It is now 2026, and prompting AI models is very different from when ChatGPT first came out in 2022. Using AI well is one of the most impactful skills you can develop. And people that are not yet at the cutting edge of AI usage often run into AI generating frustrating outputs. I want to make sure you're an expert prompter and can take advantage of today's AI tools, which are much more powerful than they were even a year ago. Let's take a look at two different experiences, the AI novice and the AI power user. Many AI experts have learned to use it to answer hard questions. In contrast, many people, including AI novices, may have gotten used to using AI for simple questions, as if you were prompting it like a Google search. So if you ask it, does Taco Bell still have the double decker taco? And maybe you get an answer like that, which is fine. But if you have much harder questions, you can also ask it of the AI and give it time to think. For example, if you're looking to buy a car, you can upload to most of the commercial services like ChatGPT, Gemini, Anthropic's Claude, or others, a set of documents, including car specs, quotes, insurance plans, and ask it: what are the trade-offs for these different cars I'm thinking about? And tell it to read everything and to think hard before answering. And this can cause AI to spend many seconds or even minutes to think and then compile a detailed report for you. I find this a huge time saver for a lot of things I have to do. Another example. AI power users have learned to provide the right context or the right background information to the AI to set it up for successfully answering your question. In contrast, I see some some AI novices use a short prompt and hope the AI will fill in the blanks. But if you think of AI as maybe being akin to a really smart fresh college grad—highly motivated, but that doesn't really know that much about you yet, then a short prompt sometimes doesn't give it enough information or enough background context to answer your question accurately. So if you tell the AI "please write a good self-review to send to my boss", the AI doesn't know what you've actually done over the last year because you haven't told it yet. And it might write a very generic self-review, which isn't that helpful. In contrast, I find that AI power users almost have empathy for the AI. I don't want to overly anthropomorphize the AI, but if you could put yourself in the shoes of someone getting a set of instructions from you, you can ask yourself: will they actually know enough about you to do a good job on the task you're assigning them? So an AI power user in comparison might upload a lot of information to the AI, maybe give it a screenshot of a project tracker, showing what you've worked on, recent project docs, maybe voice memo notes where you talk through the projects and then tell it to write a self-review to send to my boss. And that could do a much better job capturing what you're most proud of. One of the things power users have learned to do is how to prompt AI to get honest feedback. A big problem with AI is it often wants to please you. In fact, many AI systems were trained to try to make their users happy. And if you ask it a biased question, it will often give a biased answer because it's trying to tell you what it thinks you want to hear. For example, if you say: I have a great business idea, mobile tie dyeing. Critique it. Because you called it a great business idea and you're saying it's your idea, the AI will naturally want to please you and say, what a great idea. We sometimes call this sycophancy. And it's well known that if you give even a hint of what answer you're hoping for, there's a good chance the AI will just reflect back your preferences or your preconceptions. In contrast, AI power users tend to ask neutral questions that don't give any hint to the AI for what answer you're hoping for or not hoping for. Or if you give it a rubric or grading criteria to tell the AI how to form the basis for its answer, that also forces it to be more objective. For example, if you were to say: please analyze the following business idea objectively: mobile tie dyeing. And don't just make up a bunch of things for what you think. Use the rubric, or the grading criteria above, such as: is there a problem? Is there a market? Do I have a competitive advantage? If you give instructions like this to the AI, then the AI doesn't know: are you hoping it'll tell you it's a great idea or that it will save you from spending a lot of time on the bad business idea? And it's much more likely to then tell you something like, oh, this idea is an 8 out of 100 and also why this scores low. In case you run a mobile tie-dye business, I wish you really best of luck. And AI could also ask some useful questions to help you think through how to make the business even better. Lastly, I found that AI novices and AI power users ask AI to write in very different ways. Novices will just ask AI to write stuff, like write a blog post about the BlackBerry. And it will generate a bunch of text that maybe looks like this, which sounds like AI slop. It's a bunch of generic text that's just not that interesting and takes up a lot of space. In contrast, an AI power user will often not ask the AI system to just jump in writing directly, but instead ask AI to first outline an article and then critique the outline and maybe iterate a few times with the outline to shape the article. And only then ask AI to start to draft the final article. So give it a set of uploaded notes as context. An expert may say, outline a blog post about the BlackBerry based on my notes so it knows what you want to talk about. And the AI may start by giving an outline. And you might then give feedback to the AI about what you like and what you don't like about the outline. And even iterate a few times, have a few back and forth rounds before you have an outline that you're satisfied with. And maybe only then expand the outline into bullet points. And maybe even go back and forth a few times to critique the bullet points before you're satisfied with that, and then expand it into the final text. This type of power user workflow is much more likely to generate some text that you're happy with, as opposed to AI slop. And in this type of workflow, you're treating the AI as a thinking partner to almost help you brainstorm and explore different options for what you might want to write. AI systems do make mistakes, but maybe fewer than most people think, especially if you prompt it well. They made a lot more mistakes back in 2022 or 2023 than they do now. But a lot of widely publicized mistakes that AI has made, some of which went viral on social media, has made people think that AI maybe makes even more mistakes than it actually does. There's a well publicized one where people asked it how many R's are there in the word strawberry, and it thinks there are two R's. And here's one that I found amusing: I want to wash my car. Should I walk or drive there? And the AI says: walk, which would leave you there without your car. But these viral examples are not representative of AI capabilities. In contrast, power users know that AI can deliver significant value through tasks like doing deep research and writing research reports, or taking your personal data, like your health or heart rate or running time data, and analyzing that for you. Or, something we'll talk about later, even building websites for you. I've seen being an AI power user tremendously benefit individuals, as well as their businesses. It'll save you time and improve your professional and personal lives. It'll help you to build lots of cool things. You'll learn how later in these videos. And being able to prompt AI at an expert level is a highly in-demand job skill, no matter what job role you're in. In the rest of these videos, I hope to take you from wherever you are today to being an AI power user. Much has been said about AI being useful. I find using AI really fun as well, and you'll see a few examples of that in these videos too. Now, one foundational piece of knowledge that helps you work with AI is understanding where it gets its knowledge from, so that you can better predict when it'll get something right, and when you maybe shouldn't count on its answer. Let's go on to the next video to learn about how AI gets its knowledge.