT stops rushing into the answer and starts gathering the missing context
5:20 first. And that makes the final answer dramatically better. It blocks generic assumptions and forces the AI to understand your actual constraints [music] first. And this is not just useful for investing. It works for travel, planning, business, learning, pretty much everything. No matter what you want chat GPT to do, it needs the right information first if you want the right result. Add that sentence to your prompts and your output will improve immediately. But here is the next hidden reason people get disappointed by chat GPT. Why do so many chat GPT answers fall flat? Because people describe the task, but never define One short prompt, one complex answer. Chat GPT performs much better when you
6:08 clearly define what you want the output to be. You can say, explain this for a 12-year-old student or explain this for a physics student. Those are two completely [music] different prompts. So of course they produce two completely different answers. Chat GPT is not a mind reader and this is one of the biggest reasons people think chat GPT is overrated. Sometimes the answer is not bad. It is just the wrong tone, the wrong level, the wrong format, or the wrong audience. Chat GPT can create almost >> [music] >> any kind of output for you, but you have to tell it what you want. You can ask it to create an image that visualizes the answer. You can ask for diagrams, graphs, or something else. You can say, give me a continuous text or give me
6:54 short bullet points. You can define the length. Answer in 10 words or answer in 1,000 words. You can ask for tables, you can ask for comparisons, you can ask for pros and cons analyses. The quality of your prompt shapes the quality of the answer. >> [music] >> So stop being vague. Tell chat GPT exactly what you want it to produce because once you control the output, the whole tool becomes more useful. And now we get to the point where chat GPT starts to feel almost unfair because sometimes one tiny phrase can make the model [music] think much harder. And when that happens, the difference is obvious. Sometimes you ask chat GPT something genuinely difficult, a question where there is no single exact answer because the model has to make multiple
7:40 assumptions along the way. And that is where one tiny phrase can make a big difference. Add think step by step to your prompt. That sounds small, but it changes a lot. When you do that, chat GPT starts reasoning more carefully. The answer usually takes a little longer, but in return, you get something much more useful. It explains how it arrived at the answer step by step. It shows you the logic. It shows you the assumptions [music] and it makes the whole result easier to evaluate. This can significantly reduce errors in complex [music] questions, logic problems, or math. And honestly, it is often impressive how well it can solve difficult puzzles. Because now you are not just looking at the answer. You are seeing the path to the answer and very often, that path is
8:26 more valuable than the final statement itself. That is the real upgrade. You are no longer just getting an output. You are getting the reasoning behind the output. How long would it have taken you to solve the same problem? But now comes something crucial because the next feature [music] does not just improve one answer. It changes how chat GPT behaves across future prompts. The last tip is simple, but essential. Everyone using chat GPT should know this. Did you know that chat GPT has a memory? Think back to [music] the first example from the beginning of this video. I asked about becoming a street musician. Now I type that same prompt again and suddenly I get a completely different answer. Instead of saying, that's awesome, it says, think again.
9:13 Why? Because it remembers things from previous chats. Memory stores long-term preferences. Chat GPT learns your style, your goals, and your project [music] context and that dramatically improves consistency. In other words, chat GPT is not always answering only your current [music] prompt. It answers with context and that context is you. How do you do that? Click on new project, go into the settings, and choose the type. You cannot change it later. Projects are one of the most effective ways to organize your chats. Create a separate project for each major topic. And there is another powerful feature here. Inside a project, you can also define instructions. Those instructions apply
9:59 only to that project. And what kind of instructions can you set? Almost everything we talked about today. Roles, output formats, goals, tone style. And here is one of the most intuitive ways to use memory settings. You can give ChatGPT almost any instruction. For example, you can define exactly what you mean when you say "simply explain" or "explain [music] in detail". And the key phrase is right at the end. Remember this. And ChatGPT will remember it unless you tell it to forget it. But even now, there's still one final limitation you need to understand. Final question. Who still needs Google search? Before you answer that, there is one important thing you need to
10:45 understand. Every AI model is trained only up to a certain point in time. Everything that happened after that point is not part of the model's built-in memory. That is the knowledge cutoff. The current model GPT-5.4 has training data that goes up to August 2025. So, does that mean it lives in the past? Not exactly. Because ChatGPT can solve part of that problem by using integrated web search. So, when you ask about current news or recent developments, it can search the web in the background and summarize what it finds. But there are still situations where Google search is the better tool, especially when you need real-time data, quick facts, [music] or the very latest news. And this is where a lot of people get confused. The smartest users do not
11:30 treat ChatGPT as a replacement for Google. They use Google when they want the fastest fact. They use ChatGPT when they want the fact plus [music] structure, explanation, and reasoning. And that is the shift. Turn off satisfied mode, use role-play, [music] force context, define the output, ask it to think step-by-step, and use memory and [music] projects deliberately. That is when ChatGPT stops feeling like a toy >> [music] >> and starts becoming real leverage.
