WEBVTT 00:00:00.000 --> 00:00:03.700 Have you ever felt that quiet hum of acceleration, 00:00:03.960 --> 00:00:06.740 like AI shifting gears, moving really fast, and 00:00:06.740 --> 00:00:08.339 you're kind of trying to keep pace? Maybe you're 00:00:08.339 --> 00:00:10.960 using tools like ChatGPT, but there's that lingering 00:00:10.960 --> 00:00:12.960 thought, am I really getting the most out of 00:00:12.960 --> 00:00:16.420 this thing? That subtle sense of being just a 00:00:16.420 --> 00:00:19.339 little overwhelmed, maybe, navigating this powerful 00:00:19.339 --> 00:00:21.379 new landscape. Oh, absolutely. It's such a common 00:00:21.379 --> 00:00:24.039 feeling, isn't it? But here's the, well... the 00:00:24.039 --> 00:00:26.339 interesting paradox. While the sheer scope of 00:00:26.339 --> 00:00:29.399 AI can feel complex, unlocking its true potential, 00:00:29.800 --> 00:00:31.899 especially with something like Chad GPT, it actually 00:00:31.899 --> 00:00:34.640 comes down to surprisingly simple, smarter interactions. 00:00:34.880 --> 00:00:37.039 We're talking about a focused set of insights, 00:00:37.479 --> 00:00:40.179 like 35 tips maybe, that can genuinely transform 00:00:40.179 --> 00:00:42.259 how you use it day to day. Think of it like becoming 00:00:42.259 --> 00:00:45.359 the seasoned pilot of your own AI co -pilot. 00:00:45.479 --> 00:00:47.840 Welcome back to the Deep Dive. Today, our mission 00:00:47.840 --> 00:00:49.740 is pretty clear. We're not just aiming to use 00:00:49.740 --> 00:00:52.460 Chat GPT. We want to help you master it, to truly 00:00:52.460 --> 00:00:54.920 make it a personalized extension of your own 00:00:54.920 --> 00:00:57.880 thinking, a co -pilot that understands your specific 00:00:57.880 --> 00:01:00.619 needs. We've taken a deep dive into the practical 00:01:00.619 --> 00:01:02.920 wisdom that bridges that gap from, you know, 00:01:03.039 --> 00:01:05.739 novice to AI power user, helping you really command 00:01:05.739 --> 00:01:08.719 this tool. And we've charted a very deliberate 00:01:08.719 --> 00:01:11.439 path for you today. First, we'll navigate the 00:01:11.439 --> 00:01:13.540 different access plans, help you find your best 00:01:13.540 --> 00:01:15.900 starting point, then we'll get into fine tuning 00:01:15.900 --> 00:01:18.900 your account settings for maximum privacy and 00:01:18.900 --> 00:01:21.780 effectiveness. Super important. Next up, it's 00:01:21.780 --> 00:01:24.620 about picking the right AI model for each job 00:01:24.620 --> 00:01:27.939 and strategically organizing your work with this 00:01:27.939 --> 00:01:30.420 really powerful feature called Projects. It's 00:01:30.420 --> 00:01:33.200 kind of less known, but amazing. After that, 00:01:33.319 --> 00:01:35.000 we'll unpack the art of prompting how to talk 00:01:35.000 --> 00:01:37.340 to it effectively and explore some advanced conversation 00:01:37.340 --> 00:01:39.859 strategies. And finally, we'll cover working 00:01:39.859 --> 00:01:42.319 with files, how to verify information, which 00:01:42.319 --> 00:01:44.819 is crucial, and even getting the AI to help you 00:01:44.819 --> 00:01:47.060 write better prompts for itself. This whole deep 00:01:47.060 --> 00:01:49.060 dive, it's really about gaining control, building 00:01:49.060 --> 00:01:51.879 confidence, and ultimately just amplifying what 00:01:51.879 --> 00:01:54.549 you can do. OK, so we've acknowledged that initial 00:01:54.549 --> 00:01:57.290 feeling, maybe being a bit overwhelmed. Let's 00:01:57.290 --> 00:01:59.349 tackle one of the first questions people usually 00:01:59.349 --> 00:02:01.810 have with chat GPT. Do I really need to pay for 00:02:01.810 --> 00:02:05.290 it? Or is the free tier enough? Yeah, that's 00:02:05.290 --> 00:02:08.069 a really practical question. And the answer is, 00:02:08.090 --> 00:02:10.530 well, it depends, because OpenAI offers these 00:02:10.530 --> 00:02:13.810 distinct plans. The free plan is obviously zero 00:02:13.810 --> 00:02:16.530 cost. It's great for just trying things out for 00:02:16.530 --> 00:02:19.270 students, anyone just dipping their toes in. 00:02:19.550 --> 00:02:21.629 You usually get access to their latest modeled 00:02:21.740 --> 00:02:24.300 let's call it GPT -5, thinking ahead, but with 00:02:24.300 --> 00:02:26.800 a message limit. You get basic image generation, 00:02:27.139 --> 00:02:30.560 but no sort of video stuff. So for quick emails, 00:02:30.939 --> 00:02:33.520 simple questions, short summaries, it's a solid 00:02:33.520 --> 00:02:35.479 start. Right. So it's like an entry point, but 00:02:35.479 --> 00:02:37.620 with some guardrails maybe. You got it. Then 00:02:37.620 --> 00:02:40.400 there's the plus plan. That's $20 a month. This 00:02:40.400 --> 00:02:42.740 is often the sweet spot for professionals, content 00:02:42.740 --> 00:02:45.099 creators, people using it a lot. You get way 00:02:45.099 --> 00:02:47.639 higher message limits, broader access to advanced 00:02:47.639 --> 00:02:50.340 reasoning, which is key for complex stuff, data 00:02:50.340 --> 00:02:53.319 analysis, planning. Plus, you get better, faster 00:02:53.319 --> 00:02:56.020 image generation. Access to Sora for making short 00:02:56.020 --> 00:02:58.360 videos from text, this thing called Canvas for 00:02:58.360 --> 00:03:01.139 editing together, and agent mode for multi -step 00:03:01.139 --> 00:03:03.080 tasks. If you're hitting those free limits or 00:03:03.080 --> 00:03:05.020 need those advanced features, this is probably 00:03:05.020 --> 00:03:07.979 the one. Hmm, okay. That sounds like a pretty 00:03:07.979 --> 00:03:10.680 significant jump in capability, especially if 00:03:10.680 --> 00:03:13.000 you're integrating AI into your daily work. It 00:03:13.000 --> 00:03:16.479 really is, yeah. And then for most users, the 00:03:16.479 --> 00:03:19.400 top tier is the Pro Plan, 200 bucks a month. 00:03:19.740 --> 00:03:21.939 Now, this is really designed for small businesses, 00:03:22.240 --> 00:03:25.240 researchers. anyone where AI is absolutely central 00:03:25.240 --> 00:03:28.020 to what they do, heavy workloads. You're talking 00:03:28.020 --> 00:03:31.319 basically unlimited GPT -5 use, the best reasoning 00:03:31.319 --> 00:03:34.240 without caps, unlimited and faster high quality 00:03:34.240 --> 00:03:37.520 images, longer Sora videos, priority access, 00:03:38.199 --> 00:03:41.560 the works. For when AI isn't just helpful, it's 00:03:41.560 --> 00:03:43.680 like indispensable to how you operate. OK, so 00:03:43.680 --> 00:03:46.080 with those options laid out, if someone's just 00:03:46.080 --> 00:03:48.419 starting brand new to this, what's the most sensible 00:03:48.419 --> 00:03:50.919 first step? Definitely start with the free plan. 00:03:51.129 --> 00:03:53.110 See how you go. If you start hitting the message 00:03:53.110 --> 00:03:55.490 limits or needing more power, then upgrade to 00:03:55.490 --> 00:03:57.729 plus. Sensible way in. OK, that makes sense. 00:03:58.009 --> 00:04:00.050 So we've covered the plans. Now let's get into 00:04:00.050 --> 00:04:02.430 making chat GPT feel less like a generic tool 00:04:02.430 --> 00:04:04.750 and more like your personal assistant. Customizing 00:04:04.750 --> 00:04:06.629 the account settings, this is where it gets really 00:04:06.629 --> 00:04:09.050 interesting, right? Absolutely. This is crucial. 00:04:09.210 --> 00:04:12.490 First up, protect your privacy. Seriously. Think 00:04:12.490 --> 00:04:14.889 of this as your personal data firewall. In data 00:04:14.889 --> 00:04:18.009 controls, turn off. Improve the model for everyone. 00:04:18.470 --> 00:04:20.470 You're telling OpenAI, hey, my chats are private. 00:04:20.569 --> 00:04:23.529 Don't use them to train your AI. Imagine accidentally 00:04:23.529 --> 00:04:25.910 feeding your company secrets into a global model. 00:04:26.430 --> 00:04:29.490 Yeah, not good. This click prevents that. It's 00:04:29.490 --> 00:04:32.449 about owning your data. And there's also temporary 00:04:32.449 --> 00:04:35.230 chat. It's basically incognito mode. Chats aren't 00:04:35.230 --> 00:04:37.370 saved, aren't used for training. Perfect for 00:04:37.370 --> 00:04:39.810 sensitive stuff. And importantly, you can always 00:04:39.810 --> 00:04:42.470 export or delete your data. You're in control. 00:04:42.790 --> 00:04:44.670 That level of control really does offer some 00:04:44.670 --> 00:04:46.870 peace of mind, especially if you're handling 00:04:46.870 --> 00:04:50.930 confidential info. Exactly. OK, next, under personalization, 00:04:51.490 --> 00:04:54.319 you'll find Customize your experience. This is 00:04:54.319 --> 00:04:57.579 super impactful. You tell the AI about you, your 00:04:57.579 --> 00:04:59.639 name, your job. Like, are you in marketing or 00:04:59.639 --> 00:05:02.000 are you a programmer? It'll tailor answers. You 00:05:02.000 --> 00:05:04.199 can set your preferred response style, short 00:05:04.199 --> 00:05:08.040 bullets, detailed explanations. You choose. And 00:05:08.040 --> 00:05:10.160 you can give it specific rules or formats to 00:05:10.160 --> 00:05:12.879 always follow, like that example of Min, the 00:05:12.879 --> 00:05:15.970 student. They might say, call me Min. Use startup 00:05:15.970 --> 00:05:18.170 examples for academic stuff. Always summarize 00:05:18.170 --> 00:05:20.850 in bullets. Bold the main ideas. Ah, so you're 00:05:20.850 --> 00:05:23.050 basically programming its baseline understanding 00:05:23.050 --> 00:05:25.230 of how you like to communicate and what you need. 00:05:25.370 --> 00:05:27.490 Couldn't have said it better myself. Yeah. Then 00:05:27.490 --> 00:05:29.730 enable advanced features. Go into settings, turn 00:05:29.730 --> 00:05:32.170 on web search. This is vital. Otherwise, its 00:05:32.170 --> 00:05:33.850 knowledge is stuck in the past before its last 00:05:33.850 --> 00:05:35.649 training cutoff. It won't know current events. 00:05:36.129 --> 00:05:38.470 Advanced voice lets you have natural spoken chats. 00:05:38.870 --> 00:05:41.009 Canvas is like a collaborative doc with the AI 00:05:41.009 --> 00:05:43.550 helping you edit. And Code interpreter. Don't 00:05:43.550 --> 00:05:45.410 let the name fool you. Let's it analyze data, 00:05:45.569 --> 00:05:48.529 do calculations, run code snippets. Super powerful. 00:05:48.990 --> 00:05:51.370 I find that code interpreter amazing. It's not 00:05:51.370 --> 00:05:53.790 just for coders, is it? It's like having a mini 00:05:53.790 --> 00:05:57.170 data scientist on call. Precisely. And finally, 00:05:57.629 --> 00:06:00.449 set up memory and personalization. If you enable 00:06:00.449 --> 00:06:03.529 referenced save memories and referenced chat 00:06:03.529 --> 00:06:06.949 history, chat GPT starts to learn from your chats. 00:06:07.529 --> 00:06:10.410 It remembers things like Oh, you prefer a humorous 00:06:10.410 --> 00:06:12.550 tone for social media posts, so you always ask 00:06:12.550 --> 00:06:14.910 for sources. And you can manage these memories, 00:06:15.209 --> 00:06:17.310 delete ones you don't want, so it only remembers 00:06:17.310 --> 00:06:20.129 what's helpful. That sounds great, but honestly, 00:06:20.370 --> 00:06:23.709 does that AI memory ever get in the way? Like, 00:06:23.709 --> 00:06:26.209 does it sometimes make it too rigid or lead it 00:06:26.209 --> 00:06:28.149 down the wrong path because it remembered something 00:06:28.149 --> 00:06:30.779 incorrectly? It's a really sharp question. Yeah. 00:06:30.939 --> 00:06:32.699 Yeah, it can't happen, though it's mostly helpful. 00:06:33.079 --> 00:06:34.639 An old preference might occasionally lead to 00:06:34.639 --> 00:06:37.220 a less than ideal answer in a new context. That's 00:06:37.220 --> 00:06:39.019 exactly why being able to manage those memories 00:06:39.019 --> 00:06:40.819 is key. You can just delete one if it's causing 00:06:40.819 --> 00:06:43.480 issues. It reminds us AI is intuitive, but definitely 00:06:43.480 --> 00:06:46.220 not perfect. But overall, it means way less repetitive 00:06:46.220 --> 00:06:48.500 instruction makes it feel much more like it just 00:06:48.500 --> 00:06:50.899 gets you. Right, less repeating yourself over 00:06:50.899 --> 00:06:53.860 and over. OK, so the account is tuned, privacy 00:06:53.860 --> 00:06:55.620 is locked down, it's starting to feel personal. 00:06:55.879 --> 00:06:58.120 Now, let's talk about picking the right brain 00:06:58.120 --> 00:07:00.860 for the job and, crucially, how to keep things 00:07:00.860 --> 00:07:03.100 organized. Right. Often you've got a choice of 00:07:03.100 --> 00:07:06.420 models. For instant or auto tasks, think quick, 00:07:06.740 --> 00:07:09.740 simple stuff. What's the capital of France? Translate 00:07:09.740 --> 00:07:12.060 this sentence. These don't need deep thought. 00:07:12.399 --> 00:07:16.019 But for anything needing more cognitive horsepower, 00:07:16.600 --> 00:07:18.980 complex logic, math problems, detailed planning, 00:07:19.139 --> 00:07:22.240 you want the thinking models. These models engage 00:07:22.240 --> 00:07:24.399 in a much more deliberate reasoning process. 00:07:24.819 --> 00:07:26.740 You get more robust, accurate answers when the 00:07:26.740 --> 00:07:29.379 task is genuinely tricky. You might also see 00:07:29.379 --> 00:07:32.000 legacy models in the settings, maybe for specific 00:07:32.000 --> 00:07:34.180 older tasks they were good at. Okay, so simple 00:07:34.180 --> 00:07:36.839 facts are instant, but mapping out a whole marketing 00:07:36.839 --> 00:07:39.439 strategy needs a thinking model. It's exactly. 00:07:39.939 --> 00:07:42.540 Summarizing the plot of, say, the movie Inception, 00:07:43.060 --> 00:07:46.060 instant auto is fine, but asking for a detailed 00:07:46.060 --> 00:07:48.959 five -year marketing plan for a new coffee shop 00:07:48.959 --> 00:07:51.899 in Ho Chi Minh City? With budgets and audience 00:07:51.899 --> 00:07:55.100 segments, that absolutely needs a thinking model. 00:07:55.579 --> 00:07:58.220 Now, for organization, let me introduce projects. 00:07:58.459 --> 00:08:00.300 Honestly, one of the most useful features and 00:08:00.300 --> 00:08:02.379 so many people don't seem to use it. Think of 00:08:02.379 --> 00:08:05.079 projects like creating dedicated specialized 00:08:05.079 --> 00:08:07.879 AI assistants. Each one is custom built for a 00:08:07.879 --> 00:08:11.149 specific workflow or topic. dedicated AI assistants 00:08:11.149 --> 00:08:14.230 for each project. That sounds incredibly powerful. 00:08:14.410 --> 00:08:16.610 It really is. Projects let you group related 00:08:16.610 --> 00:08:18.769 chats together. You can store relevant files, 00:08:19.069 --> 00:08:20.930 PDFs, images, notes right inside that project. 00:08:21.329 --> 00:08:23.569 The AI then uses only those files as its immediate 00:08:23.569 --> 00:08:25.850 knowledge base for that project. And crucially, 00:08:26.209 --> 00:08:28.250 each project can have its own custom instructions, 00:08:28.449 --> 00:08:30.170 only separate from your main account settings. 00:08:30.589 --> 00:08:32.889 And here's the real game changer. It creates 00:08:32.889 --> 00:08:35.649 project only memory. What you discuss in one 00:08:35.649 --> 00:08:37.649 project, the preferences it learns there. They 00:08:37.649 --> 00:08:39.730 don't leak into other projects. No more crossed 00:08:39.730 --> 00:08:43.529 wires. Whoa. OK, hang on. Imagine having a separate 00:08:43.529 --> 00:08:46.409 AI expert for every single project you're working 00:08:46.409 --> 00:08:48.389 on, each one with its own specific knowledge 00:08:48.389 --> 00:08:51.070 base and instructions. That's like Maso building 00:08:51.070 --> 00:08:53.490 a whole team of specialist consultants. It's 00:08:53.490 --> 00:08:55.529 honestly that transformative. I remember when 00:08:55.529 --> 00:08:57.629 I first started using them, I was juggling like 00:08:57.629 --> 00:08:59.789 three totally different client projects. One 00:08:59.789 --> 00:09:02.629 needed this really quirky and formal social media 00:09:02.629 --> 00:09:05.590 voice. Another needed super precise technical 00:09:05.590 --> 00:09:07.870 language for finance. The third was academic. 00:09:08.090 --> 00:09:10.970 Before projects, ChatGPT kept messing them up. 00:09:11.190 --> 00:09:13.470 It would try to inject jokes into the financial 00:09:13.470 --> 00:09:15.850 report. Projects completely sorted that out. 00:09:16.110 --> 00:09:18.700 Saved my sanity, frankly. So you can have a YouTube 00:09:18.700 --> 00:09:21.179 content assistant project, upload your brand 00:09:21.179 --> 00:09:23.700 guidelines, successful video scripts, audience 00:09:23.700 --> 00:09:26.279 data, instruct it. You're a YouTube strategist, 00:09:26.419 --> 00:09:28.840 give me three catchy titles, use an energetic 00:09:28.840 --> 00:09:32.200 tone for young viewers, or maybe a university 00:09:32.200 --> 00:09:35.220 academic advisor project, feed it syllabi, notes, 00:09:35.740 --> 00:09:38.200 sample essays, tell it, you're my tutor, explain 00:09:38.200 --> 00:09:40.559 microeconomics simply, use examples relevant 00:09:40.559 --> 00:09:43.000 to my course. You can even do a weight loss coach 00:09:43.000 --> 00:09:44.919 project with your food logs and workout plan. 00:09:45.279 --> 00:09:47.889 The key pro tip. Always enable project -only 00:09:47.889 --> 00:09:50.889 memory for that laser focus. So projects basically 00:09:50.889 --> 00:09:52.990 puts up walls between different topics, stops 00:09:52.990 --> 00:09:55.110 the AI getting confused when you switch contexts. 00:09:55.529 --> 00:09:58.009 Exactly. It ensures that focused expertise for 00:09:58.009 --> 00:10:02.070 each specific task. No more knowledge contamination. 00:10:02.549 --> 00:10:05.370 Mid -roll sponsor, read placeholder. Okay, we've 00:10:05.370 --> 00:10:07.110 set up our environment. We've got the specialized 00:10:07.110 --> 00:10:09.809 project assistance ready. Now for the main event. 00:10:10.039 --> 00:10:13.179 how we actually talk to the AI. Because the quality 00:10:13.179 --> 00:10:15.399 of the answers really does depend so much on 00:10:15.399 --> 00:10:17.440 the quality of our questions, doesn't it? Spot 00:10:17.440 --> 00:10:20.120 on. First rule. Talk like a human. Seriously. 00:10:20.460 --> 00:10:23.039 Forget weird syntax or keyword jamming. Just 00:10:23.039 --> 00:10:24.759 chat like you would with a really smart colleague 00:10:24.759 --> 00:10:27.240 instead of something robotic like, request provision 00:10:27.240 --> 00:10:29.740 of Vietnamese culinary recipe optimized for under 00:10:29.740 --> 00:10:32.679 30 minute prep cook time. To diners. Just say, 00:10:32.899 --> 00:10:34.799 hey, I want to cook a Vietnamese dish for two 00:10:34.799 --> 00:10:36.899 tonight, but I'm short on time. Can you suggest 00:10:36.899 --> 00:10:38.659 something tasty that's easy to make in under 00:10:38.659 --> 00:10:42.399 30 minutes? Need the recipe too. See? Natural 00:10:42.399 --> 00:10:44.500 to clear. Yeah, it feels more like starting a 00:10:44.500 --> 00:10:48.279 conversation than issuing a command. Next up, 00:10:48.620 --> 00:10:52.039 be direct and specific. If you're vague, you 00:10:52.039 --> 00:10:55.259 get vague, generic answers. Tell me about writing 00:10:55.259 --> 00:10:57.580 a CV is just too broad. It doesn't know where 00:10:57.580 --> 00:11:00.950 to start. Better. How do I write a CV for a junior 00:11:00.950 --> 00:11:03.190 marketing exec role? I don't have formal work 00:11:03.190 --> 00:11:05.450 experience, just university clubs. What sections 00:11:05.450 --> 00:11:07.090 should I include? How do I highlight the club 00:11:07.090 --> 00:11:09.950 stuff? Much better. So the more precision you 00:11:09.950 --> 00:11:11.950 give it, the more tailored the result you get 00:11:11.950 --> 00:11:14.649 back. Precisely. Also really important, provide 00:11:14.649 --> 00:11:17.590 rich context. Think about the classic five W's. 00:11:17.710 --> 00:11:20.549 Who, what, when, where, why. The more background 00:11:20.549 --> 00:11:22.750 you give it, the more accurate and genuinely 00:11:22.750 --> 00:11:25.429 useful the answer will be. For example, act as 00:11:25.429 --> 00:11:28.360 my personal finance advisor. I'm 25, freelance, 00:11:28.559 --> 00:11:31.299 my income's unstable. Average is maybe $650 a 00:11:31.299 --> 00:11:33.960 month. Goal is saving for a house deposit in 00:11:33.960 --> 00:11:36.419 10 years. Can you suggest a simple monthly budget? 00:11:36.600 --> 00:11:38.379 My living costs in Ho Chi Minh City are about 00:11:38.379 --> 00:11:41.360 $430. See how much info that gives it to work 00:11:41.360 --> 00:11:43.379 with. That's incredibly detailed. I can see how 00:11:43.379 --> 00:11:45.659 powerful that is. But maybe for someone new, 00:11:45.679 --> 00:11:47.580 could that feel a bit much? Like, what's the 00:11:47.580 --> 00:11:49.559 absolute most crucial piece of context to start 00:11:49.559 --> 00:11:51.419 with if you're overwhelmed? That's a fair point. 00:11:51.820 --> 00:11:54.259 If you had to pick one, I'd say the why, your 00:11:54.259 --> 00:11:55.899 ultimate goal, or the problem you're trying to 00:11:55.899 --> 00:11:59.220 solve. Start there. Then layer in the who, your 00:11:59.220 --> 00:12:01.740 roller perspective, and the what's, the specific 00:12:01.740 --> 00:12:04.870 task. Also, you absolutely need to specify your 00:12:04.870 --> 00:12:07.470 desired format. Tell it how you want the output. 00:12:07.889 --> 00:12:11.610 A table, JSON, a bullet point, a checklist. This 00:12:11.610 --> 00:12:13.450 makes sure the info comes back in a way you can 00:12:13.450 --> 00:12:15.750 actually use easily. For that finance example, 00:12:15.750 --> 00:12:18.110 you could add, present the plan in JSON format 00:12:18.110 --> 00:12:20.850 with fields for income, savings, expenses. Great. 00:12:20.850 --> 00:12:22.990 So it's not just what info, but how you want 00:12:22.990 --> 00:12:25.049 it structured. You got it. And here's a really 00:12:25.049 --> 00:12:27.769 powerful technique. Give examples, good and bad. 00:12:28.250 --> 00:12:30.289 Show the AI what you want it to do, and just 00:12:30.289 --> 00:12:33.059 as importantly, what not to do. Like, if you 00:12:33.059 --> 00:12:35.419 want social media hooks for sunscreen, follow 00:12:35.419 --> 00:12:38.059 this good style. 90 % of people apply sunscreen 00:12:38.059 --> 00:12:40.120 wrong. Here's why. That makes people curious. 00:12:40.580 --> 00:12:43.220 Avoid this BD style. Sunscreen is important for 00:12:43.220 --> 00:12:46.059 skin. Too generic. Hetcom. This directly steers 00:12:46.059 --> 00:12:47.860 it towards the kind of output you're after. You 00:12:47.860 --> 00:12:49.860 know, I still wrestle with prompt drift myself 00:12:49.860 --> 00:12:51.940 sometimes, even after using these tools for a 00:12:51.940 --> 00:12:53.820 while. It really is an ongoing conversation, 00:12:53.899 --> 00:12:56.600 isn't it? Sometimes it just wanders off. It absolutely 00:12:56.600 --> 00:12:59.799 is. And providing those clear examples, good 00:12:59.799 --> 00:13:02.860 and bad, is one of the best ways to keep it anchored, 00:13:03.179 --> 00:13:06.059 especially for creative or nuanced tasks. So 00:13:06.059 --> 00:13:08.720 yeah, it boils down to treating it like a very 00:13:08.720 --> 00:13:11.940 smart but very literal dialogue partner. Be clear, 00:13:12.299 --> 00:13:14.799 be specific, give a context. Exactly, a dynamic 00:13:14.799 --> 00:13:17.629 partnership. Clarity is definitely king. Okay, 00:13:17.750 --> 00:13:19.450 let's take these interactions up another notch. 00:13:19.610 --> 00:13:21.690 Let's talk advanced conversation strategies and 00:13:21.690 --> 00:13:24.529 also how we handle information reliability, especially, 00:13:24.529 --> 00:13:27.309 you know, checking facts. Right. So, for advanced 00:13:27.309 --> 00:13:29.690 conversation strategies, number one, don't accept 00:13:29.690 --> 00:13:33.009 the first draft. Ever. Chat GPT shines when you 00:13:33.009 --> 00:13:35.710 iterate. Always push back. Rewrite this more 00:13:35.710 --> 00:13:37.990 concisely. Make the tone more professional. Add 00:13:37.990 --> 00:13:40.029 some humor. Can you explain that with a simpler 00:13:40.029 --> 00:13:43.240 analogy? Then, use the branching feature. This 00:13:43.240 --> 00:13:45.940 is cool. After a response, find the little three 00:13:45.940 --> 00:13:48.860 dot icon, click branch in new chat. It lets you 00:13:48.860 --> 00:13:51.580 explore different paths from the same point without 00:13:51.580 --> 00:13:54.159 messing up your main chat. Like draft a job email, 00:13:54.519 --> 00:13:56.899 then branch it one friendly, one more assertive, 00:13:57.139 --> 00:13:59.000 see which works better. Oh, that's brilliant. 00:13:59.179 --> 00:14:01.279 Like A, B testing different approaches without 00:14:01.279 --> 00:14:04.139 losing your original thread. Exactly. Also for 00:14:04.139 --> 00:14:07.299 big tasks. Chunk big projects into steps. Don't 00:14:07.299 --> 00:14:09.720 ask for a whole business plan in one go. That's 00:14:09.720 --> 00:14:12.159 overwhelming for you and the AI. Break it down, 00:14:12.279 --> 00:14:16.500 part one. Executive summary, part two. Market 00:14:16.500 --> 00:14:19.799 analysis, part three. Marketing, part four. Financials. 00:14:19.960 --> 00:14:22.299 Tackle each piece separately. And this is crucial 00:14:22.299 --> 00:14:25.539 for alignment. Ask ChatGPT to state its assumptions. 00:14:25.759 --> 00:14:27.580 Before it jumps into planning your five -day 00:14:27.580 --> 00:14:30.120 New York trip, ask it. OK, before you start, 00:14:30.460 --> 00:14:32.299 what are you assuming about my budget, interests, 00:14:32.500 --> 00:14:34.679 and how I plan to get around? This gets you both 00:14:34.679 --> 00:14:37.500 on the same page. Finally, try this. Let ChadGBT 00:14:37.500 --> 00:14:39.559 ask you questions. If you want to help improving 00:14:39.559 --> 00:14:41.340 time management, instead of asking for tips, 00:14:41.620 --> 00:14:44.059 say, ask me five diagnostic questions about my 00:14:44.059 --> 00:14:46.059 current time management challenges. Then based 00:14:46.059 --> 00:14:48.179 on your answers, it can suggest a tailored strategy. 00:14:48.360 --> 00:14:50.399 Oh, that flips the script. It becomes a diagnostic 00:14:50.399 --> 00:14:52.980 partner first, then a solution provider. Very 00:14:52.980 --> 00:14:55.789 smart. Yeah. Now, shifting to working with files 00:14:55.789 --> 00:14:58.289 and media, the big one. Verify facts and check 00:14:58.289 --> 00:15:01.230 sources. AI can hallucinate. It can make stuff 00:15:01.230 --> 00:15:04.309 up, but sound totally confident. Always ask for 00:15:04.309 --> 00:15:06.529 sources. Click the links. Cross reference with 00:15:06.529 --> 00:15:08.710 other reliable sources. Be extra skeptical about 00:15:08.710 --> 00:15:11.669 recent events or sensitive topics where accuracy 00:15:11.669 --> 00:15:14.389 is paramount. Right. So rule one for critical 00:15:14.389 --> 00:15:17.669 info. Never just trust the AI output blindly. 00:15:18.350 --> 00:15:21.169 Always double check. Always. However, you can 00:15:21.169 --> 00:15:24.370 upload and analyze documents. This is incredibly 00:15:24.370 --> 00:15:27.370 useful. Upload PDFs, research papers, reports, 00:15:27.789 --> 00:15:30.490 images, charts, screenshots, text files, data, 00:15:30.509 --> 00:15:33.350 notes. You can ask it to, say, summarize a Q3 00:15:33.350 --> 00:15:36.190 financial report PDF, identify key trends or 00:15:36.190 --> 00:15:38.429 risks mentioned, and even cite the page numbers 00:15:38.429 --> 00:15:41.269 for its claims. You can also use Canvas for collaborative 00:15:41.269 --> 00:15:43.370 editing. Remember that. It puts an editable document 00:15:43.370 --> 00:15:45.289 right next to the chat. You can write something, 00:15:45.409 --> 00:15:47.590 edit it directly, and get AI suggestions right 00:15:47.590 --> 00:15:49.590 there. That really makes it feel like a shared 00:15:49.590 --> 00:15:52.559 workspace, not just a Q &A box. Totally, and 00:15:52.559 --> 00:15:55.100 this extends to image and visual capabilities. 00:15:55.879 --> 00:15:59.039 ChatGBT can interpret images now, upload a photo, 00:15:59.700 --> 00:16:02.240 ask it to count things with a pinch of salt on 00:16:02.240 --> 00:16:04.860 accuracy, use OCR optical character recognition 00:16:04.860 --> 00:16:07.340 to pull text out of a screenshot, describe a 00:16:07.340 --> 00:16:10.279 photo, or extract data from a chart. Pro tip, 00:16:10.759 --> 00:16:12.940 perform OCR on this image and give me the text. 00:16:13.419 --> 00:16:16.960 Super handy, and just a hammer at home. For any 00:16:16.960 --> 00:16:20.000 sensitive discussion, visual or text, use temporary 00:16:20.000 --> 00:16:22.840 chat for privacy. Not saved, not trained on, 00:16:22.960 --> 00:16:25.259 not stored in memory. OK. So the habit we're 00:16:25.259 --> 00:16:27.960 building is this constant loop. Interact, refine, 00:16:28.179 --> 00:16:30.519 ask questions, verify, iterate again. Absolutely. 00:16:30.519 --> 00:16:32.899 It's a dynamic partnership. You need to be engaged, 00:16:32.980 --> 00:16:35.000 not just a passive recipient. It's not a one 00:16:35.000 --> 00:16:37.120 -way street. All right. Now for some really advanced 00:16:37.120 --> 00:16:39.860 moves. Turning the AI into a prompt writing assistant 00:16:39.860 --> 00:16:41.620 for itself and looking at some ready -to -go 00:16:41.620 --> 00:16:43.960 templates to save time. Yeah. This is where you 00:16:43.960 --> 00:16:47.519 really start leveraging AI smartly. Meta -prompting. 00:16:47.720 --> 00:16:51.279 Using AI to improve your prompts for AI. It sounds 00:16:51.279 --> 00:16:54.460 complex, but it's simple. Just ask ChatGPT to 00:16:54.460 --> 00:16:57.200 write better prompts for you. For example, I 00:16:57.200 --> 00:16:59.539 want to brainstorm ideas for my Instagram channel 00:16:59.539 --> 00:17:02.279 on vegan cooking. Can you suggest five detailed, 00:17:02.700 --> 00:17:04.660 effective prompts I could use to get post ideas, 00:17:05.220 --> 00:17:07.339 recipe outlines, and short video concepts from 00:17:07.339 --> 00:17:09.599 you? It will then give you well -structured prompts 00:17:09.599 --> 00:17:12.220 you can just copy and paste. So you're basically 00:17:12.220 --> 00:17:14.119 asking the AI, how should I ask you questions 00:17:14.119 --> 00:17:16.440 better? Precisely. It's surprisingly effective. 00:17:16.700 --> 00:17:18.839 And then there are ready -to -use prompt templates. 00:17:19.259 --> 00:17:21.759 These are huge time savers. For product comparison, 00:17:21.839 --> 00:17:23.799 you can set up a template. Create a detailed 00:17:23.799 --> 00:17:26.099 comparison table for product A versus product 00:17:26.099 --> 00:17:29.839 B. Include criteria, price, key features, ideal 00:17:29.839 --> 00:17:32.779 user pros, cons. Then just plug in, say, Grammarly 00:17:32.779 --> 00:17:35.789 versus Quillbot. for learning new topics. Explain 00:17:35.789 --> 00:17:38.589 complex topic to me like I'm 12, use simple analogies, 00:17:38.710 --> 00:17:40.369 then give me a three question multiple choice 00:17:40.369 --> 00:17:42.650 quiz with answers. Great for getting up to speed 00:17:42.650 --> 00:17:45.410 on something like cloud computing. That's fantastic 00:17:45.410 --> 00:17:47.710 for quickly grasping something complex, especially 00:17:47.710 --> 00:17:49.910 if you need to learn it fast. Definitely. For 00:17:49.910 --> 00:17:52.569 creative writing, write a genre, you example 00:17:52.569 --> 00:17:55.529 sci -fi short story about topic, a lonely robot 00:17:55.529 --> 00:17:58.470 with a surprising element, you example twist 00:17:58.470 --> 00:18:02.240 ending where it finds a dog, then ask. Suggest 00:18:02.240 --> 00:18:05.119 three ways to develop this story further. And 00:18:05.119 --> 00:18:07.180 there are loads more useful templates out there. 00:18:07.460 --> 00:18:10.519 Email improvement, document summarization, brainstorming 00:18:10.519 --> 00:18:13.440 ideas, organizing data, planning presentations, 00:18:13.900 --> 00:18:16.339 content editing, data analysis, translation, 00:18:16.900 --> 00:18:19.119 even personal coaching frameworks. They give 00:18:19.119 --> 00:18:21.660 you a solid structure, a powerful head start 00:18:21.660 --> 00:18:23.779 for all sorts of tasks. This really does feel 00:18:23.779 --> 00:18:25.539 like having a whole toolkit ready, doesn't it? 00:18:25.599 --> 00:18:27.500 One that even helps you refine your own thinking 00:18:27.500 --> 00:18:29.819 about what you need to ask. It's all about using 00:18:29.819 --> 00:18:32.539 the AI to make your own interaction with AI smarter 00:18:32.539 --> 00:18:35.690 and more efficient. So... Pulling this all together, 00:18:35.829 --> 00:18:37.789 what does this really mean for someone listening 00:18:37.789 --> 00:18:41.309 who wants to truly master chat GPT, to really 00:18:41.309 --> 00:18:43.769 elevate how they use it? It boils down to maybe 00:18:43.769 --> 00:18:46.650 five core ideas, I think. One, communicate naturally. 00:18:46.789 --> 00:18:48.470 Talk to it like a smart person, not a search 00:18:48.470 --> 00:18:51.829 engine. Two, use projects. Seriously, organize 00:18:51.829 --> 00:18:53.789 your work into these focused specialized areas. 00:18:53.990 --> 00:18:57.789 It's a game changer. Three, always verify facts. 00:18:58.109 --> 00:19:00.109 Don't trust it blindly, especially for important 00:19:00.109 --> 00:19:03.369 stuff. AI can be confidently wrong. It hallucinates. 00:19:03.589 --> 00:19:06.440 Four, Iterate and refine. The first answer is 00:19:06.440 --> 00:19:09.119 rarely the best. Push it, ask for changes, polish 00:19:09.119 --> 00:19:12.720 it. And five, be specific. More context, more 00:19:12.720 --> 00:19:15.359 detail leads to exponentially better, more relevant 00:19:15.359 --> 00:19:17.539 results. It really sounds like it's about building 00:19:17.539 --> 00:19:19.880 a more conscious, more interactive relationship 00:19:19.880 --> 00:19:22.420 with the AI. Less passive prompting, more active 00:19:22.420 --> 00:19:25.339 dialogue. Exactly. Chat GPT is an incredibly 00:19:25.339 --> 00:19:27.940 powerful tool, but look, it's not magic. Mastery 00:19:27.940 --> 00:19:29.960 takes practice, takes engagement. Start free. 00:19:30.200 --> 00:19:32.740 Try these techniques. Focus on natural language. 00:19:33.220 --> 00:19:35.140 Set up a project for one common task you do. 00:19:35.359 --> 00:19:37.700 And always, always double check anything critical. 00:19:38.059 --> 00:19:39.839 Yeah, the goal isn't really to become some kind 00:19:39.839 --> 00:19:42.380 of AI guru overnight, is it? It's more about 00:19:42.380 --> 00:19:45.740 gradually building ChatGPT into a tool that truly 00:19:45.740 --> 00:19:48.480 works for you, understands your needs. It should 00:19:48.480 --> 00:19:50.839 amplify what you can do, not replace your own 00:19:50.839 --> 00:19:53.160 thinking. The better you communicate, the more 00:19:53.160 --> 00:19:55.680 powerful it becomes in your hands. So the question 00:19:55.680 --> 00:19:59.329 to ponder is, What new possibilities will honing 00:19:59.329 --> 00:20:02.210 these skills unlock found? What will you create 00:20:02.210 --> 00:20:05.250 or solve or discover when AI becomes a true extension 00:20:05.250 --> 00:20:07.569 of your own intellect? Your AI powered journey 00:20:07.569 --> 00:20:09.069 really starts with your very next conversation. 00:20:09.250 --> 00:20:10.990 Go experiment. Out T -Row music.