EP 310: The One ChatGPT Mistake That We‘re All Making

This is the Everyday AI show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business and everyday life.

There's one chat GPT mistake that we're all making. Well, not just chat GPT but large language models in general. And it actually has very little to do with prompting and it has everything to do with mindset and with our skills and with our abilities. And if I'm being honest, I think large language models are actually making us lazier and they're making us kind of dumb. But it shouldn't be like this. There's more to using large language models than meets the eye. And I've literally helped thousands of people learn chat GPT and other large language models. And I'm going to be sharing with you today the one chat GPT Chat GPT mistake that I think we're all making and how we can change it. All right, I hope that got your attention. I'm excited for today's show. So if you're new here, what's going on? My name is Jordan. I'm the host of Everyday AI and this is a daily live stream podcast, free daily newsletter helping us all learn and leverage generative AI. This is a live stream. This is unscripted, unedited, the real estate and artificial intelligence. So if you haven't already, why the heck not? Go to your everydayai.com Sign up for the free daily newsletter. We'll be recapping today's show and a whole lot more, including bringing you the AI news. So it is our hot take Tuesday. On Monday, we bring you what's happening in AI news and why it matters. And on Tuesday, we bring you hot takes. So for our live stream audience, at least let me know. I could get a little cranky today, you know, should I keep it hot? Should we go kind of soft? Let me know. I already said large language models are making it dumb, so apparently I'm a little spicy. All right, but let's go ahead and before we get started, let's go over and start as we do every day with going over the AI news and hey, live stream audience, take a look at your, take a look at your screen there. All right, let's. In AI news, a lot going on today, but two powerful new AI video tools have just been launched. So first there's dreamflare AI. It's a startup that aims to assist content creators in producing and monetizing short form AI generated content. Dream Flare offers two types of animated content. Flips which are comic book styles, stories with AI generated short clips and images and spins which are interactive. Choose your own adventure short films. The other new AI tool that just launched is Odyssey, a San Francisco based startup that just came out of STEALTH mode with $9 million in seed funding from GFE and other GV and other investors, which is aiming to revolutionize the film industry with advanced AI technology. They're looking to be a Hollywood grade AI video tool in line with Sora, Runway, Cling and others. So we'll obviously have previews of that in the newsletter. Here's a fun one, y'. All. The New York Times may be suing Open AI, but it is reportedly using the AI tool to help write its headlines. Yeah. All right, so according to reports, the New York Times has been using OpenAI's technology for headline generation in enforcing its style guide as uncovered in a leaked code repository on GitHub and first reported by the Intercept. So the leak code exposed the Times experimentation with AI tools, including a headline generator and style guide checker. So the Times alleged use of OpenAI highlights the evolving role of AI in newsroom tasks traditionally carried out by human editors. So this one obviously just goes to that legal battle right between the Times and OpenAI and underscores the significance of the copyright infringement allegations in the context of AI integration into journalism. So yeah, if this is true, that might really hurt the New York times case against OpenAI if they're still using its platform. I don't know when they did request that the technology be destroyed in this lawsuit. All right, one more piece of AI news and still, hey OpenAI, a lot of, a lot of big news from them today. So OpenAI and Thrive have just launched a new AI powered health company. So Thrive AI Health, a new startup backed by OpenAI's venture fund and Thrive Global aim to democratize access to expert level health coaching using AI to address health inequities and chronic diseases. So decarlos Love, a former Google product leader, joins as CEO to lead the development of a personalized AI health coach targeting behavior change across sleep, food, fitness, stress management and connection. So this is going to be leveraging resources from OpenAI and Thrive Global. The company plans to empower individuals through AI driven health coaching focusing on prevention and disease treatment optimization. All right, so a lot of, a lot of news there going on and yeah, but that's not why you tuned in. Shout out to our live stream audience. Get your vote in now on this secret poll. But let's just get straight to it. All right, let's talk about the one chat GPT mistake that we're all making. And again, to reiterate, this is not just chat GPT, but I'd say it's the most widely used large language model. But here's our, here's our takeaway. Here's the end. We're using large language models to become more efficient in our current skill set, right? But that's the absolutely wrong way to use a large language model. Doesn't sound like a smart take from me. Right? Like, wait, isn't the whole point of using generative AI Isn't the whole point of using large language models is to increase productivity and to increase efficiency? Kind of. But we're becoming over reliant and I think that as a society maybe. And you know, obviously I'm speaking to our audience here, right? I'm speaking to, you know, our live stream audience, people such as, as Michael and Fred and Tara and Jay and Colby and Roland and Ben. Right. All of us out there who are using AI on a daily basis. And I'm putting myself in there as well. And this is something I always have to remind myself because it's a big mistake. I think we're choosing or we're opting for short term productivity and we're sacrificing long term core skill development. Yeah. I have to pause and think about this because I think it's actually problematic. Right? And I think as humans we can be lazy, right? Kobe said to bring the heat, so did Tara. So maybe I'll bring some heat here. As humans, we are inherently lazy, right? You always look for the fastest way to do something. You look for the easiest way. You look for shortcuts, Right? That's why I think generative AI has been such a huge boom, Right? Because it allows people to do their jobs so much faster. Right. And especially for those that, you know, their generative AI use is kind of under the table, or companies are like, yeah, you can use it, but they're really not pushing and they're not educating their employees. Those that have figured it out are all of a sudden like, yo, like, I can get my job done in a fourth of the time. This is amazing. Right? These reports take minutes instead of hours, Right? This Data analysis takes 30 seconds instead of half a day. The people that are using generative AI are choosing short term productivity. And there's nothing wrong with that, Right? Because prior to generative AI, that's what it was all about, right? It was all about, hey, you know, how can you use marketing automation? You Know, how can you use all these new, you know, software as a service? And you might be thinking, okay, well, what's the difference? And why is that a problem? Jordan, if we're using large language models to just be more efficient and more productive, why is that a problem? Because historically, that's how it's always been, right? You use the latest technology, you use the Internet, you use cloud, you use mobile, right? You use what tools and technologies are available to become more efficient and more productive. Yes, but generative AI is a little different, right? Because as an example, the Internet didn't really do your job for you, right? I mean, some people, yeah, sure. But for the most part, right, when we look at knowledge workers here in the US that's the majority of our audience. So if you sit in front of a computer, you every single day and you are paid for your expertise, you are paid to whatever your expertise may be. It could be data entry, it could be project management, it could be marketing automation, it could be so many of these things. But historically the Internet didn't change that too much. Maybe it made it easier. But the difference is with generative AI is it's actually for the first time ever, starting to do our job for us, right? And therein lies the problem for our long term skill development. And this is where I think we're overlooking even the meaning of a large language model. They are literally trained in their system prompts to be helpful assistance, yet we're not using them to be helpful assistance. We are just giving them certain blocks of our work, trying to find the shortest, kind of like the combination of the easiest prompt that has the best results. And then we are trying to give the models as much of our work as possible. And again, it's not bad, but it's the wrong way. All right? And hey, I'm gonna, I'm gonna say this right now. Take a little 10 second commercial break here. You're probably going to want to repost this episode. All right, so if you're joining us live on LinkedIn, go ahead, click that little repost button now, because this problem, I'm gonna help a couple people, right? So if you're listening on the podcast, make sure to check out your show notes. We're going to have a link to this LinkedIn post. We go live on LinkedIn, Twitter, YouTube, but you can't share these on YouTube. So make sure to go check out the LinkedIn post and repost this, all right? Because we're going to help. I think we're going to do three Kind of short one on one sessions to help you get over this problem because we've helped thousands of people. But more on that later. But let's dig into a little bit more here on this problem because like I said, this is how almost everyone is using large language models right now. Right? Here's my work. Go do it. So let's talk, let's go through an example, right? And probably the easiest example, I know this might be a lazy example, but let's say you are a writer, a marketer, someone in communications, pr, etc, and you are using, let's just say chat GPT to do your writing for you, right? This is one of the earlier use cases of generative AI technology, right? This is even what our team started using the GPT technology for, right? When it first became publicly available in late 2020, right. It was essentially a copywriting tool early on through the GPT3 technology in tools like, you know, copy AI, Jasper, you know, we mentioned some of those because our team was using those almost daily back in 2020. Even my background, right, talk about this a little bit, but my background's in journalism. I've been getting paid to write professionally for more than 20 years. Some of the biggest brands in the world, commercials, you know, big brands like everything. I've been getting paid to write for 20 years. So, you know, these early tools, it was easy to be like, oh, wow, we can hand off a lot of our copywriting. Even some of the earlier tools, I mean, they, they're looking back at them now, they were absolutely terrible. But they can handle a skill like copywriting. It's something large language models are great at. So let's use that as an example. All right? So let's just say for whatever reason you are a writer and let's say your current skill set, or maybe, you know, you're a marketer and writing is one of your big jobs. But let's say that you currently have a skill set, 6 out of 10 in writing. All right? Hopefully, hopefully. We're tracking, we're tracking here on this one. So you have a skill set 6 out of 10. But then guess what? These large language models are pretty good, right? And as a writer, especially writers who've been doing it a long time, you can spend hours, literally hours on one paragraph. You could spend an hour toiling over one word, right? To make it better, right? I gotta cut the fat. I have to, you know, this, this, these two paragraphs, we need to transition between these two paragraphs. It's kind of abrupt or right like there's an art and a science of writing, but let's just say you're a 6 out of 10 and you're handing it over to chat GPT and you're like, whoa, this chat GPT things pretty Good. Or this Claude 3.5 sonnet is, is pretty good. I'm handing over my copywriting, all of it to them. All right, but then we have to think about why, right? Because then essentially you're just using a large language model to write more copy faster with a simple copy and paste prompt, right? That's ultimately what, what we do. Like I said, we're lazy humans. We look for shortcuts. We say, hey, what's the best prompt or the best, the best methodology that's quick and easy, that I can get the best results, that I can just put this copy and paste this and oh, boom, my eight hour job is now one hour. And maybe I'll tell people, maybe I'll take on more work, maybe I won't, right? But that's what so many of us are doing. Okay? So I mean here's, here's the, the good, the good part is you'll be able to wait, you'll be able to write way more copy at that current skill set. Like I said, you're a six out of ten. Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on Genai. Hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead. And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to to use gen AI. So whether you're looking for chat GPT training for thousands or just need help building your front end AI strategy, you can partner with us too. Just like some of the biggest companies in the world do. Go to your everyday AI.com partner to get in contact with our team or you can just click on the partner section of our website. We'll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on the gen AI. What about the downside though, right? We don't think about that if we become over reliant on large language models, which is part of the problem, right. It's part this is, this is the big issue here is we are becoming too reliant. So if you are that skilled copywriter, 6 out of 10, guess what? Your skill set is going to go down over time, right? If you think that the more and more you use chat GPT or use a large language model to do your work for you, let's say you're in data analysis, let's say you are in web development and, and you're, you're coding. Let's say you are a strategic creative, right? Whatever it is, there's always a con. If you are using chat GBT just to do your work, just to replace your skills, a simple copy and paste prompt, your own skill set is going to start to deteriorate. So if you are a 6 out of 10 and you've, you know, learned to pump out some, you know, C plus, B minus content in 1/10 of the time, you're gonna keep doing it, right? What happens to your skill set after a week, Are you still a six out of ten? After a month, are you still a six out of ten? After a quarter, are you still a six out of ten or do you start to lose your skills, right? So many of the high demand skills that we get paid and we get rewarded for being experts, we do so by keeping those skills sharp, right? By practicing them daily, hourly, by the minute. That's what got us in the position where we are today. Whatever your role is, whether you've been there for, for two months or 20 years, there's a good chance that you're there because of your expertise. Because you've gone for maybe years, maybe decades, you've gone through these manual steps of building your expertise, becoming a subject matter expert in something. So now we're at this crossroads, right? In this new economy, in this new AI first AI native economy, where things are different. We're not going to, in the future. I've said this in the future, we're not going to be rewarded, promoted for what we know. We're not going to be promoted for our current skill set or rewarded. We're going to be rewarded and promoted for how we use that with AI, can we get the most out of AI leveraging our former skill set, right? But you're going to lose it. If you just hand your skill set over, you're going to lose it. I've talked about this before, right? That's one of the reasons why I still write the daily newsletter by hand. Yeah, it's, it's almost like vintage, right? You know, I Was I was joking around with a good friend of mine about a month or two ago, you know, and we said, hey, in the future, it might be, you know, let's just say you buy, I don't know, an app on your phone, I think was his example. And he said, you know, there might be a market in the future where, oh, this was made by humans. This was made without code. Right. And it's almost like, you know, like vintage or like a throwback. Right. Like buying a record. Right. But still, right now, I write the daily newsletter for our. Right. When I'm done here, I'm gonna go and write the top takeaways that we got to in this episode. I might answer a couple of the questions that come in from our audience. I might answer that in the newsletter, but I write it right? Why? Because that's my core skill set. One of my core skill sets is being able to tell stories, being able to write, being able to craft compelling copy. Can Chat GPT do it better than me? Absolutely. But if I just hand off and do I use it for other aspects? Absolutely right. But if I hand off one of my core skill sets, if I'm a 6 out of 10, I'm going to start to lose that skill set. So here's what we need to do, y'. All. Here's what we need to do. We need to turn Chat GPT into a consultant. All right, I'm going to walk you through what that means. And if you've recently taken our PPP course. Recently, right, because we updated it a couple months ago. That's our free prime, prompt, polish, prompt engineering course. This next part might sound familiar. So if you're on the podcast, make sure to just literally or anyone, if you want access, probably the easiest way. I mean, you can type in PPP in the chat here. And, you know, if you have taken the course recently, let me know what you think in the comments. If you're watching live, the easiest way, if you want to get access, we don't put it on our website. It's kind of secret. It's hidden. Just subscribe to our newsletter. Your everyday AI.com reply to any of the emails. Just put PPP and I'll send you the link. All right. But if. If you have taken our course, this next part will probably be familiar. So, like I said, we've taught. I have to actually count. We're probably close to 6,000. 6,000 people live. We do this live multiple times a week. I'm going to be doing this in about three Hours. Right now we teach people the basics of prompt engineering, but the biggest thing is turning Chat GPT into a consultant, right? Because going back to our previous example, if I'm, if I'm a copywriter, that's a 6 out of 10. And if I just am handing my work off, my skills are going to go down, right? I'm sacrificing short term productivity for long term skill development. I'm eventually going to be a 5 out of 10 and a 4 out of 10 the more I use AI, right, but and the other thing too, if I'm a 6 out of 10 and I'm using a simple prompt, right? But think in your own head. Whatever you use a large language model for, right? Sales, prospecting, research, competitive analysis, like whatever you're using it for, if you're just finding the simplest copy and paste prompt, your skill sets actually going down. Not only that, but your output or what Chat GPT or a large language model is capable of is going to be capped as well. It's going to be capped at your current skill set if you're just doing a simple copy and paste prompt. If you turn Chat GPT or a large language model into a consultant, your skill sets are going to increase and the output of the model is going to improve, y'. All. It's simple math, it's simple science. But we overlook it because we are lazy humans and we say what is the fastest, shortest, easiest way to give my work to this large language model and have it be done. But remember, it's a helpful assistant. So in our prime prompt polish course, we talk about priming and how essentially your output out of a large language model is dependent on the work that you put in, right? In these models. There's a lot of garbage in there, right? When we talk about these ultra jumbo models with trillions of parameters that have just scraped the entirety of the Internet. There's a lot of bad information out there. So depending on how you're using it, you might just get not that great outputs. Especially if you're just trying to copy and paste prompt your thing. If you're trying to zero shot your way to something usable, if you're trying to few shot your way to business growth. That's not how it works. You have to turn Chat GPT into a consultant. We teach Refine Q. It's a priming technique. So before you ever ask for an output, you got to put in, you got to put in a lot of work, a lot of back and forth. Essentially a conversation we had This a podcast episode a couple of months ago with Abram from OpenAI and we talked about this exact concept, right? And how people are getting bad results out of large language models because they're just putting a simple prompt in there and they're not taking the time to essentially train the model or train the response or coach or teach a skill set. That is what we need to do. And when you use the Refine Q method, right, so that's assigning a role, giving examples, fetching information, asking for insights, narrating the pain points, explaining the expectations, and ending with questions. Right? So that is Refine Q. The biggest part is asking questions when you are using a large language model. So if you don't get anything out of this long winding podcast live stream today, get this. Do not ask for an output. When you start a conversation in Chat GPT, turn it into a consultant, give it all of the information it needs and then tell it to poke holes. Right? That is what refineq is. You do not want an output. You want to turn Chat GPT into a consultant. Because again, not only is that going to make your output exponentially better, whatever you're doing, and then you will be able to reuse it and scale it, but you as a human will get smarter. It's problematic. AI is so good. It's problematic. We're handing everything over to AI, but it can make us dumb. This is, I think the single biggest problem of large language models is we're not using them the right way. Right. So we need to turn it into a consultant. And I'm gonna not poke fun, but hopefully I'm gonna illustrate a point here because, you know, some of the biggest consulting companies in the world, when you go back to 2021 or 2022, when Chat GPT came out, a lot of the biggest consulting companies in the world wrote off generative AI. They wrote off large language models. They said, this isn't a threat. They literally go back and read it. We, we have receipts all day. Right. They literally advise their clients, do not use AI. This is not a technology you should be using. Right. I poked fun of them, you know, a year and a half ago on the show, when even still, right in early 2023, companies were like, some of the largest consulting companies in the world were like, Anna, this large language model thing, don't worry about it, don't think about it. Why? Because they understood if used correctly, it could do a bulk of the work that they were doing. Yeah, there's kind of some, some early articles floating around, you know, when people finally understood the basics of prompt engineering. Right. Oh, I have to work with the model a little bit and I can turn it into a consultant and it's going to make me better. You know, know, people started saying, oh, large language models are like having a consultant in your pocket. Right? Yeah. And eventually, eventually these big consulting companies realize, oh, we're losing a bunch of clients, you know, and, and the public consulting companies were like, oh, our forecasts aren't going very well. You know, our shareholders apparently aren't happy when we say, ah, this AI stuff, don't pay attention to it. Large language models, they're dangerous. Right. They're not very good. We're the best. Right. Of course, consulting companies were writing it off. Right. They felt threatened. If you don't understand a technology or if it's really good, you feel threatened and you tell people, ah, ignore this, this AI stuff, not very good. Hey, small medium business. Hey, enterprise company. That's been one of our overpaying clients for decades. Don't use AI. You need us chat GPT. This thing is large language models, they're just for blog posts. No, they're not. Absolutely not. Large language models, when used correctly, outperform the smartest humans, period. In narrow tasks. There is no when used correctly. That's the thing. 99 of people aren't using large language models correctly. The ones who are, are growing their companies, they're growing their careers, they're growing their departments. But eventually these consulting companies, ah, they couldn't run from it anymore. Right. And they realized, oh yeah, these large language models, if you do a little more than just asking for a simple prompt, they're actually really good. Yeah, right? Yeah. After writing them off for so long, they're like, oh, yeah, Turns out the first companies that started investing in AI and they started mentioning it and, oh, all of a sudden their business grew exponentially, bringing in tens of billions of dollars in more revenue because they started to, hey, we should use AI. And it's actually pretty good as a, as a strategic partner. Right? As a helpful assistant. Not as just copying and pasting your current skill set. Because now look what's happening. Oh, PWC is investing a billion dollars in AI. Deloitte investing $2 billion in AI. Accenture investing $3 billion in AI. Guess what? They figured it out eventually. The future of work is with generative AI. And when you use large language models correctly, when you use them, what they were built for, to be a helpful assistant, they are not only going to grow your own skill set, but your output in the model is going to be unrecognizably better. So, humans, can we all just do something? Can we stop being lazy? I'm calling myself out as well. I mean, is it nice to jump into a, a co pilot or a Gemini or a Claude or a Chat GPT and put in one little prompt and have it do, you know, three hours of work in three minutes? Absolutely. That's great. Right? But at what cost? Right? Especially if you are using it consistently for that one skill set that is your expertise. And I want to clear this up. I'm not telling you. Right. I gave the example of myself. Yeah, I write the newsletter because that, that part for me is important. Writing and clear communication is important because prompting large language models requires clear written communication. So that part's important. All right, so I'm not trying to dissuade you from using large language models to do your work. It's the exact opposite. You should be doing it, but do not do it in a copy and paste fashion. Do not do it in a lazy fashion. Do not do it. Put in a 6 out of 10 effort, the, the, the easiest copy and paste and just ask for an output. Because you are missing out on the most powerful aspect of, of large language models, which is it will help you be better. Right. When you tell Chat GPT or Claude or Gemini to be a consultant, when you go through, as an example, our refined Q process, it makes you think all of a sudden, if you do this correctly, you have a seasoned consultant at PwC. On the other side, you have a seasoned strategist at Accenture. On the other side, you have a, a, a wise Deloitte leader on the other side asking you questions, poking and prodding, forcing you to do more research. That's the key. It's saving you time, making you smarter and improving your output all at the same time. So if you do it correctly. So don't just use Chat GPT to do your work. Use it in other large language models to make your work better, to make you smarter, to make you the AI leader in your department, in your organization, in your company. Use it the right way. Don't make the same mistake that everyone else is using or that everyone else is making. Right. Use large language models the way that they were built to be used to make you smarter and to make your work better and to save you time. All right. I hope that was helpful. You know, a little bit of a hot take Tuesday. We got a little bit spicy, right? But for our live stream audience Appreciate you tuning in as well. Podcast audience. Hey, I told you, go check out the show notes. Come back to this LinkedIn. Click repost. So we're going to be doing three consults that we generally charge hundreds of dollars for and we're going to work with you one on one to help you avoid making these mistakes. Whatever your large language model, whatever your large language model of choice is, we can help you get better at it. We can help you work through problems, right? So whether it's your, your prompting technique, how to bring in more accurate, up to date information, how to, you know, maybe as an example, you're using Claude and you're like, oh wow, I've heard about this new feature, but I don't know how to use it. Whatever your large language model question is, we can help you. All right? And yes, we companies or individuals generally pay us hundreds or thousands or tens of thousands of dollars, right, for some of our time. So we spend a lot of time here at Everyday AI to bring you high quality information. So we just ask if this was helpful. Click that repost button, takes you four to five seconds. All right? And we spend four to five hours sometimes on shows like today's. So anyone that reposts this on LinkedIn, you will be entered into a drawing where we're going to be giving away three quick one on one consults to get more out of large language models. So I hope that's you. Takes takes a couple seconds. So if you're on the podcast, please, super simple. The link is going to be in the show notes so you can go to this exact episode, click that repost button and you will be entered to win. So I hope this was helpful. Hey, in live stream audience. Let me go back here. Let me go back here. Sorry, we got a lot of flashing going on, but tomorrow you're deciding what's on the plate, so let me know. Thank you for tuning in. Hope to see you back tomorrow and every day for more Everyday AI. Thanks y'. All.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going for a little more AI magic. Visit youreverydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.