Alastair McDermott 0:00
There’s a lot of talk about AI. There’s a lot of people using it to generate amazing things. There’s a lot of people using it to generate absolute rubbish. And putting that out there on social media and misusing it in different places. So when my guest this week was unavailable for the interview, we had scheduled, somebody suggested to me that I create an interview of myself with AI. And so what I decided to do was do that and have a little bit of fun with that. And so that is one segment of this show. But the rest of the show is kind of like a discussion about how I actually use AI to generate high quality content, I’ll let you be the judge of that. And I’m going to show you how I actually use it to brainstorm and that will be towards the end. If you watch this on YouTube, you will see I’ve generated some of the segments are actually AI generated video as well. So stay tuned, I think you will find this one interesting, I hope you find it useful. I hope it helps you to use AI to help you be more productive and to get more done.
Voiceover 1:02
Welcome to The Recognized Authority, the show that helps independent consultants and domain experts or ignore by building authority, helping you command premium fees, work less hours, enjoy your work more. Here’s your host, Alastair McDermott.
Alastair McDermott 1:16
Okay, so I have something a little bit interesting for you, and a little bit different. It’s an experiment, I’ve been playing around with technology, I wanted to demonstrate and show the limits of the existing technology to you give you a taster of the current state of play with some things and show you some of the things that you can do with the current AI tools that we have. So the first thing I’m going to do in this podcast is I’m going to play a segment of the podcast, which is almost entirely AI generated in its own right. And I’ll tell you afterwards about some of the inputs that I gave it to, to actually help generate this. So the voices that you hear the interviewer, and the guests are both AI generated. And that will be pretty clear when you listen to it. And so there was about six minutes of this segment. And the entire conversation is AI generated. But the guest here is me. And it has been trained on some inputs that I’ve given it that reflect what I think and are based on some of the things that I’ve written. And including one of those things is my AI policy. And the other one of those things that I inputted into it was my side have a conversation about AI with with Frank Prendergast on this podcast. In a previous episode, I just gave up my side of the conversation, because I just wanted to give it my take. And so I got the AI to generate an interview between me and a radio DJ radio interviewer, and the output, or what it’s given us is this. And I put it through some other tools to actually great both sides of that conversation. So first, I’m going to play the conversation. And then I’m going to talk to you a little bit about it afterwards. And then tell you more specifically in my real human voice about how I’m using AI and show you some of the things if you’re watching the video version of this, you’ll see it on screen, but I will be describing it for everybody listening on audio. I’ll just show you how I’m actually using AI at the moment and give you an idea about how you might be able to do that too.
AI Interviewer 3:23
Today we have a special guest Alastair McDermott, an expert in using AI for content creation. Alastair, welcome to the show. Let’s dive right in.
Alastair McDermott 3:33
Thank you. It’s great to be here.
AI Interviewer 3:35
How did you first get into using AI for your writing and brand building?
Alastair McDermott 3:39
I’ve always been fascinated by technology and its potential. When tools like chat GPT and Claude came along, I saw an opportunity to streamline my content creation process. It’s been a game changer for me. What were
AI Interviewer 3:53
your initial thoughts on AI generated content? And has your opinion changed over time?
Alastair McDermott 3:59
Initially, I was both excited and skeptical. The technology has its limitations, especially in capturing tone and style. But as I’ve used it more, I’ve found ways to make it incredibly effective, especially when you understand its capabilities and limitations. Can you
AI Interviewer 4:15
give us a quick rundown of how you integrate AI into your content creation?
Alastair McDermott 4:20
Absolutely. I use AI for brainstorming, reformatting and drafting some content, especially titles. The key is to use it as a collaborative tool, not a replacement for human creativity.
AI Interviewer 4:33
why should other business professionals consider using AI for their content?
Alastair McDermott 4:38
If you’re not using it? Yeah, you’re missing out. It’s like a an accountant refusing to use a spreadsheet. I can do a lot of the grunt work for you, saving you time and potentially improving the quality of your output.
AI Interviewer 4:54
You’ve mentioned that quality over quantity is important. How does AI help maintain that quality
Alastair McDermott 5:01
AI can help you draft and revise quickly. But it’s crucial to have human oversight. I often go through 10 iterations with the AI to get the content 90% of the way there. It’s about balance and taking advantage of AIS capabilities while maintaining a human touch.
AI Interviewer 5:19
Speaking of a human touch, how do you ensure your content still feels personal and not robotic,
Alastair McDermott 5:25
I use specific prompts and instructions to guide the AI. For example, I might say, as an expert copywriter, rewrite this, I also use it to emulate my writing style, which is more conversational and succinct. It’s all about the prompts you use
AI Interviewer 5:40
any pitfalls or mistakes people should be cautious of.
Alastair McDermott 5:43
Absolutely. One of the dangers is AI hallucinations, where the AI makes stuff up, it can severely impact your credibility, always review and fact check the content. Whoa, hey, so what happened there? Well, this is real Allaster. And the reason why the voice changed like that is because there’s a variability setting in the voice generation tool. And I had that variability setting quite high. And this is a setting that allows you to change how much variability there is in the tone, so that it doesn’t sound monotonous and robotic. And unfortunately, when you set that, above a certain threshold, or if you push the limit of that, it gets into this red zone. And obviously, it can go a bit haywire. And so that’s what it did. It went off on a tangent. And so it gave us this completely different voice. So what I did was I regenerated the rest of the interview with that variability setting lower. But I wanted to leave in that segment, just to show you some of the issues that happened with these kinds of tools.
AI Interviewer 6:53
Has using AI made it easier for you to produce more content? What impact has that had on your brand? It
Alastair McDermott 7:01
has indeed, it’s like having an assistant that can draft, revise and brainstorm. It’s helped me maintain a consistent output, which is crucial for brand building.
AI Interviewer 7:12
Switching gears a bit. Any ethical concerns with using AI for writing?
Alastair McDermott 7:18
Ethical considerations are essential. using AI to fully create text with minimal human oversight can result in ineffective, even misleading content, it’s crucial to find a balanced approach.
AI Interviewer 7:32
How do you measure the impact of using AI on your brand?
Alastair McDermott 7:36
It’s a bit subjective, but the time saved and the ability to maintain consistent high quality output are good indicators. Plus, the positive feedback from my audience tells me it’s working.
AI Interviewer 7:50
Any success stories you’d like to share?
Alastair McDermott 7:54
I’ve seen professionals, especially those for whom English is not their first language, or who struggle with writing benefit immensely, they’ve been able to communicate more effectively and build their brand.
AI Interviewer 8:08
What’s your top tip for someone considering using AI for content creation?
Alastair McDermott 8:12
Understand what AI can and can’t do? Use it as a tool to augment your capabilities, not replace them. And always, always review the content it generates.
AI Interviewer 8:22
Can you talk about the actual use cases for AI and content creation?
Alastair McDermott 8:27
Beyond the obvious applications, like blog posts and social media? AI can be used for a variety of tasks? For instance, it can help draft emails, create training plans, or even assist in writing scripts. The key is to start with human input and let the AI augment it.
AI Interviewer 8:49
And that’s fascinating. How do you go about prompt engineering to get the most out of AI?
Alastair McDermott 8:53
Prompt engineering is all about being specific with your instructions. For example, if I want the AI to write in a certain style, I might say rewrite this in the style of Sheryl Sandberg, as I have a similar writing style to hers, the more specific you are with your prompts, the better the output.
AI Interviewer 9:14
You mentioned last time that AI can sometimes make stuff up. How do you handle that?
Alastair McDermott 9:20
That’s a big concern. You have to be vigilant and fact check everything. AI is great at generating plausible sounding information, but it can be wrong. Always double check before publishing anything.
AI Interviewer 9:34
What about the writing style? How do you get AI to match your tone?
Alastair McDermott 9:39
I’ve experimented with this a lot. I’ve even asked the AI to analyze my writing style, and then use those insights to guide the AI in generating content. That’s more me. It’s not perfect, but it’s getting there.
AI Interviewer 9:54
How do you see AI evolving in the content creation space?
Alastair McDermott 9:57
I think as the technology matures, So we’ll see better tone and style matching, fewer hallucinations and more nuanced outputs. It’s an exciting time to be in this space.
AI Interviewer 10:12
Do you think there’s a risk of AI taking over human jobs and content creation?
Alastair McDermott 10:17
That’s a common fear. But I see AI more as a tool that augments human capabilities. It can handle the grunt work, allowing us to focus on strategy and creativity. So, no, I don’t see it replacing humans anytime soon. What
AI Interviewer 10:34
about people who are skeptical or afraid of using AI? What would you say to them?
Alastair McDermott 10:39
Skepticism is healthy. But outright dismissal is not. AI is a tool. And like any tool, it can be used well or poorly. The key is to understand its limitations and possibilities. Don’t be afraid to experiment,
AI Interviewer 10:56
any new projects or experiments you’re working on that involve AI.
Alastair McDermott 11:00
I’m always testing new ways to integrate AI into my workflow. Currently, I’m exploring how AI can assist in market research and customer engagement. It’s still in the early stages, but the results are promising.
AI Interviewer 11:15
Finally, any additional tips for our listeners who are considering diving into the world of AI for content creation,
Alastair McDermott 11:22
understand what AI can and can’t do? Use it as a tool to augment your capabilities, not replace them. And always, always review the content it generates.
AI Interviewer 11:35
Alistair, thank you for another enlightening conversation. I’m sure our listeners have gained valuable insights.
Alastair McDermott 11:41
It’s been a pleasure. Thanks for having me on your show.
AI Interviewer 11:45
If you’re as intrigued by Alistair’s insights as I am, you’re in luck, stay tuned, because up next is The Recognized Authority hosted by the real Alastair McDermott.
Alastair McDermott 11:57
So here’s how I did that. And I’m going to show you, I’m not going to go into the details of the tool. The tool that I use for the video apart is called Hey, Jen, the the tool that I used for the audio part is called 11 labs. And those are tools that you can use to generate content, video and audio content from text. The first thing I’m going to do is I’m actually going to show you how I created the text part. And so one of the things that’s quite interesting is that the interview that was generated, the questions are all really good. And the actual answers that I gave are probably 95% accuracy for what I might actually say myself. And I found that really interesting. So let me just talk you through and there are people listening to this who don’t see the video. So I’m just going to talk you through what you’re seeing here.
Alastair McDermott 12:50
So the first thing that I did, I decided to change the gender of the interview or because because I couldn’t find a good voice. So I found a male voice that I want to use, but I originally called the interviewer Rachel. And so I said to a tool called Claude, which is like Chachi Beatty, I said, I want you to create an interview transcript for an interview between me, Alastair McDermott, and Rachel the interviewer, who is a veteran expert interviewer, and an entrepreneur who hosts a business talk show on Saturday mornings. Now a couple of things, I want to talk about why I said those things. So I wanted to make it clear that the person is a veteran expert interviewer because I wanted it to be a good interview. And I wanted it to be as if it was a really good interviewer who was asking those questions. But I also wanted it to be light. And I wanted it to be I wanted to describe it as a business talk show, because I’ve heard some of those kinds of talk shows and Saturday mornings, where it’s an interesting talk show, but like, you know, the, the interviewer doesn’t grill, the person they’re interviewing, they don’t put them on the spot too much. It’s kind of a keeps it light and just kind of fluffy, kind of high level overview. And I felt like that’s kind of what I was trying to go for here. I didn’t want to get too deep on something. So I said write an interview, entire conversation, both sides for a 10 to 15 minute interview, approximately 10,000 to 12,000 words make it flow naturally.
Alastair McDermott 14:10
Now one of the things bad all of the AI tools as they do find it difficult to generate large output. And so I’m asking for a lot when I say 10 to 15 minutes of an interview or 10,000 words, that’s a lot to be asking the AI to to produce. And I said make it flow naturally, I wanted to make sure that the questions flow side to side. So Below are some of the questions I might use. Please rewrite them to make it flow. So I have some questions here that I was giving us suggestions. And I’ll show you where I got those question from. But I did mention that I wanted it to reflow those because what happens is when somebody is being interviewed, the interviewer changes the next question depending on what the previous answer was. So they don’t just robotically go through the list of questions. So I wanted to rewrite the questions to make it flow more and make it make it sound like a real conversation. And so I’ll just show you where I got those Questions from? So I asked these I got these questions generate them in chat GPT. And you’ll see I’m switching back and forth I go back and forth between Chuck GPT and Claude, for different things. And Claude is is very similar chatty btw except it can take larger inputs, which is important in a minute, you’ll see why. So in Chachi, PDI said, you are an expert interviewer in the b2b Professional Services space, you are going to interview an author about his use of AI in creating content, like blog posts and books that are created to build authority and personal brand. So basically, you’re interviewing me, and you’re going to be interviewing about the use of AI.
Alastair McDermott 15:35
So I said, Give me 12 questions, you might ask him make the questions flow in a natural order where the answer from one might lead to the next question. Just give the questions as a numbered list and write them in natural language. And so Chachi Beatty has given me 12 questions here. So I have taken copy paste these into Claude, in my, in my request here in Claude, the reason why I got I find that chatty btw tends to give me better output on the most part, but Claude is better at taking larger input. So that’s why I go back and forth between the two. The second, so So going back to Claude, I said, read the entire interview conversation. And here’s some of the questions. So I said, the following text is in three parts. Part one is a list of the questions. Part two is my usage, my AI usage policy that I created for my website. And so I’ve got the text of that, because this gives insight into what I think about AI. So this is helping it to create an interview that is realistic based on me. And then the third, part three is my side of the conversation from one of my podcasts, where I discussed generative AI like Chachi Beatty. And so how I got that. So for every podcast episode that I put up on The Recognized Authority, there is a transcript not transcript is automatically generated in a tool called otter. And so what I did was I’m just gonna look for Frank, because Frank was the other person in the episode. So let’s see. See if I can find Frank’s episode here. And so what we did so here’s the chat GPT episode, this was a conversation I had with Frank. And so this is the entire transcript of that. So what I did was I took the output, I took this transcript. And I want to use my side of this as an input to the conversation because I wanted to create an interview with me. So what I did was, I took that transcript, and I gave that I have another window here, I won’t, I don’t have it open. So I’m not sure what but I had another chat window. And I said, Please pull out my side of this conversation only, and give it back to me. And so it gave me my side of the conversation only. So that’s where this part three came from.
Alastair McDermott 17:59
So now here’s the like, here’s all of the inputs into this conversation. So the first thing is, we’ve set up that it’s a, it’s an interview transcripts for an interview between me and this radio interviewer. And it’s going to be like a fairly light, high level, Business Talk Show type. And we have the input, which is the questions and then my AI usage policy, because that gives insight into what I how I use AI. And then my side of the conversation from where I talked to Frank, about two or three months ago. And so that also included a lot of my thoughts on how I use AI. And so all of that is input here into this. And so and then the rest of this prompt is just all of that. So there’s my AI policy. And then here is all of the text. So what this gave us as output is this 10 to 15 minute interview between Rachel and Alastair McDermott, and then we have the text here. Welcome to the show. It’s great to have you on thanks for having me on. I’m excited to chat. And that’s what you just heard or so if you’re watching the, or listening to the podcast episode.
Alastair McDermott 19:10
So that’s our, that’s all of our text. Now, our text input. So the next step, then, and and, by the way, dislike this output here, it’s it’s definitely not 10 15,000 words. And it’s not I think it’s about six minutes long. It’s so the output is much shorter than but you could say, you know, let’s go again, or I think I think when I was experimenting with this, I said, I came back the next week and had a further conversation. Please write the transcript of that conversation. And it’s able to generate more that way so you can end up with longer output. So that’s how I got the text of the of the of the interview. The next thing then was to take 11 Labs, which is a tool that takes text input. And what I did was I trained it on my voice. So let’s see if I can find over here 11 labs.
Alastair McDermott 20:10
And so what I was able to do with this was I was able to record some recordings, I think of 60 seconds each, and I think I did five or six of them and asked for, and then it clones my voice based on that. And this is a bit creepy. Like the sound of it, it doesn’t sound quite like you, it also gave me a bit of an American accent. And I think they’re getting better. Now I see different kind of accents, or multilingual and things like that are being added to this. But this tool can then generate the, the audio. And so what I did was I generated the audio with this. And the other tool that I use them, and I use this for my sort of conversation. And then I was playing around with different voices to get the interviewer. And that’s when I decided to change from Rachel, which was the name of the original interviewer, a female, I found a male voice I thought would be quite good. And then when I was playing around, I wanted to make a video elements to it. And I found a kind of like an old black and white, kind of 60s American talk show avatar that I used to create the video for that. So that’s where that whole interview came from. I was able to kind of stitch all those pieces together. And that created this interview. Now, the interesting thing is the interview is actually really good like it, it’s pretty much like I would answer those questions. There are only one or two sections of it that are not exactly as I would answer.
Alastair McDermott 21:43
But I would say 95%. It’s accurate for what I would say. Probably the most inaccurate is the shortness of the answers, I would probably answer I would probably take maybe two or three minutes answering some of those questions, whereas it gave like a 15 or 25 second answer. So I would be a bit more wordy with that. So yeah, that is how I actually created the interview and created the video of the interview. The video is a bit creepy. So if you check it out, you can have a look, I do have some notes on the video about what to look out for, you know, lip sync issues. But you know that the head is moving when the rest of the image or the rest of the body doesn’t kind of move naturally. But it’s like, you know, this is the worst that AI is ever going to be everything that we see today. It’s only going to get better and better. I think it’s personally I think it’s going to be a long, long time before AI generated video, at least on a mass scale is something that humans would record would not recognize as computer generated, I think is we’re a long way from that. I think that there’s just too many different cues and tics that show that something is not a real live human. So okay. That’s the that’s just explaining the interview, and how I generated that. And I think it’s kind of an interesting experiment. But it’s not how I actually use AI on a day to day basis. So I thought that I would show you a little bit about that as well. Because I think that’s interesting. And and you know, the biggest thing that I use AI for is probably reformatting text and turning text into, you know, lists and brainstorming and things like that. So I’m going to show you over here. Okay, first thing in order to, in order to promote this live stream, I wanted to create an event description. So the first thing here’s a simple prompts. I needed a title and an event description for a live stream to my audience where I demo how I use AI to create an entire text, audio and video interview with an expert. I will also show some advanced use of chat GPT included to increase productivity, brainstorm and reformat text inputs.
Alastair McDermott 24:07
So these prompts that you write are super important. And one other thing that I do that you may or may not know about is a thing in GPT, which are called Custom instructions. I’m just going to show you what my custom instructions look like. So the first there’s two questions and in customer instructions, what would you like chat GPT to know about you to provide better responses. And the second question is, how would you like chat GPT to respond. And so I say my target audience consists of subject matter experts, executives, business leaders consultants aged 4065. In the b2b Consulting sector, their primary goal is to amplify their visibility and solidify their authority. Despite the RAS experience, they grapple with content creation challenges and the aim to break beyond their local networks. They’re active on LinkedIn and regular podcast listeners. The main key points are lead generation inbound marketing, messaging, positioning content creation, and so I I put it in there. I’m not gonna read the whole thing but my content pillars cover establishing leveraging authority purse, brand development, podcasting, content strategy, social media, content marketing tactics. So have I have this filled out to give ChatGPT context, So I don’t have to give the context in every single conversation. That’s why I have that. Then the other thing is, how would you like ChatGPT respond. So I now this is something that doesn’t work so well, unless I tell you otherwise, always follow these rules, the unless I tell you otherwise, giving chatty pretty negative instructions. It tends not to not to follow those names, or actually, it doesn’t follow that very well. But I have given it the rules. And so it tends to apply these all the time. No filler or fluff right in a direct concise writing style, give actionable advice, avoid hyperbole or sounding hyperbolic. Avoid using these words, whenever you can family fam delve thrive, I find that an overuse is love these kinds of words. And so this is so so in terms of success. But it’s, you know, I want to give it these instructions. And then I have my tone of voice. The voice should be conversational, personal, open, engaging as if the author speaks directly to the reader using first person pronouns, vocabulary should be simple and accessible, make the text easy to read for a range of readers use very sentence structure. So I’m telling it how I wanted to write the output. And this is very much how I write.
Alastair McDermott 26:36
And so one of the ways that I got it too, I asked it to analyze on my text and tell me how I write. And so then I can describe it back to it. And so I have a spreadsheet over here, which is, this is like a kind of a prompts, texts prompts that I copy, paste. And I have a bunch of different ones in here. Analyze the writing style of the above author, and write about the topic in a similar way. You could also do analyze the writing style of this author and tell me how to describe it, you know, you can you can get that feedback. So I gave it some other instructions here. And it’s sometimes follows these and sometimes does not, I asked it to add an opening hook. So open the piece of the hook that touches on a relatable challenge, promises tangible value, or starts to an engaging question. So this is really important in all of your social media content, and all your content really, you want to hook people from the start, it’s a bit like the cold open on TV shows. You see, do you ever see the start of Breaking Bad? It’s Walter Weiss, standing in his underwear with a gun in the middle of the desert, and with a gas mask on his head. And you can hear sirens coming from from over the hill in the distance. And that’s the very first thing you see in the entire series of Breaking Bad. And it just piques your interest, you know, and then it cuts to, you know, 24 hours earlier, whatever this. And so, it’s the same with with films like movies, like any of the movies like Fight Club, there’s a guy tied to a chair at the top of a skyscraper, and then there’s bombs or whatever.
Alastair McDermott 28:15
And so you know, it’s all about grabbing attention. And that’s what your opening hook is. So I would suggest that you use opening hooks in all of your content anyway, it’s like it’s really important to learn, and then closing with a call to action, adding a call to action. So I’m telling ChatGPT Beatty, that I wanted to do these things. And it’s really good actually following these two instructions. So and you’ll see that in some of the content I show you, so Okay, so I told ChatGPT for this particular stream, I said, I need a title, event description for this. And I want it give me that. And then it’s given me the title, you’ll see that it’s actually called out the opening hook. It’s it’s told me what the opening hook that this is an opening hook. And then at the end, it’s got the call to action, CTA and it’s actually called that out as well. So this is interesting that it’s marking those out. So it is following the instructions when I give it that. So that’s one way that I use chat GBT in using those custom instructions to make the output more kind of effective, and not have to repeat yourself time after time. Now the other thing that I use it for a lot is I use it for brainstorming. And I’m not sure.
Alastair McDermott 29:26
Let’s see. Let’s see what I have here. Because I have some older conversations, see which ones might be might be useful. So one of the things that you’ll notice when I’m writing prompts for chat GPT is almost always the first thing I wanted to do is act like and I tell it to act like an expert in a certain field, quite often a b2b marketing copywriter. So I said here act like a b2b professional, sorry, a b2b marketing copywriter with huge experience in helping writers publish successful books on Amazon. First, I want you to help me renamed This book, I want you to give me a catchy book title, suggestions in the form of Title column longer subtitle, use the term super practical guide in 10 of the title suggestions, give me 15. Others. Here’s the working title. And I give the working title of one of my, my books or my booklets. And this is information about it. So I’ve given it lots of input. I’ve told it, so I’ve given it I’ve told it, how to act. I’ve told it, what I wanted to do, and the exact format that I want the output. And then here is, here’s the kind of the more and more context. And that’s where it starts to give. Now this is how I use it for brainstorm, because now I look at these and see, do I like any of those? And you can call out by number, Hey, I like number three?
Alastair McDermott 30:43
And number seven, can you give me variations on those? Or can you merge those and see, so and then it’s given me some likes gave me what I asked to give me some other other options as well. So that is, so I said, Give me 10 variations on this. Because now I’m iterating. So that’s the thing here is iterating on the on the previous output, and now I tend to go in in a very iterative process when I’m following with GPT. So I say give me 10 variations on this concentrating on switching up the subtitle only. And I gave it and it’s given me more. So that’s how I use it, to get ideas and to kind of iterate and get get more ideas. Now there is another thing that that you can do, which is quite interesting for brainstorming. And so I’m gonna pick, let me pick a topic.
Alastair McDermott 31:34
Okay, and then pick value pricing, value based pricing, because it’s kind of it’s an interesting topic I’ve interviewed Jonathan Stark. I’ve interviewed Christo, Alan Weiss, and then the grandfather of value pricing, Ron Baker, and I’ve interviewed all of those on my show. On The Recognized Authority, they are all experts in their fields. So I know I know those. So I’m going to ask Chachi Beatty, please. So I’m going to say I am inviting. I’m going to give it the list. So Jonathan Stark, Christo. Who else did I say Ron Baker? Otherwise, to a panel discussion? On value pricing? Who, who that? So I’m gonna say there’s, there’s also also to special special guests, with experts with expertise in this field. So this is I’m actually doing this live. Now I’m, that’s why it’s a little bit slow here. But bear with me. So. So the prompt I’m giving it is I’m inviting Jonathan Stark, Chris No, Christo, Braun Baker and Alan Weiss to a panel discussion about the pricing. There’s also two special guests with expertise in this field. They, they have a they have a heated conversation on how best to apply value pricing. To let’s let’s pick this, let’s pick something, let’s say to, to in a in a business in the context of selling, let’s say business coaching, services. Please write the transcript of their conversation. Now this this is an interesting one. So what it’s doing is it’s picking up from everything that it’s learned in its database. It’s picking up everything that it knows. And it’s stitching together. What Mike, Jonathan Stark, and Christo and RON BAKER And Alan Weiss is perspectives be on this on this topic.
Alastair McDermott 33:51
And it’s kind of auto kind of prompting or projecting what those could be trying to figure out who else might be people who will have an input on this. And then it’s, it’s realized, okay, so if I said heated conversation, then it’s got to have somebody disagreeing with somebody else. So what might be the kind of polarizing sides of an argument, and then it’s applying all of that context to business coaching. And so it’s going to come up with some really interesting ideas for us here. And so now I found this, like, I’m not going to use this verbatim, but what I might do is I might use this to give me ideas about a blog post that I might want to write about, about value pricing, or maybe about business coaching, you know, it’s, I’m using it as input. So and So here we’ve got right so Jonathan is talking about understanding the client’s problem so deeply, you can offer a transformative solution. Chris comes in about positioning as an authority. Ron is talking to about to us about outcomes. So special guests one special guests too. So I didn’t name them I should have, I should have told you I want them to name. So. So let’s, let’s see, who were who were the special guests. Let’s just see. And then, let’s see. Let’s assume the special guests are Blair Enns. Brilliant. And Jody grundon. Okay, so, so now we have some peoples and so that we know by like, these are possibly, you know, the top people in this space that might be worthwhile. So, so now I’m gonna, I’m gonna say, let’s say.
Alastair McDermott 35:37
So Perry Marshall, who’s also a kind of consultant, or an expert marketing expert. John Perry Marshall chimes in from the audience, again, and makes an interesting point, what did he say? Okay, so I’m saying to jack up now. So Perry, Marshall was pretending or Perry Marshall was in the audience, he makes an interesting point. So what did he say? And so he comes in and talks about the 8020 principle. So you’re all talking about value positioning outcomes, but don’t forget the 8020 principle, which is absolutely something Perry Marshall might say. And so this is interesting now, because this gives us some ideas. So okay, so we’re writing about value pricing, should I should I think about 8020 principle, and bring that in. So what this does is this, this, I just use this as a brainstorming tool to give me ideas. Now, let’s say I want to turn this, I’m going to pretend that I want to turn this into a blog post, I’m not actually going to use this as blog posts, I’m just going to use this as an example of how I might turn this into a blog post. So I’m going to say, take the previous take the entire previous conversation
Alastair McDermott 37:02
and turn it into a into a detailed blog post outline. And so it’s now creating this in the context remember, of my custom instructions for my audience, it’s, it’s using that. So I told her to stop generating there because it started talking about the panel discussion. So, so ignore, so I’m gonna say, I’m going to change my my input and say, ignore the fact that it was panel, don’t mention that. That or the people just include, just include the insights from the conversation. Okay, so I’ve realized that the output that it’s giving me isn’t quite what I want. So I’m going back and editing the, the one I’ve given it already, and asking him to change that. So now it’s gonna give me it’s given accordingly, custom instructions, give me an opening hook. And then it’s starting to give me the, the outline, section one with some clients problem, section two, talking about positioning, section three, focusing on outcomes, aligning interests, shifting the conversation cost value. Okay, so now I could say, Okay, now right. Now, right, the first, the first 500 words, including section 123. So now I’m telling you, I’m starting at the start, start actually writing the content. And so this, I put now is going to be a bit better, because you’ve you’ve, you’ve given a lot of inputs, and you’re kind of molding it. Now, I still wouldn’t be happy with this, I also will almost never include introductions and conclusions in the way that it does it get gets very formal in the way that it does that. That’s not my personal writing style.
Alastair McDermott 39:11
But that is something that chat GPT will almost always output. But it’s interesting to see how it how it gives us, it gives us some some things here like action steps. And the output here is actually pretty good. This output is way better than if I had just launched into, I want you to write me a blog post about value pricing in the context of business coaching. So it’s the output here is going to be far superior to that. So now what you can do is you can do other things with that. So one example would be rewrite that. Write that in the form of a 10 page linked carousel document for LinkedIn. So you’ll notice that I’m not giving it the act as instructions. Oh, it’s, it’s going in checking Canva, because I have Canva turned on. So it’s giving me some templates as well. So and it’s gonna show those. So it’s given me the, the, the templates here from Canva. Now, I didn’t ask it for those, but I guess I must have said some trigger words for that Canva plugin. The Canva plugin is not great for Chachi PT, what it does is it just searches the templates in Canva. But I guess that’s, that’s useful. But what I want to know now is is it going to forget the rest of my. So let’s, let’s say I’m going to tell it use the first one. This is this the first time I’ve been using that. So here’s how the content will be structured in a 10 page, LinkedIn car, so now it’s giving me the actual content. So the cover page that’s giving me the cover the title, the subtitle, the opening hook for page two, the text for page two, the text for page three, page four, page five. And so it’s given me that the text release. And what it’s done is, I’ve got the so I’ve got the blog post. And now it’s giving me this is giving me this as an as a as a as a carousel, it’s given me the text for all the pages. So now, this, this reformatting is how I use GPT. And these tools, but typically, the inputs that I give it are different. So quite often, what I’ll do is I’ll actually narrate I do this, sometimes when I’m driving, I’ll stick on record. And I will record while I’m driving about a particular topic. And so it could be a blog post, it could be some ideas for you know, for a book or for changing a book or something. And I’ll record that in the otter app, which is otter that I showed you earlier on. It’s the voice narration tool. And so I’ll take some notes in there some voice notes, and sometimes I’ll actually write the the chat GPT prompt, I’ll actually say that in my conversation in that I’m recording. So I take that. And by the time I’m driving, by the time I get home, it’s already been transcribed into text. So I’ll take the text from otter, and then I’ll put that straight in to Jack GBT as the input. And if it’s too long, because sometimes it’s too long, I’ll put it into Claude. And I’ll give that as the input. And then the output I’m going to get, it’s actually based on my expertise and my knowledge.
Alastair McDermott 42:33
And so that means that the output is going to be far superior to kind of the generic because what ChatGPT is doing to create the, the average content is it’s giving you back the average of all of the knowledge that it’s been trained on, and the databases that they’ve been trained on, they’re not perfect. It’s not the sum of all knowledge for the entire human race that some people think it is. It’s not magic. It’s it’s a training database that, uh, that has a lot of information in there. But it has biases in there. And it has incorrect information sometimes. So you have to be really careful about just using its input verbatim. But if you control the inputs more than if you get more input, what it’s going to give you out, it’s gonna be much better because of that. So one of the other things I did with Claude, because cloak can take more input. So let’s see if I can find this here. I took the entire text for my book, 33 ways not to screw up your business podcast. And I took the entire text of that, and pasted that into cloth as text. So that is I think it’s 26,000 words, 30,000 words, something like that. I think I think I skipped the appendix and those sections. And I think I’ll include those. Okay, so So yes, literally, the entire book text. And I said that this was when I was playing around, I was originally going to do an interview based on the author of the included book, me and the the interviewer that I talked about. And so I got to to write up an interview based on that. We might use that some other time, I don’t know. But that all of that those answers are generated from my book. And where I think this is going to be really interesting is when I as an author can give it all of the text of all of my books, and all of my blogs and all of my social media posts, and then say, okay, here are some thoughts on this topic. I want you to write a blog post based on everything that I’ve that I’ve written. Now create something in my writing style, and give me give me a detailed outline of that. Now, that’s going to be that’s kind of the next level of this. But it’s we’re still a long way from that.
Alastair McDermott 44:55
And you know that Chat GPT is doing ChatGPT enterprise We’re just going to be allowing for a little bit of that, you know, not just for an individual like me, but for a larger corporation where you can have these, Chris Do is doing something similar to this. So then there are other kinds of ethical issues like for example, if I was to use as input the transcript of all of the texting interviews that I’ve done, then I’m also including in that database, I’m including my guests in insights and information in there. And so the question is, then, you know, do I have the right to put that into there. And all sorts of, you know, kind of moral issues around that. So that gets really interesting. There’s, there’s a whole bunch of other stuff that you can do with GPT. And Claude, and I don’t want to make this too long.
Alastair McDermott 45:47
So one of the, one of the last things that I want to show you is just the two part prompt. So Oh, yeah, before I do that, just just on how to write like yourself, I have given it instructions. So right like me, writing their direct concise writing style, and use the following tone, voice vocabulary and sentence structure and tailor it for the following target audience. So I say tone conversational, it should convey a sense of openness, voice should be personal and engaging as if the author speaks directly to the reader using first person pronouns and a narrative style. Vocabulary shall be simple and accessible. Without technical jargon, or overly complex terms. This makes the text easier to understand for wide range readers. Now, I got this from ChatGPT, by inputting, I took a bunch of my texts. And I said, Please tell me about the tone, voice, vocabulary and sentence structure of the following author. And then I gave it my writing, I gave it like five blog posts, and it pulled back and told me what those look like. And then I said, Okay, now please turn that into a prompt to help you to write in my style. So that’s where that comes from. It still does not do it perfectly, but it’s pretty useful. The other kind of more advanced one I want to show you here is the two part prompt. And there’s lots of different variations on on this one. So I’m just going to show you how you might do it. So first thing is act as an expert in blah, and help me to create, and whatever it is that you want. So act as an X expert. And help me to create why I want you to help me brain dump and get all of the necessary information down and give me your response in two parts. And this is the key part of this and why it’s interesting.
Alastair McDermott 47:28
Part One will be an updated list in the form of a table you can do in a table whatever way you want to do that. The columns you can give it if you’re asking for tables, you can add columns and things like that. Part two of the response will be another question to help us further refine, and update. Now this is this is really cool. You can use this for a lot of stuff. I’m sorry, the last part is keep asking questions, until I tell you that all the items are fully up to date. Okay, I’m just going to show you how you could use this. And so I’m going to make this so act as a productivity expert. No, say act as a productivity expert. And help me create a concisely written book complete to do list. Okay. So I want you to help me break up again, all necessary information down, give me response in two parts.
Alastair McDermott 48:20
Part one of the responsible will be an updated to do list. And part two of responsible in other questions helps further update the list. And keep asking questions. So I’m just using this to create a to do list I’m going to I’m going to just make some stuff up for the purposes of this, this show. So okay, because I’ve got custom instructions turned on, it’s actually started to to create this for me. So this is where sometimes you need to turn off those custom instructions. So I’m just going to copy paste that prompt again. I’m going to go in and I’m going to custom instructions, and then they’ll say turn off. And now I’m going to do new chat. And now I’m going to use that again. So now what I’m doing is I’m asking chat GPT to to be my to help me create a to do list. So let’s do a tutorial. So we’re gonna have three things. So please go ahead and list all the tasks that you think so I’m gonna say get milk at the shop, cook dinner and record a podcast episode. Okay, so I’ve told it. Three things I need to do. So now it’s going to it’s told me it’s going to start organizing these into a table. Remember, this is the two part prompt. So it’s given me part one it’s given me my to do list of my three things I need to do. Now part two is has given me refinement questions. Do we have a specific recipe in mind? For the podcast episode, do you need to prepare a script? Are there any other tasks so I need Since I’m going to tell I need to invite a guest, I’m going to say, I need to get spaghetti. For, for the meal. And I’m going to say I need to, I need to get get petrol. And let’s gas for the Americans in the audience.
Alastair McDermott 50:25
So now it’s updated. Because this is again, this is a two part prompt. The first part is it gives us the updated list. And so now it’s what it’s doing. It’s given me the full list back. Now there’s four items on the to do list, get milk, cooked dinner, record a podcast and get petrol. And so now it’s asking me more questions about that. And so this two part prompt is really useful, because you can kind of use this as a brainstorming tool, when you’re creating content, or to do almost anything where you want to get stuff out of your head. And what it’s doing is you’re seeing the new updated list of items every time as you add more and more to it. Now, what I find with chatty beauty and any of these tools is as the conversation goes on, and the further you get from the initial prompt, the more degraded the possibility is of it getting stuff wrong, of it degrading its knowledge, I think it’s to do with the token length, that which is the amount of stuff it’s able to hold in memory, as you get further and further away from that initial in terms of the amount of text. So you could find that this only works, you know, you might be able to use this only eight times, or 10 times or something like that, depending on what the list is. But it’s a really useful way to kind of have that conversation where it’s, it’s, it’s almost driving the conversation. And the key to that is the two part prompt. So you say part one, so give me your response in two parts. Part one will be an updated output. And by the way, you can use this doesn’t have to be a list, this could be an updated blog post, or it could be an updated event description. So part one is an updated, whatever it is. And then part two will be another question to help us out further update and refine.
Alastair McDermott 53:53
Thanks for listening, I know that you’ve got a choice of podcasts and shows that you can listen to so I really do appreciate your time and your attention. If you did find this episode interesting. I would truly appreciate if you could take 30 seconds to rate the show in your podcast player or even leave a text review. It won’t take you long but it has a huge impact on the growth of the show. And it also helps to motivate me and continuing to do it. So it’s right where you’re listening to the show. You can also find a link in the show notes which will take you to rate and review. Thank you again. See you in the next one.
Alastair McDermott 52:03
And then you tell it, keep asking questions until I tell you that everything’s up to date. So that is super advanced in terms of of chatGPT prompting, I hope you find it useful. The other things that I’ve showed you here, I’m happy to if you send me a message, I’m happy to share my prompts and things like that. I use this a lot. It’s really important to control the quality by checking all the output and giving it as much input as you possibly can. So you’re not relying on it, generating stuff just from the database, because it’s really poor quality control. And it’s really good at faking. Like it knows that being confident that what it’s giving you is correct. So you’ve got to be really careful. So thanks for tuning in. I hope you found this one useful.
If you’re listening just the audio version, please do go check the video, I think you’ll find some of the video particularly the start were showed the autumn, the AI generated video, I think you’ll find that interesting. So I’d love to get your feedback on this if you found it useful. If you found any of the AI prompts useful. Please check those out. And I’ve also put up a bunch of stuff on Amazon in terms of my books and booklets. And some of those are free. So if you search my name on Amazon, have a look there and you’ll see some about consecration getting quick wins from content and some things like that. So yep, thanks for watching. Thanks for tuning in. Thanks for listening. See you next time.
