So it

seems like the with chat GPT or some of these other tools, like a lot of it is built around creating a good prompt. So one, like maybe defined kind of describe for us for people who haven't fiddled with these tools, like what is a prompt? What are we talking about here? How does that work? And then also like the key to it is making good prompts. And so like what what's the difference between a good prompt versus a crap prop? And ultimately, like it's kind of like, like I talked to our speakers about this within the speaker labs, sometimes just like if you ask a generic question, you're going to get a generic answer. So the more specific you can be tends to lead to like a more specific better tailor to answer. But it forces you to like really think about the best possible question to ask to get the best possible answer. And I would assume it's similar with prompt. So to find like, again, kind of what is a prompt? What are we talking about there? And then what are some tips for creating good prompts that lead to good outputs? Yeah, so a prompt is the updated vernacular for what we would also think of as like a Google query. When we all

learned how to use Google before Google got smart enough to kind of build these things in, do you remember that we had to use quotation marks to keep words together? And then the all caps and or not? But you know, those things, even I think you can still kind of do that a little bit on Google, but the tool is evolved so that we don't have to, right, that it's intuitive. So prompting is queries kind of like, I don't know what we're on 6.0, right? And so prompting is is asking of the tool, giving the instructions to get back the correct output. And remember, so I said that that chat GPT and other AI models are trained with the the purpose of predicting the next best word in the sentence to put this all together. So it doesn't have the context, it doesn't have the intuition that we built over our careers that we built over our education. So it's a great assist, but you also you have to be such a good delegator and be so specific. So to be a great prompt engineer, you have to really set the tone. So for example, if you're in chat GPT and you're just messing around for the first time, you're going to create a conversation thread. That conversation is kind of in its own silo. Chat GPT is not referring to any other conversations you've ever created under your account, whether you have one or not. And so you have to provide it all of this information from the get go, which means you need to tell the tool what role it needs to play for you, right? What what do you need it to be? Do you need it to be your marketing copywriter? Do you need it to be your VP of finance? Do you need it to be your writing coach? Right? So give it a role, give it some explicit tasks, and then tell the model that you're going to give it the additional context that you need. Now I have a prompt on my website, and it's my kind of my hook in, but I'm happy to give it to your listeners as well. And it's it's a really awesome way to learn how to be a prompt engineer. And it kind of sets for you a way to conversate with chat GPT to find out what it's really looking for in order to make better prompts. So you might tell it like I want you to become my prompt engineer, and your goal is to help me craft the best possible prompt for what I'm looking for, because the prompt will be used by you, chat GPT, to give me the output I'm looking for. And it kind of goes through this song and dance, where you give it some information, and it'll write a prompt, but then it will also give you four or five questions that would make the prompt better. Yeah. And that's how I started writing and learning how to write prompts, and understanding that a an okay prompt, like you were saying, like, we're going to ask better questions, an okay prompt is one that is just a couple sentences, right? A really good clear prompt that's going to get you 90% of the way with your output is probably several paragraphs. Yeah. And with explicit information. And if I want someone to write, like, if I want to write a sales page in your voice or in the voice of Amy Porterfield, I can provide exactly what that looks like, write me a sales page that looks like this, except for my course, my talk, my such and such, right? And then the close the more context I can provide, the better it gets. And you have to give more feedback. So I was prompting yesterday around some like launch items for my new business fan wagon. And I mentioned that my birthday was coming up somewhere in the prompt. I don't know, we were having a conversation. Well, at the end of the output, Chad GPT says, I'm glad this was helpful. Don't forget to come back if you have any more questions around this NPS happy early birthday. And I'm like, that's one, both like, terrifying into like, fun that, that it understands. There was a study done with, with patients like health care patients. And we're having virtual visits with doctors, you know, kind of that live chat virtual visit experience. Some of the patients saw a or corresponded with an actual MD, some of the patients corresponded with a chat GPT model. The patients who corresponded with the chat GPT model actually rated it as a better experience, because chat GPT showed up as more empathetic. Yeah. And it's response and then a normal MD, they would start with, Oh, I'm sorry, you're having a headache, or I'm sorry, you're having this problem. And if that's the way that the model is trained to like first respond versus an MD who might be like, tell me more about the symptom. What else can I do for, you know, like a much more nut, it's not that the human has less empathy is that we forget to share it. But the model is taking right now, some of the better parts of humanity and mirroring those back at us.