Episode 265.

Late Night Internet Marketing.

This week on the Late Night Internet Marketing Podcast, we're going to talk all about the recent release of ChatGPT 4.0. That's the letter O, not the number O. We'll get into that and more about how you can use ChatGPT and other tools like it in your marketing in 2024. All this and more on the Late Night Internet Marketing Podcast.

The Late Night Internet Marketing Podcast. Yes, you can do it right when it's late at night. At the end of the day, your dreams burn in your sights.

Keep it up and you will find that you're building your business one night at a time and now broadcasting late at night from a little studio in the big state of Texas, Mark Mason. Hey, hey, hey, how is everyone doing? I am your host, mark Mason, and I'm coming to you from the little studio in Dallas, texas. I almost said live, but we're not live. This was recorded earlier. If you're listening to this, it's because I recorded it earlier, not live, but it's great to talk to you.

I hope you're having a fantastic week, and this week I just had to talk about AI in marketing and how it's shaking things up Because, as I'm sure you've heard, several days ago OpenAI released Chat GPT 4.0. And by O I mean the letter O, which they use to stand for the word omni, which means that they're combining the large language model, the text part of AI, with vision and some other stuff to create a ubiquitous, omnipresent model and getting us one step closer to a world where AI behaves like a person, one step closer to general AI, and I tell you it's a pretty exciting announcement. If you haven't seen that demo by now, I definitely recommend you check out the show notes over at late night. I am dot com forward slash two, six, five. I'll post them there and make sure that you can find that original open a press conference Very impressive. So that made me think, hey, how should people like us best use AI in marketing? Because, after all, if you're listening to this podcast, you're a small business owner, quite likely, and you are doing your own marketing or you have a small team, and AI can be leveraged in those situations to enhance growth, to make your team more efficient, to innovate in ways that probably none of us have even thought about, and I was thinking about this today. It's really changing the future of content creation. A lot of people, I think, they fear AI because they don't know what to expect, in the same way that factory workers feared robotics, that they think change is coming and people, I think, naturally fear change. And today I recognize, boy, it really is coming, because I have this problem in my house. This is kind of an aside related to AI, but I have this problem in my house. This is kind of an aside related to AI, but I have this problem in my house.

I have a lot of home automation and, generally speaking, in my home a lot of the automation is a little bit outdated in the sense that the protocol that's used for some of the switches in my house are based on a technology called Z-Wave. Z-wave is not the most modern protocol. It works great, but it's not the most modern protocol and generally speaking, it's not strictly true, but generally speaking. Homekit, which is super security focused, does not play well with Z-Wave, and so one of the things that you need to do is you need to put stuff in the middle to get Siri HomeKit to put stuff in the middle to get Siri home kit to talk to Z wave devices, and you can use hubs, and there are several of these home automation hubs on the market. So I asked chat GPT just on a Lark what it thought were the best home automation hubs and why. I just just asked it that question and it created dynamically for me essentially what amounts to a review article with pretty up-to-date information. I was actually quite surprised and that was super helpful to me and I had absolutely zero need to go to Google and look at this. I just went straight to Amazon to look at the different things that it recommended. I bypassed Google and blog posts and the internet completely. So this is going to be an interesting thing as it develops. I think a lot of the things that have traditionally helped us get visitors to our websites and to our content. I think a lot of that's going to change us get visitors to our websites and to our content. I think a lot of that's going to change, and I'm not exactly sure how, but I know Google, gemini and tools like ChatGPT are going to play a big role in doing that. And then you know there's other ways that this is impacting us. The other day in my day job, one of my day job responsibilities is to move AI forward in my part of the company, and so I was running some experiments with AI, and not just for content creation but for actually doing work.

So in my business in the electronics industry, we have a lot of industry standards that we need to comply with, and they're long documents with numbered paragraphs that sound a lot like laws when you read them, or legal documents, and we have to comply with those and we have to show evidence that we comply. So the way that we do that is internally. We have a list of rules that we use to run the company, our quality specs that we use to tell us how we have to do things, and they're written in such a way that they automatically mean that if we follow our own rules, we'll comply with this pile of different standards that cover different topics. And so one of the things we have to worry about is are all the requirements called out in a big stack of standards, are they all comprehended and covered in our own internal rules? And so I ran an experiment where I gave actually, in this case it was Claude Opus I gave Claude Opus a stack of industry standards and I said identify every requirement that applies to my company that's in this giant pile of standards and I'm talking about, you know, a hundred or 200 pages of of gobbledygook standards. And I fed it to Claude Opus. I said identify every thing that we need to worry about in this pile.

And then I asked Claude Opus now read my internal specifications and identify where we comply in our internal specifications and tell me how we comply and if there's any gap between what we're doing according to our own internal specifications and what we're asked to do by the industry specifications, and report that as a table. And it generated this giant table of items that we needed to go look at. And either, in most cases there was no action and it explained why, but in a couple of cases it interpreted these industry specifications in a way that made us go hmm, maybe we need to look at that. This is very interesting. Now, this is the kind of work that people have been doing in this space for decades, but it would have taken days and days to do that with a human, and AI did it literally in 20 minutes. It was quite amazing and the results came out as an Excel spreadsheet that then this team could go line by line and validate what the AI had said and take actions to close gaps.

So all of this and more and if you're really interested in AI, you know this is not an AI podcast, this is a marketing podcast. If you want an AI podcast man, have I got something for you? My very good friend podcast man. Have I got something for you? My very good friend, michael Stelzner, who is the world famous owner of Social Media Examiner and is the host of the Social Media Marketing Podcast, which has been going on. All of that has existed for more than a decade. Thousands of people go to his social media conference every year. He has started a new podcast called AI Explored. There's going to be a link to that in the show notes over at latenightimcom forward slash 265. And I definitely recommend that you go listen to that.

Michael Stelzner is one of the smartest guys that I know I know a lot of smart people, okay. He is insightful, and sometimes so much so that I think he can kind of see the future. He's very good at anticipating things to come, and so I'm really excited about everything that he's got going on over at the AI Explored podcast. Go check that out. Okay. So let's talk about three different things. Let's talk about what you can use AI for today to help grow your business, what you should avoid, and then, as a third thing, let's talk about creative ways that you might consider using AI as we move forward into the future.

So obviously, the thing that everyone talks about in fact, we've already talked about it on this episode is the use of AI tools for content creation, right? So this all started. You probably remember Jarvis, which I think Disney forced them to rename. Jasper Jarvis. Ai, and later Jasper AI, was the first version of these large language models to help people generate blog posts and copy and social media content, and we all kind of looked at that and said, yeah, that's kind of interesting, but it's not that great. The very first versions of Jasper weren't great. They hallucinated a lot, which is what you say in AI when you talk about a large language model making things up, and their language wasn't very good. And, quite frankly, ai created by those tools was pretty easy to detect, required a lot of editing, so I thought those were good sometimes for creating lists of interesting ideas, but not really all that great for creating content directly. Of course, jarvis is much better.

Now there are other tools like WriteSonic and many other tools. A lot of people are using ChatGPT and Claude to create things directly, to create things directly. But I think you know when you have a purpose built situation like you want to create content specifically for blogs for a specific purpose, I think a lot of these specific tools can do a better job than just having you sit in front of chat GPT, even though they're using chat GPT and Opus as their backend. And the reason is most of these tools, the way they work is they will ask you questions about what you're trying to do, specific questions as an input to the process, and then they will use that to engineer the prompts that they're using on the backend of this process so that they can get a really, really good result with a specific level of quality and specific format. The best example that I know of that is my buddy Ray Edwards tool called clarity scribe AI. It is a copywriting tool for copywriters and it writes amazing copy and it uses the framework that Ray teaches for copywriting, called the pastor method P A S T O R. We've talked about that on the podcast many times and it creates amazing copy.

And of course, the benefit of this is speed and cost. Right when I started creating content, we were paying one penny a word to get things written offshore that we had to edit heavily and Jarvis kind of put those people out of business. And now tools like ChatGPT are putting real writers out of business in the sense that the content is starting to get so good that a lot of people it's good enough for what they're doing, and you can increase your blog post productivity and output by sometimes a factor of 10 in some cases, and for me, the difference is just between starting with a blank piece of paper or starting with an outline that is created by chat GPT and even if I throw half of it away okay, even if I junk half of it I'm still way ahead of where I would be if I just started with a blank piece of paper. So this is a real thing and I think you know content creation is something that is here to stay. With regard to AI and, in fact, the Content Marketing Institute just released a study recently says that businesses that using AI for content creation have seen at least a 60% increase in efficiency in their content creation processes, and I'm here to tell you that as AI gets better and better, that's going to go up another 40%. They'll get 10 times more efficient than they were.

At the end of all this, the second thing I'd say is that can use AI-driven tools to improve SEO, traffic and audio engagement. One of the interesting ways this has been happening is SEO tools are now using AI to decide how to recommend for you what to write about, and then write it with SEO in mind. Then write it with SEO in mind. So, for example, surfer SEO, which is a pretty famous SEO tool, now has a tool that uses AI to analyze the top ranking pages and provide database recommendations on exactly what to write, and then uses AI to actually write the article with all of the SEO in mind. That's really cool. I haven't tested that, but I hear people are getting fantastic results, at least for the moment, when it comes to ranking, and, again, that's not any different than what we've been doing for years. We've been telling people for years hey, if you want to rank number one for a keyword, go look at what your 10 competitors are doing and do that, but just do it better. And that's what Surfer SEO is doing. They're just using AI to do that. So, instead of taking a week to create a really strong blog post that's designed to rank for a particular keyword, you can do it in a minute and that's really amazing.

Clearscope is another tool where you know it takes content and tries to make it more relevant by suggesting keywords and optimizing the article so that it will rank better. It looks at your content with regards to seo and makes recommendations. And you know, uh, people have reported this isn't just for blog posts. Works for works for YouTube as well. We've heard reports of people increasing views by 40, 50% using Surfer SEO, and I know one of my buddies actually hires a consultant to help understand what the title of YouTube videos should be, what the thumbnails should be, and I know those consultants are using various kinds of AI to help them win at YouTube SEO, and they're layering on top of that their own expertise. But I think increasingly that expertise is just more validating what these tools are coming back. Clearscope has a study on their website where they had a user report a 30% increase in organic traffic just by refining their blog content using the ClearScope tool. Brightedge said that businesses that implement AI-driven SEO strategies are seeing 50% higher click-through rates, which is very interesting, as the AIs modify the call to action and the pre-selling that happens in those posts to get the click-through All super interesting ways.

And then the other way that you can use this is in analytics. You know Google Analytics has always used machine learning not always, but GA4, for sure, has used machine learning, which is kind of the underpinnings of things like ChatGPT. So ChatGPT is a large language model, right, it's a program that recognizes and generates text and does some other things too, and it's trained on a huge set of data. That's where the large part comes from. So it's a large language model because it's built on a huge set of data. Llms, these large language models they're built on top of machine learning and specifically there's this kind of neural network technology underneath this, called transformer models. So, simply, these large language models have been fed enough examples to be able to recognize and interpret human language, and we use machine learning, called deep learning, in order to take advantage of that. So that's how all those things are related. Google Analytics has been using this underlying machine learning technology to see patterns in data for a very long time, and that's how they do things like optimize who sees your ads and stuff like that.

Hubspot and Marketo both of those are really good places to learn more if you want to see how AI is being used in marketing. There are lots of good articles over there. This applies to email, too, and we've heard of people doubling their email open rates with HubSpot using their personalization algorithms, where, instead of segmentation, they dynamically personalize the email kind of person by person. That's kind of where we're headed right, instead of sending 10,000 emails and segmenting them into two or three segments, like in my case one for people who haven't started their business yet, one segment of new business owners and one segment of people who have lots of experience in online business. Why do that? If I've got enough data, why don't I just have the AI personalize every email to every subscriber before it goes out with stuff that I know about them based on blog posts. They've read purchases, they made things they've clicked on. That's where we're headed with email marketing. So lots of opportunities to use these kind of tools. We're just at the very, very beginning of that, so that's how we can use them.

Today let's talk about the mistakes that we make, and the obvious one is people want something for nothing. They over rely on AI to create content. They lose their personal touch and really the authenticity of their voice, and really that's our only defense against where AI is going. If you want to differentiate yourself from everyone else who's using AI, when AI is at the limit and can create content just like you, your personality and the authenticity of your voice, that's pretty much all you're going to have left. And we've heard some examples of companies that turned AI loose to answer their social media posts and people detected that and there was backlash there. People started unfollowing. Now AI is going to get better and I think eventually you won't be able to tell. I really do believe that, but it's something you need to be really careful about. And then there's some ethical and legal considerations. We're probably going to see some legislation eventually where you've got to disclose. If you're presenting yourself to be human and you're not human, you'll have to disclose that. I would imagine that the Federal Trade Commission or other regulatory bodies will start putting rules in like that.

And then there's always the question of if AI creates a blog post. What's the copyright situation on that? I mean, I didn't create it. Ai is creating it based on other people's work that it read on the internet, so it's derivative. That's usually okay from a copyright standpoint, but does it belong to me? Does it belong to open AI? I think there's a lot of work to do there in the legal space, and I think Gartner, which is one of these data reporting companies. They did a survey and reported that 30% of businesses have faced some sort of legal challenge due to AI, and I know in my day job, our lawyers are all over this, making sure we understand what the legal implications are of the use of AI, not only from a data security standpoint, but also from just the aspect of who owns what's generated. All of those things Pretty interesting.

And then there's the issue of quality control. I think a lot of people are putting out AI generated content that contains inaccuracies, because they're writing about stuff that they don't know anything about, which is a feature of AI. It allows you to talk about things you don't know anything about, which is a feature of AI. It allows you to talk about things you don't know anything about. We call those hallucinations. In the AI space, the AI hallucinates something that's not true. That's getting better, much, much better, with the current models, and one of the really cool things that we've been experimenting with in my day job are these so-called RAG models for AI, where each thing that the AI says has to be annotated back to a particular piece of data in a controlled data set.

I think that's a really powerful idea. So here's an idea. Here's a place where you could differentiate. Let's say that all of my podcast episodes were transcribed and fed to an AI. You could ask Mark Mason a question that AI could answer based on things that I had said, or maybe things that it could determine based on what I had said, but only things that I had said in previous podcast episodes. Then you start to talk about personalized AI, where my AI gives a different answer than chat, gpt or someone else's AI. That's really interesting.

Now I think you know there's all kinds of questions about people generating content that then gets busted by AI detectors kind of like we used to do with copy scape back in the day and I tell you my son has even had this problem. This is hilarious. My son's a really good writer. Copyscape back in the day, and I tell you, my son has even had this problem. This is hilarious. My son's a really good writer. He's a junior in high school and that's just one of the things he's good at. He reads voraciously. He's probably read a hundred books this year. He's into spy novels, he loves to read, and so he tears them up. I mean, it costs me a fortune to keep him in books when he starts ripping through one of these spy novel series I'm talking about, like Tom Clancy, that kind of stuff.

So he was in English class and he had to write an essay. So in their class they have Chromebooks which makes me cry because Chromebooks are horrible, but they have. They have Chromebooks which makes me cry, because Chromebooks are horrible, but they have Chromebooks and so they have these special browsers where you can't get to the internet and all this stuff. So they're locked down and so he writes this essay in some special software that monitors you while you're right. And he turned his essay in. He wrote it in class. Class is an hour and a half long. It's a Tuesday, thursday class, really interesting.

His high school is set up kind of like college and he has classes that he goes to every other day. Anyway, he writes this essay, turns it in. It's I don't know 1,500 or 2,000 words, something like that. He turns it in and the teacher calls him out in the hallway because the software flagged it for AI creation. Here's the thing. Right, he was on his school Chromebook. Ai is blocked by every firewall you can imagine. The browser is locked down. He looked at the teacher and he's like and how would I do that? Exactly? And I think she didn't even really understand what the AI tool was telling her what the technology did. I mean, she's just like way behind. She didn't understand that it was impossible. So she's like oh, I didn't know that. But he got busted for writing AI content.

So the story is Google's probably going to make that same mistake. I think, as Google tries to sort this out and remove the AI content, I think we're going to have some problems with noise in the rankings because Google is going to think some really great content is too AI generated and try to demote it. This is going to be a challenge. I don't know what Google is going to do. You know Google is facing some challenges here and it's going to be really interesting to see how they continue to maintain their ad model, how they pay for content that they're using to train their models. All of this is really going to be interesting going forward. But getting busted by Google is definitely something you need to worry about, just like my kid, zach, who got busted by his teacher for doing nothing. Just to be clear, we're maintaining Zach's innocence he did nothing.

Okay, so let's talk about some unexpected uses of AI. I think this is where it's to your advantage to not just use AI like everybody else is using AI, but to get clever and use AI in clever ways. We have all kinds of tools now that are creating pictures from text, and I think what's now are tools that create videos from text. In fact, one of the things that I've been experimenting with. Some of you may know that this podcast, in addition to being an audio podcast, which is how the vast majority of you find this podcast, I also release it to YouTube now, and there's a video of my talking head looking into my camera at my desk and that's interesting for about 13 seconds. And so I've been experimenting with B roll, mixing B roll in and having the B roll be relevant to what I'm talking about. Well, mixing B roll in and having the B roll be relevant to what I'm talking about Well, that's a perfect job for AI. So those kinds of things where you're synthesizing video content from text or other content, that's coming, and some of those tools like pictory are already starting to do that. There are tools out there that are generating AI avatars for video content. I think this will also be the thing. We're already starting to see companies who have customer service avatars that will talk to you with a talking head that's fake, you know, kind of a max headroom sort of talking head. So this is all coming. How that content will rank in YouTube is a question I'm not sure.

I've talked many times about Opus Opus Clip, which creates these little short clips out of the content that I create for publication of social media. It does that mostly automatically. I'm using Descript to record this podcast, and one of the things that Descript does that's really cool is it will remove retakes. So if I say something like oh oops, I messed that up. So one of the things Descript does really good is it removes retakes. Okay, so you see how that was a retake of a retake. Descript will detect that and remove it out. And in fact, to get that in the final copy, I'm going to have to go put it back in, because no doubt Descript detected that as a retake and it just cast it to the side.

And then another use is an example. I'm starting to see things like baseball videos where you can use AI. So you know, if I record a baseball game, I'm interested in just the times when the ball is in play. I don't know what percentage of the time in baseball the ball's not in play, but it's most of the time. And so what I usually have done in the past especially like when my son's pitching and I want video of every pitch is I have a keystroke sequence that I use and I scrub through the video and manually, one by one, delete every sequence between each pitch, so I'm left with just the pitches. Well, that's the perfect use for AI. So all of that kind of stuff is coming, and we've all been using Dolly and Art Breeder and tools like that to do this. So that's all coming and I recommend that you explore that.

I think the other thing that's coming is more interactive content. So let's say I go to your review article about home automation hubs and I've got a question. Well, wouldn't it be cool if there was an AI bot right there on the page that I could interact with to get your opinion about my question? That would be really neat. It makes a lot of sense and I think AI coaching is also something that's going to come. I've been experimenting with this with the current version of chat GPT in my car. This is a really cool way to use AI.

Next time you're driving somewhere, instead of listening to classic rock on the radio and hearing a song that you've heard 50,000 times, put your earpiece in and start talking to chat GPT about a problem that you're working on. For example, let's say you've got a project, get chat GPT to help you break that project down in steps to identify resources that you need to think of, creative ideas to create to do lists and all these kinds of things, schedules, to write emails to people that need to help you all of this stuff and then, when you get to your destination, bring up chat GPT on your laptop and right there will be the transcript of that entire conversation. Cut, copy and paste the stuff that you want. I think that Siri, when it's announced at WWDC in June, is going to have that kind of capability, and I'm very excited about that. That's my guess is that, finally, siri 2.0, which, for all I know, will even have a new name, although I doubt it, but might have a new name it will perform like chat GPT 4.0, which is what you're seeing, what you saw in that video above that I mentioned from OpenAI. Lots of exciting stuff you can do, but my tip there for you is start talking to chat GPT like it's a person and ask it for advice Super interesting.

I'll tell you another really interesting use that you can use chat GPT or other AI tools for Go, do a survey of your audience and ask them open-ended questions. Get as many pieces of feedback as you can and, instead of asking these one through five questions, which have a lot of bias, ask good questions like you would ask at a coffee shop, and get people to respond to them by typing Now you can't do too many of those because those take a long time, but ask them two or three questions, ask hundreds or thousands of people, and then feed that data into a large language model and ask it to generate a report regarding trends and ideas and summaries of the data. You'll be amazed at the capability of these large language models to pull big ideas out of open-ended content like that. It's really quite remarkable. I definitely think that you can start using more surveys in your marketing which surveys by themselves, or quizzes, can make great lead magnets and then use AI to both improve those for better conversion rates and to extract insight from the data. Really interesting stuff.

So what have we talked about today? Really, we've talked about the fact that AI is coming and, instead of being scared, we should be excited. We should immediately enhance our efficiency with AI and check out Michael Stelzner's podcast over at AI Explored. There's going to be a lot of discussion about that. We talked about the fact that we need to avoid the common mistakes and pitfalls that people seem to be falling into with AI these days, particularly just spewing AI garbage out there. That's just not going to work. It's not a sustainable business strategy. And then, finally, we talked about a few just a few of the many novel and unexpected use cases for AI.

I hope this is helpful for you. I hope, if you have a question, you'll send me an email at feedback at latenightimcom. Tell me how you're using AI, tell me about the ideas that I missed that you think I should have covered, and I would love to do a second episode based on all of the things that are going on that I didn't even mention in this episode. Until that time, I hope you have a fantastic week. We'll talk to you next week. I hope your business is amazing between now and then, and I hope something in this helps you make it even more amazing. Ciao.

You can do it right when it's late at night. You've been listening to the Late Night Internet Marketing Podcast. Be sure to visit LNIMpodcastcom today to leave feedback for Mark, Download special bonus content, access the show notes and more. See you there. Until then, go and make some great progress on your internet business one night at a time. One night at a time.

Yeah, this AI stuff is amazing and my advice to everyone is understand change and adapt to it. Don't fear it. Look for opportunities in the change. That's how people have overwhelming success is they see disruption as a way to move forward and do amazing things that have never been done before, to do them before everyone else. Take that mindset. Don't spend time thinking about how woe is me, AI is going to take my job. Spend your time asking the question that Cliff Ravenscraft taught me how to ask All this AI stuff what does it make possible, right? Don't spend time fretting over what is gone. Think about what is to come. It is an amazing time to be alive and this makes an unlimited number of things possible. We haven't even started scratching the surface, Ciao.