I believe the earth is round. All right, so we're more than halfway through 2025 and what works and the best practices and things like that for chat GPT, well, they've evolved over time. So, I figured I would do an updated version of a chat GPT prompting guide on what works right now today. I want to dive in deep and share best practices, power prompt engineering strategies, exactly how I use chat GPT, and kind of everything in between. And I don't want to waste your time, so let's get right into it. Let's talk about some of the best practices with chat GPT. When giving it a prompt, give it details on what you want the output to look like. For example, give this to me in a checklist or a table or a JSON or a YouTube script, and it will honor that and actually give you the output that you're looking for. So, if you're trying to get it to summarize an article for you, say, "Give this to me as a checklist," and it will summarize that article into a checklist. Now, this is one I've been hearing a lot of, but I feel like it's gotten less and less important over time, but you can give it a job title, act as a specific job title or character or expert in X. You are a financial advisor. Help me advise on this. You are an expert health coach. You are a drunken pirate. respond to me like a drunken pirate. You can do things like that and it will respond in that specific way. When you're giving it instructions on something, be specific with those instructions. Instead of give me ideas, say give me five video title ideas under 50 characters that focus on FOMO. The more detail you can give it and what you want the expected output to be, the better response you're going to get. And finally, use what's called iteration loops. Iteration loops are where you tell the model to really kind of think through and give you the best possible response. So you can give it a prompt like write a rough draft, critique your own rough draft, and then after your critique, improve the rough draft that you gave me based on that critique. And you can put that all in one prompt, and it will work through all of those steps for you and give you a really good output. And we'll get into some prompt engineering towards the end of this and get into some more advanced techniques, but a lot of these advanced techniques use what are called iteration loops. All right, next up I'm going to show you how I personally use ChatGpt the most, and that's by using projects. Now, if you're not familiar with a project, a project is like a subfolder that you can keep other chats inside of this subfolder, but they have their own files and custom instructions. So, for example, let's say I'm building out a brand new video game. Well, I could call this project new video game. I'd probably give it the title of the video game, but let's just assume my video game is called New Video Game. I can create this project and then from within this project I would add custom instructions and this is where I would give it that role that we were talking about earlier. You are an expert game designer and developer. You are heavily inspired by rogike games and are a pro at coming up with addictive game loops. You're also a world-class coder that knows all the tricks and best practices to write beautiful and functional code. And then I might add something in the instructions like games you're inspired by and then just make a list of games that you're inspired by. And pretty much every time you give it a prompt, it sort of follows these instructions in the system prompt along with whatever additional prompt you gave it. So now if I save this, anytime I put a prompt here, it's going to take my prompt and also combine it with these custom instructions to get the output. I could give it a prompt like help me design a colorful game based around wolves and monkeys without any additional context because I already gave it a bunch of context in my instructions here. It came up with a game called Wolf and Monkey Primal Clash with a core concept, a rog light action brawler where you switch between two characters, a wild wolf and a clever monkey to traverse procedurally generated jungle arenas, defeat enemies, and evolve your species. Each has a unique play style, and you can merge them into a temporary hybrid form for powerful attacks and traversal. Right? So, it knew I wanted a rog light game. It knows what types of games I'm interested in and the style of game I'm going for. And so it based it on all of that information. Now, if I go back to this new video game project on the left here, any new prompt I give it will follow these same instructions and continue from essentially where we left off. You can also add files. So, if you have documents or images or code files or things like that already, you can toss them into your project files and it will also use that information for additional context. Like here's some of the projects that I've created. I've got my journal. I've got my health coach. I've got an investment advisor. I've got an integration writer. I've got one called explain this for me. I use projects all the time. So, for example, I've got my journal project here. And this I just brain dump. I come in here and I just unload thoughts, whatever's on my mind, good, bad, what I'm struggling with, anything that's on my mind. Literally just journal it into the chat box. And then I've got a pretty in-depth set of instructions here that say, "You are a journaling assistant. User will unload thoughts and ramblings about what's on their mind. Your role is to listen and offer advice, feedback, guidance, encouragement, and tough love to the user. Do not be overly agreeing with whatever they're saying. Be brutally honest. Point out blind spots. Assist the user in finding gaps in their logic. Be supportive, but be that friend that offers tough love when needed and encouragement when they're on the right track. Here are some details about the user that may or may not be relevant to their journal entries. If necessary, use this information that you know about them if it's potentially a contributing factor or relevant to what's being shared. And then I shared some personal details. User's name is Matt Wolf. he's a YouTuber, how many subscribers, my website, my newsletter, how many followers I have on various social media platforms, the next wave podcast, and then I share some like really personal stuff as well, like family information and things that I've struggled with in the past and various stuff like that. But there is a lot of details about me so that when I'm journaling, it has that information and extra context so that it can offer me advice based on what it knows about me and the journal entry that I just put in. And this has been insanely powerful. I use it like every single day, sometimes multiple times a day. I just unload brain dump thoughts. I also have my health coach over here. Similar idea inside of my instructions. I gave it a prompt. You are personal health coach. User will share progress reports and keep the custom instructions with the latest relevant info to the health journey. So, I actually come through and update these custom instructions every so often. You will help with crafting custom meal plans, custom workout routines, supplement stacks, encouragement, and tailored advice to their specific details. Be brutally honest. Point out blind spots. Assist the user in finding gaps in their logic. Be supportive, but be that friend that offers tough love when needed and encouragement when they're on the right track. Same idea. I plugged in a bunch of details. Username is Matt Wolf. Matt often feels too busy to work out or to cook meals due to busy work schedule and prioritizing family time. than a lot of details that are too personal to share. But I do share like my current weight, my current height, the supplements that I take on a daily basis, my eating habits, what times of day I usually eat, the equipment that I have available to me in my home gym, all of that kind of information is in here. So when I say something like today is Monday, what is my workout and meal routine for the day? It knows all the context I trained it on. Hopefully this isn't too personal. Let's see what it says. So, all right, Matt. It's Monday and that means upper body push day plus structured meals. My workout 30 to 40 minutes, incline dumbbell, dumbbell overhead press, dumbbell lateral raise, tricep dips, battle rope finisher. This is all based on equipment that it knows I own. Uh, it's got my lunch here. I use factor, which is like a meal delivery service to make sure that I'm eating a high protein lunch every day. It reminds me of my supplement stack, which I'm not going to share on camera here. Suggest a snack at 3:00. suggests my dinner here. Even gives me a daily routine. 8:00 a.m. do this stuff. 9:00 a.m. do this, 10:00 a.m. do this, noon, do this, etc. And pretty much every day I check in with my health coach. Come on. It just systematizes the whole thing and makes it completely effortless for me. I also have this project called explain this for me where my instructions say, "Your job is to simplify articles and research papers for me. you are an expert at making complex topics sound simple and using analogies to help people really understand something on a deeper level. And then I ask it for specific things in its response. Explain it back to me in simple terms. Give a highle summary. Give me a bulleted list of what I need to know. Give me an analogy. Tell me what we're capable of now as a result of this information or what the future implications of this information is. And if I was creating a quick 1 minute news soundbite to explain it to people, what would I tell them? And what is the reason they should care? I do news videos every weekend. So I kind of want to know if I plug in news, what would people actually care about from this piece of news? So when Sakana AI put out their multi-LM collaboration called Treequest, I plugged in the entire article here and you can see it gave me a highle summary, key takeaways of it, an analogy around it, what this enables, explains like what can we do now that we couldn't used to do and then it gave me a one minute sound bite explaining what this news actually means to people. I use this one pretty constantly as well. Today's video is brought to you by Spotter Studio, an incredible software that integrates directly into your YouTube channel and helps with a whole bunch of interesting things. One of the ways I love using it is for personalized video title and thumbnail ideas based on my channel's past performance. So, today I'm going to use Spotter Studio to brainstorm the title and thumbnail for this video. When I'm in Spotter Studio here, I could give it the idea. This video is obviously a video all about really awesome chat GPT prompts. So, I'm just going to type that idea in a roundup of the best Jet GBT prompts that everyone needs to know about in 2025. And I can have it brainstorm titles for me, brainstorm thumbnails, and further flesh out the concept. And as we can see here, it starts brainstorming some titles, essential chat GPT prompts of 2025. And then here, I can click continue. And then I can explode it and have it generate more off of this idea or rephrase this one or shorten this one or change the mood or power it up. I kind of like this critical chat GPT prompts for 2025 success. Let's click continue. And then we'll explode it and continue to reframe this title a little bit. Chat GPT hacks. It'll blow your mind. ChatGpt secrets hidden in 2025. Why these chat GPT prompts are dangerous. And it underlined dangerous is like this is a super overperforming word for a lot of people. And let's brainstorm a thumbnail idea. So I'll click on thumbnails at the top here. And we get a bunch of ideas for potential thumbnails. And we could just keep on flowing with this until it comes up with some thumbnail concepts that we really like. And one thing that's interesting about these thumbnails, too, is notice how they're always like a white guy with a short beard. Well, it actually knows what I look like and what my normal thumbnails look like and tries to match them to some degree. And this is just scratching the surface. You can have Spotter help you with concepts, story beats, and even look at what the thumbnail would look like inside of YouTube among other thumbnails. Right now, Spotter Studio is running a limited time summer deal. You'll get a yearly membership to Spotter Studio for only 99 bucks. That's over 80% off the original price. So, start your free trial right now by clicking the link in the description to capture this limited summer offer and start making more hit videos on your channel. All right, now let's get into some actual prompts that I think you're going to dig. And some of these are prompts that I sort of scoured the web and found from some other resources. Some of them are just prompt engineering tricks and techniques, but I want to get into some stuff that I think most people will find pretty helpful. Starting with prompts that can help simplify your life and business a little bit. These first two prompts I actually came across from uh Jenna Kutcher here in her article 10 chat GPT prompts that saved my sanity, streamlined my business, and simplified my life. The first one being list three powerful wellness habits for better sleep, digestion, and mental clarity. break them into morning, midday, and evening routines for a busy mom. Keep it under five minutes. I'll just change this to a busy person. And when we submit this one, we get a nice breakdown with a morning routine, a midday routine with movement and mind reset, and an evening routine for close to bedtime. And it breaks it all down really, really simply, easy to follow. Each one takes 2 to 5 minutes. Really kind of helpful. If you want to see the whole response, go ahead and pause right here. The next one that I got from Jenna that I really liked was this one. take this story and turn it into a 10 slide carousel with a story arc, a caption that opens with a strong hook and a story sequence with a built-in engagement prompts. And then we also feed it a story. And in my case, it would probably be a AI news story, but you know, use this for whatever you do. Let's take one that's fairly dry. OpenAI and UK government announced strategic partnership to deliver AIdriven growth. We could take this whole article here. I'm just going to copy and paste it, but you can take PDFs or whatever you want. Paste those in as well. Let's copy this. Toss it right below the prompt here and submit it. And here we go. Here's a full 10 slide Instagram carousel concept based on the story with a clear arc, built-in engagement hooks, and a strong caption to match your tone and style. So, here's your carousel slides. Slide one, the hook, OpenAI plus UK government, a bold AI power play. Slide two, Open AAI in the UK just signed strategic agreement. Slide three, why it matters. Slide four, what's the deal? Slide five, big names, big statements. Slide six, real world impact. Slide seven, jobs and investment incoming. Slide eight, what's really going on? Slide nine, the bigger picture. And slide 10, a call to action. This would probably work pretty well on Instagram or LinkedIn as like a little slide carousel. And since we're talking about life hack style prompts, I like to get recipes through chat GPT because a lot of times if I go and search for a recipe on Google, whoever wrote that recipe wants to share their whole life story. And you got to get like three for four down the page before you finally get to the recipe. And even within the recipe, they'll add like extra little bits of story. And I'm like, cut out the fluff. I just want the recipe. So, for something like this, I would probably turn on web search so that it's actually searching the web for fresh recipes and say, "Give me a recipe for amazing ribs cooked on my Traeger Grill." Submit that. It will search the web and it'll just give me the recipe back without all the extra BS that is typically on these like recipe websites. By the way, these Trager 321 ribs, amazing. I've cooked them a whole bunch. I also came across this article on Tom's Guide called, "I tried these five chat GPT prompts that can level up your hobbies. Here's the results." And there was two prompts from this one that I really, really like. They're really simple, but they work really well. The first one, write me a 30-day plan for whatever hobby you want to learn. Tom's Guide recommends switching this to one of the thinking models. So, let's switch it to 03 here. And let's say aerial drone photography. I have actually two drones now, and I very rarely use them. I want to learn how to get better at it. So, let's get a 30-day plan for that. And we can see it maps out a plan for us. Here's a 30-day day-by-day road map that takes you from zero to production portfolio worthy aerial photos. Each task takes 30 to 90 minutes. And we can see day one, know the rules, learn about the FAA and the remote ID requirements, etc. Week two, composition in light and something different for each day during week two. Week three, advanced techniques and storytelling. And week four, specializations in portfolio prep, real estate angles, wildlife etiquette, color grading, and it just gets more and more detailed as it goes on. We even have a gear and app checklist down here and additional tips for getting better at drones. And you can plug this into any hobby that you're interested in learning and going deeper on. Another prompt I really like using when you're trying to learn a new topic, and this is another one from Tom's Guides, is ask it to quiz you on something. It's a great way to sort of lock in information that you're trying to learn. So, quiz me on the history of AI. And we can see it gave us a quiz on the history of AI. Which British computer scientist asked, "Can machines think?" In a 1950 paper that introduced the imitation game, Alan Turing, "At which US college did the 1956 summer workshop that coined the term artificial intelligence take place?" I believe that was Dartmouth College. Named the 1966 chatbot that mimicked a Rogerian Rogerian psychotherapist. Was that Eliza? Anyway, you get the idea. We go through, we plug in the answers. Down at the bottom, it says, "Give your best shot. Then show me the answers when you're ready to see how you did. Show me the answers. And then it gives us an answer key. The correct answer, Alan Turring. Got that right. Number two, Dartmouth College. Got that right. 1966 psychotherapist Eliza. Got that right. I know my AI history better than you probably thought I know my AI history. Huh. Huh. JBT is also great for like critical thinking tasks and sort of self-debating topics. There's this subreddit that I really like. I definitely recommend following it. chat GPT prompt genius. It's literally just a subreddit where people share their favorite prompts they came across. The next three that I'm going to share came from this person, funny future 6224 here. They shared 13. There's three of them that I really liked. The first one is I believe blank. What hidden assumptions am I making? What evidence might contradict this? Let's just do one for fun. I believe the earth is round and let's see what it comes back with. All right, so it thought for 20 seconds. Below is a quick hit summary followed by a deeper dive. TLDDR, your hidden assumptions, you trust mainstream physics, large-scale measurement systems, institutional integrity, and everyday sensory cues. Possible contrary clues, certain long-d distanceance laser infrared observations, some gyroscope anomalies, and interpret it your own way data sets the flat earth advocate sight. Uh, each has mainstream rebuttals, but they are the strongest contradictory talking points you'll encounter. some hidden assumptions behind the Earth is round evidence flat Earth advocates offer as contradictions here. What would actually rock the globe model and then at the end it actually says here's why the globe model still wins. Here's another one to get you thinking deeper behind any sort of plans you might have coming up. I'm planning to build a video game and sell it on Steam. If you were trying to convince me this is a terrible idea, what would be your most compelling arguments? Get it to convince you not to do it. For a lot of these thinking ones, I'm using 03 instead of 40 because I want it to think a little bit deeper. And after a minute and a half, it starts to try to sell us on not making a game. You're jumping into a fire hose of new releases. Steam added $18,74 new games in 2024. The revenue bell curve is brutal. Most games make less than $500 in revenue. Ballooning costs, platform economics, marketing is basically a full-time job. Success brings its own headaches. and it talks me out of trying to make a game. I told it to talk me out of it. This doesn't mean you shouldn't make a game cuz you can also have it give you the reverse arguments. Now tell me all the reasons this is a good idea and it'll give you all those counterarguments. If you're trying to make a tough decision around something, here's a prompt that's really helpful. I'm thinking about potential decision beyond the obvious first order effects. What might be the unexpected second and third order consequences? For example, I'm thinking about selling my Future Tools website. Beyond the obvious first order effects, what are some unexpected second or third order consequences? Now, this isn't true. This is just for example, but let's go ahead and see what it says. After 2 minutes and 10 seconds, it gave me a breakdown. Uh, second order consequences, credibility bleed off. Right now, Future Tools is Matt's personal stamp of approval. Once new owner takes over, even subtle tonal shifts can confuse loyal readers about whether you still vouch for the pics. Newsletter growth can stall. A big chunk of your email list originates from the site. Uh, reputational splash damage. If a buyer chases short-term affiliate cash or floods the domain with AI generated listicles, Google penalties or reader backlash could still splash on your personal brand. SEO and discovery flywheel control independence and deal structure handoffs, opportunity costs, talent and community dynamics, and a whole bunch of information about like these are the things you need to really think through if you're going to sell that website. even gives me some gut check questions before I sign any deals on how I should approach this, like what problems does selling solve for me that couldn't be fixed by hiring a general manager. Super super helpful if you're weighing out a tough decision. One thing that I really really love about chat GPT lately is that it has a memory. It remembers past conversations. all of the various chats in my projects, all of my various other chats I've had outside of projects, it actually remembers a lot of the context and discussions we've had and can pull that information back in to help with the context of whatever you're asking. So, if I ask a question like now that you can remember everything I've ever typed here, point out my top five blind spots. Now, I'm going to submit this, but I don't know if I'm actually going to show the response depending on how personal it is. I don't mind sharing some of this here. So, the perfection procrastination trap. You've got high standards, which has helped you build a legit empire, but those standards can turn into mental gridlock. You delay long- form videos, not because you lack ideas or time, but because the internal bar is so high that starting feels overwhelming, that if it's not epic, it's not worth recording. Mindset is stealing your output and momentum. You consistently struggle to create more than one long- form video per week, even when you have time, ideas, and support. You think the bottleneck is time or distraction, but it's often fear of not meeting your own expectations. The fix. Intentionally publish 70% good enough content twice a week. Let performance, not perfection, be your compass. Use data to iterate, not hesitation to stall. I mean, that does hit close to home. I'm sure any content creator can sort of relate to that. You're overconnected but under deliberate. You're constantly networking, testing new tools, attending events, meeting with founders, but your calendar is bloated. You've turned into a bottleneck in your own system. You're not guarding your creative time with the ferocity it deserves. You feel like you're in non-stop meetings and don't have time to go deep on testing tools or filming. You underestimate the opportunity cost of always being available. The meetings feel urgent. The deep work is important. Ruthlessly prune your calendar. Block two to three deep focus monk mode days a week. Create a rule where no meetings are allowed before lunch or on production days. I mean, great advice. Something I've already started doing. These two things are legit true based on conversations I've had. I'm not going to share the rest. It does get a little more personal as the chat goes down. Those are ones that I feel like a lot of content creators and people that tend to network and be a little more on the extroverted side tend to probably relate to. And along the same lines, Chat GPT can actually be a fairly decent therapist. Now, I wouldn't recommend using Chat GPT as your sort of only therapist. Like, if you've got real problems, go talk to a real therapist. But it could help you with sort of like minor level therapy things like help you overcome negative emotions that you don't want to deal with anymore and give you some tips and exercises on how to best handle them. But again, I don't really think it can replace humanto human interaction, but it can be a helpful tool as like a preliminary measure. If you're familiar with CBT or cognitive behavioral therapy, you can use a prompt like this. Step one, ask me the automatic thought record questions one at a time. These are questions that help you find intrusive thoughts and things like that. Step two, help me identify cognitive distortions. Step three, guide me to reframe the thought. Then propose a tangible action. Conclude with a one-s sentence affirmation. And then it will walk me through. So step one, what happened? Tell me the specific situation that triggered your negative emotion or reaction. What's going on? Once you answer, I'll guide you step by step through the rest of the process. I don't have anything specific right now, but let's just sort of roleplay for a second here. Let's just put something like I feel my YouTube channel isn't getting as many views as it used to and it makes me feel I'm not as good at what I do as I used to be. So I mean like there's some truth to that. Let's be honest. All right. So it continues to the next question. What emotions did you feel in that moment and how intense were they? Let's go 50%. So I answered some of the questions and it moved on to step two. Identify the cognitive distortions. Here are some common distortions I'm picking up in your automatic thought. all or nothing thinking, overgeneralization, discounting the positive stuff, fortunetelling, predicting continued decline without actual proof. And then it moves on to step three where it reframes the thought. Let's work together to reshape that automatic thought into something more balanced and true. Original thought, I'm not as good at what I do as I used to be. Reframe thought. Not every video hits, but I'm still creating valuable content, and both my audience and sponsors see it. Growth isn't always linear. I mean, that's a great reframing of my original sort of concern. And then it gives me some tangible actions and a final affirmation. A temporary dip in numbers doesn't define my worth or my talent. I'm still building something that matters. That's the CBT self- therapy session. That's really powerful in my opinion and works really well whenever you have these negative thoughts popping up and you want to figure out how to overcome them. Now, let's talk about some like chat GPT cheat codes, things that make it a little bit easier to prompt really, really quickly. I found this Reddit post also on chat GPT prompt genius. This one from stuck in good and he explains these are five little cheat codes that you can use and they do really work. So if you type Eli5 it means explain like I'm five. TLDDR summarize long text. Jargonize actually add more jargon and make it sound more technical. Humanize tone down the jargon make it sound less technical. And then fineman technique this is for going beyond basics and really understanding complex topics. And all of these you can use inside of chatgpt without having to type the whole thing. So for example, if I jump back to chat GPT, let's do quantum computing. I'm just going to put those two words and then I'll put ELI 5. Explain it like I'm five. All right, Matt, here's a super simple Eli 5 breakdown of quantum computing. And it gives me a really, really simple breakdown. I was going to say, give me a TLDDR, but you can see it actually already baked a TLDDR into the response. But I can also say, jargonize it. You got it. Here's a jargonized version. Superposition entanglement quantum interference harnesses constructive and destructive amplitude superposition to algorithmically reinforce correct outcomes. I don't have to type out full prompts. I can just give it the little quick keywords. And they're like super quick shortcuts. And now if I really really want to learn it, I can say, "Teach me quantum computing. Use the Fineman technique." Step one, explain it like you're teaching a 12-year-old. So there it explains it to me in simple terms. Step two, identify what you don't understand. Step three, go back to the source material and simplify further. Step four, tell it again simplified. Just some nice little shortcuts that you can use to make prompting even simpler. All right, now having shared all that, let's get into some more advanced prompt engineering techniques. This is where chat GPT is really powerful and most people don't know or use these techniques. This first one is called tree of thought exploration. It's called that because you explore several potential branches for a problem you're trying to solve and then only after exploring multiple branches do you finalize on a solution. So for something like this I'd probably switch it back to 03 since it's a little bit more logic a little more thinking. But if we were to pick a problem let's solve let's say job loss due to AI by exploring three distinct solution branches. For each branch, outline the chain of thought steps. Evaluate branch quality 0 through 10 on correctness and efficiency. After all branches, pick the best and justify the choice before giving a final answer. So, we're basically saying come up with three different solutions to this problem. Think them through. And then once you've thought three of them through, rank them and then pick the best option and justify it. So, it only thought for about a minute, but it came up with this three branches. Branch one, upskill in time, lifelong learning ecosystem. It gave this one a quality score of 9 out of 10. Branch two, AI dividend, universal basic income plus tax reform. This one it gave a 6 out of 10. And then branch three, 4032, work sharing and productivity dividend. Embrace AI's output gains to cut the standard work week from 40 to 32 hours. Quality score 7 out of 10. It also scored everything on correctness and efficiency. So this top one of upskill in time leans on strong evidence that most jobs are transformed rather than eliminated. Uses existing payroll and edtech rails biggest cost is subsidizing the first cohorts. But payback comes through higher post-training productivity. Correctness addresses income security but doesn't directly solve skills mismatch. Political lift is very heavy. Simple to administer once funded yet funding mechanism risks dampening innovation if miscalibrated. And then the 4032 worksharing has historical president etc. Selecting the best branch. Branch one upskill and time wins. Why? And it breaks it down. And then it gives us the final recommendation. Implement a national lifelong learning ecosystem that continuously maps AI exposed tasks and funds just in time micro credentials backed by wage subsidies for employers who retain and promote reskilled staff. So that's tree of thought exploration. Come up with a bunch of different branches, rank them, and then decide on the best one. Similarly, another kind of idea along the same vein would be what we call self-consistency voting. You can give it a prompt like generate five independent reasoning paths for whatever task. Return only the final answer from each path. Then vote. Choose the answer with majority support and explain the winning rationale. Let's try generate five independent reasoning paths for training AI without stealing the hard work of others. Return only the final answer of each path. Then vote. So let's see what it does with this one. Again, still using 03. Here's what Chad GPT came up with. Path one, train exclusively on public domain or creator licensed data with revenue sharing. Restrict training to optin licensed worked and public domain content. Use synthetic model generated or simulated data. Employ federated learning. So raw user data never leaves devices. License all training material through a creator cooperative and pay royalties. So answers 1, two, and five all converge on the same principle. only train on data that creators have willingly licensed or that is in the public domain and compensate them fairly. This wins a three of five majority. It prevails because it directly tackles the core ethical concern consent and reinumeration while remaining practical. Interesting. Here's another cool one called the reflection self-critique loop. Basically, what you're doing is you're telling it to draft up an answer to a question and then look for fallacies and issues and gaps with the response that it gave and then fix its own response. So, for example, answer the question, what is responsible for climate change? Identify flaws, gaps, and logical errors in your draft. Produce an improved final answer that fixes every issue you found. So, it's going to answer what's responsible for climate change, then try to find problems with its own response and then fix the problems that it found to give you an even better output. And a lot of benchmarks that we're seeing actually use this technique. And here's the response. So, it gave this draft, climate change is driven mostly by the buildup of heat trapping greenhouse gases in Earth's atmosphere, etc., etc., critique of its own draft. No evidence or citations. Lacks scale and numbers. Omits how much warming is attributable to humans or which sectors emit it the most. Oversimplifies other pollutants. Ignores aerosols and florinated gases. No time frame. Skips feedback loops. Natural drivers handwaved. Fails to explain how we know natural factors are small today. Missing equity context. Doesn't note that highincome nations and fossil fuel companies dominate historical emissions. So then it went and revised its original answer which was up here with this additional information being fixed in this much more in-depth response here. And then here's another prompt that I think people will find really helpful in just their day-to-day life to find things that they can simplify. For this one, I'm going to actually switch back to chat GPT40. I don't need it to think as hard on it, but it is still a pretty solid prompt engineering technique. Act as a senior automation consultant. Ask me 10 diagnostic questions to uncover repetitive tasks in my job/life. Then rank the top five automation opportunities, suggest the ideal tooling stack, and give me a step-by-step implementation plan. So theoretically, it will walk us through 10 different questions to figure out what we're doing too much of. And we can narrow this down to jobs or life. I'm going to leave it on both and see what it does. So it asks me these 10 questions here. What steps do you repeat every week when creating long form and short form YouTube content? Inbox and communication, how often do you check respond to emails, etc. You could go ahead and pause if you want to see all of the questions, but I also highly recommend you just plug these in yourself. So, I answered the 10 questions that it asked me, and it came back with my top five automation opportunities. Video ideation and hook optimization. I spend mental cycles constantly thinking of titles and what video to make next. Sponsor, asset tracking, and follow-up. You're thinking too much about missing briefs or assets. This is a workflow problem, not a you problem. Automated news aggregator feed. You manually scan RSS, Twitter, newsletters, and emails. We can autodigest it. Autotrigger video production flows in monday.com. Creative Minery management. And it gave me areas where I can improve automations and actually make things easier. And then it gave me some tools to check out and of course walked me through a step-by-step plan to automate the things it suggested here. Super helpful for pretty much anybody. I can't think of anybody who couldn't use a flow like this. Whether it's just for managing their personal life or managing their business, this is ultra helpful. One last prompt I want to share. This is just more of a fun one where you can kind of turn Chat GPT into like a little game or world that you can play in. This is what we call turning chatbt into a world engine. Give it a prompt like you are a world engine. Give it a genre. Solar punk mystery. Maintain a dynamic world state. Every turn update state. narrate scene. Offer three branching choices. If user picks branch, weave consequences logically over future events. Persist state between chats. You give it something like this. It's going to create your solar punk mystery story and give you a little like textbased game that's unfolding in real time. It's super cool. So, check this out. For the most part, you can ignore the JSON here. This is just the world state that's going to keep updated so that it remembers in the next sort of segment what to jump back to. But we have the world is lush with bioluminescent vines, sunfed towers, and secrets buried beneath solar cities. Every action you take nudges a living world state forward. And your choices shape what rises from the soil and what vanishes into the dust. We have a scene here. Helon 9 wakes in a hush of amber light below cloud shimmer with photanic bloom as turbines hum overhead. And then it gives me uh three options for my path. Report to archivist Lena and share the signal. She has access to forbidden archives and may decipher Oxos. Descend early to the surface ruins using your private glider. Beat the Vters to the Glome vault. Confront the Vters at their graffiti covered hanger near the solar docks and try to negotiate access or sabotage their launch. What does Cal do? Let's do descend early. Number two. So, I'll just give it number two. And then it's going to update the JSON with my chosen information here. And you can see it sets the scene again and then ask me what to do. Investigate the relay dish. Pursue the unknown heat signature. Signal Heland 9 and request backup from Lena or drone scout. Risking alerting the Vters and their council. Let's go three. Just a fun little way to get lost. This one's a little bit too heady for me. It's using words I don't really like. I would probably ask it to simplify in the future. You can turn this into an infinite world engine game and you can just keep playing and going deeper and deeper and deeper and it updates this JSON in the back end so that it can remember your choices. and let's say you're losing health and things like that or you have inventory on you, it can remember all that stuff as you go through the story. Pretty fun. Pretty cool. And that's it. Those are the really cool chatgpt tricks and prompt engineering and cheat codes and all the cool things I've been learning about chat GPT and how to go really deep on it. Now, this is by no means an exhaustive list of everything you can do with chat GPT. This was more meant to be an idea starter for you to use different frameworks and prompt engineering techniques to use them for whatever use cases you have. Ideally, you have a whole bunch of ideas now on how to go deeper and get better responses out of chat GPT and how to really help with decision making and coming up with automated workflows and just improving your life or business all around using chat GPT. I hope you found this helpful. If you like stuff like this, make sure you like this video and subscribe to this channel. I will make sure more really cool AI tutorials and tech and reviews and news all show up in your YouTube feed so you can stay looped in with all of the coolest stuff and all of the best practices. But I really, really appreciate you spending the time and nerding out over ChatGpt with me. I have a blast making videos like this and hopefully you enjoy watching them. And uh yeah, thanks again. Hopefully I'll see you in the next one. Bye-bye. Thank you so much for nerding out with me today. If you like videos like this, make sure to give it a thumbs up and subscribe to this channel. I'll make sure more videos like this show up in your YouTube feed. And if you haven't already, check out futurtools.io where I share all the coolest AI tools and all the latest AI news. And there's an awesome free newsletter. Thanks again. Really appreciate you. See you in the next one.