Melissa Rice (00:00):
Welcome everybody to the 30th episode of SEO for Bloggers. I’m your host, Melissa.
Arsen Rabinovich (00:05):
Wow, 30th. Sorry.
Melissa Rice (00:06):
I know we’re middle-aged. Today, we’ll be talking about AI-generated content and ChatGPT with experts, Casey Markee, Arsen Rabinovich, Andrew Wilder, and special guest, Brett Tabke. Brett is the founder of Webmaster World and Pubcon, the longest-running educational conference series in the SEO industry. He’s a well-deserved recipient of the US Search Awards Lifetime Achievement Award.
He is a big believer in community building, so naturally, we’re very excited to have him sharing his decades’ worth of industry knowledge with us today. Brett, thank you and welcome.
Brett Tabke (00:44):
Thank you. Appreciate you having me. Thank you so much.
Melissa Rice (00:46):
Appreciate you being here.
Brett Tabke (00:47):
Happy to be here.
Melissa Rice (00:49):
Everyone else joining, thanks too. Just a reminder, we will be having a Q&A at the end, so please feel free to drop any and all questions in the Q&A section, which you can find below. If you haven’t already, let us know where you’re tuning in from the chat box. Let’s get started, shall we. Andrew, a lot of the attendees had the question of just where do I get started? For someone who knows nothing about AI, where or how would you recommend them starting to use it?
Andrew Wilder (01:16):
Well, I think the first thing to realize is AI is already here and you’re already using it. For, I don’t know how many years you guys would know Google’s been using AI to process search results. AI means a lot of different things. I think what we’re going to be talking about today is the tools where you can have it generate content mostly, where you’re interacting with AI directly and giving it prompts and saying, “Make me this.”
The most popular one right now, the two things I’m going to mention are coming from a group called OpenAI. They’re basically sort of spearheading this, I think it’s a nonprofit actually. They’re developing these tools and they’re releasing to the public and everybody’s losing their minds. Last year, they released DALL-E 2, which was a image generation program. You can give it a prompt, you can say, “Create a 3D rendering of a dog on a skateboard, and in about 15 seconds you get four versions of that.”
Yes, thank you Garrett. Microsoft’s been funding OpenAI and ChatGPT. It takes a tremendous amount of computing power to do this. That 15 seconds, it blows my mind how much it’s actually doing there, but then you can actually then take it and have it change the image a little bit. You can say, “That image is good, but let’s change this part and remove the background or things like that.” DALL-E 2 is the image generation program. I’m going to drop a link in the chat. I’m mixing my words.
Melissa Rice (02:42):
Andrew Wilder (02:44):
Then, more recently, what burst onto the scene was ChatGPT, which is a text-based chatbot. You can give it a prompt, you ask it a question and it’ll give you an answer. You watch it, literally write it on the fly and then you can give it another prompt. The key to this is it remembers that conversation. Everything you talk about in that thread, it’s using all of that information. Then, actually as you go, it’ll save that chat and you can start a new chat, new session.
In that session it’s referential. If you say create a paragraph on this topic and then it does it and you say you’re thinking, “That’s good, but it’s too long. You can literally just say shorter and it’ll rewrite it shorter. You don’t have to give it the prompt again.” It’s iterative. I’m going to drop that link in the chat too. I think you need a free account to log in through these tools to use them. You can just create an account and log in.
If you haven’t tinkered with it yet, I suggest you wait until you have a lot of time, because it’s a rabbit hole you’re going to start and then hours later you’re going to be like, “Wait, it can do that too?
Melissa Rice (03:44):
It’s a rabbit hole [inaudible 00:03:47]
Andrew Wilder (03:47):
It’s a rabbit hole. It’s both exhilarating and terrifying at the same time. There you go.
Melissa Rice (03:53):
It’s really cool to see other people generating stuff. I think there’s like Reddit threads too that I’ve seen. I love everybody’s ideas. It’s fascinating. Another major concern of everybody was just if Google can detect AI or people using ChatGPT to create content. Brett, is this something that content creators should be worried about now or in the future, anytime soon?
Brett Tabke (04:19):
The standard SEO joke these days is it depends. They came out today and tried to clarify with a blog post what they were saying and it was clear as mud. They never actually said whether they can detect it or whether they want detect it or if they’re trying to detect it, but there’s about 10 different really good AI content detectors out there right now. Most of them have varying degrees from a little bit AI, kind of AI, mostly AI, and there’s all kinds of ways around it.
The question is, should Google actually care? What are they actually going to care about? Are they looking for you and me throwing up a blog post and running it through ChatGPT and then tweaking it to ourselves and putting it on our blog? Day after the person trying to write a million product descriptions on an affiliate site with ChatGPT. They’re probably going to flag the latter, but I just don’t think Google’s going to care a whole lot about it.
The default value of ChatGPT, the writing level is higher than the general writing level on the web. It’s actually going to improve the writing on the web if it’s used properly, so why should they care? Why should they care is my question. If it’s going to make things better for them, why not use it? Should we be worried about it? Yeah, I’m going to take steps around it if I do use ChatGPT.
I’m going to run it through an article spinner. I’m going to run it through Grammarly to tweak it a bit, then I’m going to throw my own spin on it, but I won’t be putting it up raw. There we go.
Melissa Rice (06:00):
I think they said in the article you’re talking about, use it responsibly, and I’m using finger quotes. Casey, where is ChatGPT actually getting all of its information from? We had a lot of people wondering, are they mining blogs for information, et cetera?
Casey Markee (06:21):
No, it’s just a specific room in a garage in LA. It’s illegal aliens mostly. They’re not here on good pieces, no.
Brett Tabke (06:31):
Monkeys at the keyboard.
Casey Markee (06:32):
No, we mentioned this kind of when Andrew was talking, we call it generative artificial intelligence. It’s a large language model. It was created by OpenAI. It’s trained on a diverse range of texts found on the internet websites, books, social media, podcasts. We call this a corpus and it’s basically the cutoff date of all the data was 2021. Literally, there are things you can put into ChatGPT, and it has no idea what you’re talking about because it only has information up until 2021.
That means the information available comes from sources that existed on the internet until 2021 and it’s not equipped with more recent information. Despite that limitation, and if you ask me, it’s not a huge limitation, but it is, if you’re asking it to generate a list of most recent tools, there’s a lot of tools that won’t be on there. It’s capable of generating answers that are both helpful and informative and it produces those answers in a very human-like response that makes you think, “Oh my God, it’s like I’m talking to a real person.”
I think that’s so stunning about what we’re getting in return is that it’s very nice. We can even change the tone. People can play around with it and say, “Can you generate me complete response to this as if I was talking to Mark Twain.” Could you use this in a way that would be like I was talking to a 10-year-old and that’s helped me considerably because I’m just not very bright. It’s good, I’ve been able to do that, dumb things down, especially for my kids. It’s pretty interesting. That’s how it works.
What it is a generative pre-trained, transformer fancy way of saying that this is just a large language model, it’s just something that’s trained on a large amount of information and it generates things one word at a time. It is predictive auditing, so to speak. It isn’t really true AI, it’s just predictive alliteration.
Melissa Rice (08:32):
Well, it does feel like we are talking to people and then there’s the fear now that, “Is all that kind of content going to create no room for actual content creators who want to stay creative? Brett, is this the beginning of the end of blogging as we know it?
Brett Tabke (08:48):
No, I don’t think so. I think it actually helps us quite a bit because one of the good things the ChatGPT one they’re perfect at is finding your blind spots. I’m sure we’ve all been in a meeting where the boss comes in and says, “We’re going to do project X and now we need all these great ideas on how to execute it.” The room goes silent. Nobody has a single idea what to do.
Then, somebody says, “Well, if we’re going to build roads or something, maybe we ought put up some street signs or road markings or something.” Then, you’re off to the races because somebody come up with a dumb idea. Well, ChatGPT is that guy in the room. He comes up with nine stupid ideas and one really good one that you never thought of.
Casey Markee (09:30):
Brett Tabke (09:31):
I really think it helps with creativity. Then, when you go back to what Casey was saying there, what was it trained on? Well, ChatGPT it started on Wikipedia, it took all of Wikipedia and then it took all the references from Wikipedia. Those have all been hand-filtered, human-filtered. It’s like the same way when PageRank was started what they do to seat it, they seated it with Yahoo.
They seated ChatGPT with Wikipedia, and then they went to WebMD and then they went to all the programming places, PHP Perl, all of the authoritative documentation on those programming languages to get that one central transformer. Now, the creativity part of we’re we’re ChatGPT is just awesome, is you remember in the Matrix when it was the first big movie that did the morphing that was able to morph between and come up with a picture in the middle between two pictures.
That was the amazing part about The Matrix. Well, that’s what ChatGPT does. It has these two things. It’s got something off of Wikipedia and then it’s got something off of WebMD and then you ask it a question and it tries to predict the middle. Then, from there it kind of tries to predict what should come next. On TikTok the other day, there was a doctor who asked it, “Diagnosed this patient and so it diagnosed it and it was a rare thing. It was right.
Then, it said, “Well, where did you figure that out at ChatGPT give me some sources.” It just made up a whole bunch of sources didn’t even know how it came to be.
Casey Markee (11:06):
It was like I was having an argument with my wife in real time.
Brett Tabke (11:09):
Casey Markee (11:10):
I came to the right conclusion, but I could not support my argument.
Brett Tabke (11:15):
I think creativity’s out there and effect it enhances us. I think Microsoft’s on the right idea here, calling it a co-pilot kind of an assistant to help you in everything you do. I think it helps our creativity actually.
Melissa Rice (11:27):
Awesome, I also love that Matrix reference. Thank you for that. Arsen, how do you see AI being most successfully used in food blogging?
Arsen Rabinovich (11:39):
That’s a good question. I see two possibilities right now that can be helpful to bloggers and that I am comfortable with them using the first one, and we have a team member who’s all about AI. Dan, if those of you who work with us, who Dan is, he’s been doing a lot of stuff with AI. The first thing would be to help with content topic ideation. It sucks because we have a service when we do topic ideation and we sell this service.
Let’s say you want to create a topical silo of apple pie or apple based desserts and you want to figure out like, “What are complimentary recipes to apple pie?” You can have ChatGPT come up with that for you. We actually did it, I have notes here, and then right away came back with caramel apple tart, apple crumb cake, apple galette, I don’t even how pronounce that. Apples strudel, apple dumplings, baked apple cider donuts, and gives you descriptions for that.
As a way to help with content ideation, topic ideation, I think it’s a really good tool for that. The second one is identifying entities or missed entities in your writing. You can take your entire post and drop it in there and then ask it to identify any missing entities, and we did that. We’ve done that. I think the prompt are any key entities missing from the content I submitted. Then, it came back with cooking time, oven temperature, amount of butter, amount of cinnamon.
All of these things are useful and for those of you who have gone through our coaching program, we talk about SEO Russia’s content tool, which helps with identifying those missing entities, ChatGPT will do that for you. Those two things I’m very comfortable with food bloggers using right now.
Melissa Rice (13:47):
That’s to get started, but Casey, for people looking to reoptimize, what are some ways to reoptimize blog post they have using AI or ChatGPT?
Casey Markee (13:58):
That’s a great question. Basically, I’m going to just going to paste it over here, but there’s a fantastic amount of stuff that you can do if you were just to pop that in, “The ChatGPT” Man, it’s not allowing me to paste. I think that the problem is that my responses are so awesome that I’m not able to pace everything fully into it. We’ll do smaller and see if that works.
Look at that, there we go. The issue is too big. I’ll keep my jokes on that to another day. We’ll just point that here and there you go. If you were, for example, take a look at this, this is if you were to paste over, if you were to ask ChatGPT, what are some of the best ways to optimize blog posts? This is what it will spit out to you and they’re very good answers. You get keyword research, you can use ChatGPT to generate a list of relevant keywords and phrases related to your blog topic.
You can use it for content generation, you can use ChatGPT to generate high quality unique content for your blog posts. You can use it for titles, you can use it for headline optimization, you can use it for readability analysis. You can use and use it for link suggestions. You could go in, you could find a related link that you like and say, “I can’t get a link from this site. Could you give me a list of other possible link targets that I could reach out to for?” Voila, there’s 7 to 10 options right there.
I have shown this online to multiple bloggers this week during their live audits, and they’ve all just been blown away at how much, “Oh my God, I really like this.” One of the really cool prompts that we’ve been using regularly is put in these three ingredients, please generate me a list of recipes that I can use for these three ingredients.” Then, it will literally print out 7 to 10 options for you to make these recipes. Then, the bloggers look at that, “Oh my gosh, this is saving me so much time.”
Then, they could just expand those recipes out, or one of the things that you really want to think about is saving yourself time, taking back more of your schedule. Pens, we all have pens, we all have to do a lot of detailed pending for Pinterest. If you’ve got a photo and you’ve got to write five pen descriptions, I’m going to take one pen description, I’m going to go over to ChatGPT, and I’m going to tell it to spin that and provide me four other related pin descriptions, all that says the same thing, but in different ways.
Then I’m going to paste those five options back into my individual pens. Saves you a ton of time and that’s just the tip of the iceberg. I’m going to go ahead and paste over a link to some really great options for you to see if I can find it here. I’m going to share a cheat sheet from the firm Neural Magic that provides a great overview of all the different ways you can use prompts.
Melissa Rice (16:48):
Casey Markee (16:50):
Check that out, well worth your time.
Melissa Rice (16:52):
I see we’re having issues copying and pasting all this stuff. Casey, do you mind providing some stuff for us for the recap next week so we can share?
Casey Markee (16:59):
Yeah, no problem.
Melissa Rice (17:00):
Casey Markee (17:01):
I’m not sure why it’s not pasting everything over, but good times.
Melissa Rice (17:07):
We’re working it out.
Arsen Rabinovich (17:08):
AI overlords are not happy with all this information getting out.
Melissa Rice (17:11):
We’re doing our best.
Andrew Wilder (17:13):
I think in the chat box there’s a little three little dots to say more if you click on that I think there’s a safe chat button behind that. I don’t know if everybody can see that.
Casey Markee (17:22):
No, there you go.
Melissa Rice (17:24):
Andrew, you’re the best.
Casey Markee (17:27):
I don’t care what they say, Andrew, you are the best.
Arsen Rabinovich (17:27):
Melissa Rice (17:34):
It’s the fine print. I couldn’t see it. Arsen, I know we already touched on this a little bit earlier, but it is a good place then to ask for keywords on a topic. Would you recommend it?
Arsen Rabinovich (17:44):
Yeah, absolutely. You want to take the keywords, but then obviously treat it again as a tool. I wouldn’t necessarily take it and run with it. Treat it as a tool, help it have it be your starting point, and then obviously take it into whatever your favorite tool is, Semrush, HRES, whatever, whichever one you use. Yeah, definitely a good starting point.
Melissa Rice (18:08):
Andrew, how can a blogger use AI to help them streamline or speed up their workflow in creating content? Is this a good alternative for people who are extremely stuck in writer’s block, for instance?
Andrew Wilder (18:23):
Yes, and I’ll illustrate this with a story from last weekend. We are working on updating the NerdPress website. We’ve got some frequently asked questions that were a few years old, so we’re updating all of them. We were working on one question and we had a draft in Google Docs. We started writing there and Heather and I were working on it over Zoom.
We were stuck on one paragraph and we’ve had a few people chime in where we were stuck. We were like, okay, this is kind of the right idea, but the wording is clunky and it just wasn’t coming out. I literally took the two sentences, pasted it into ChatGPT and said, “Rewrite this, but make it more casual.” It took a couple iterations, but it ended up coming up with something so better than we had originally.
We have language about we like to be a partner, not a vendor for our clients. It actually twisted that a little bit and said, “We’re a supportive partner, not a vendor.” I’m like, “What? That’s so much better.” Even if we got the rest of the sentence, it added this layer of something that we’d not had not imagined. Now, on the website, I’m really happy with the result it conveys exactly what we wanted.
It really was the perfect way to get over this hurdle, this a little bit of writer’s block. We had it, I pasted in some of the questions and I didn’t love the answers because they were a little generic. I did find it was better to work in small bits. Instead of having it write a whole long thing, the spot you’re stuck on, and then I didn’t use a lot that was verbatim, but it would have terms of phrase that were really good.
Then, if we flip it around, suddenly what we were working on would work now. It was really a lot of fun actually, and very liberating and it’s just a really powerful tool. It’s like Amy just said, it’s the world’s best thesaurus. It’s like that. Doing this actually reaffirmed for me, “Yeah, this isn’t taking over.” It’s not that good. It’ll get better, but it’s not that good.
Melissa Rice (20:20):
I feel like a lot of people are working as a one-man army when they’re running their blogs too. I feel like if this is something they could offset almost as a virtual assistant, that would be really helpful for a lot of them. I know that when we’re on calls all the time with clients, Arsen and I, there’s a lot of, “Well, where can I go to for this?” I feel like this is going to be a great option as it grows.
Andrew Wilder (20:42):
While we were talking, I did ask the question, the specific question, and it came back with a long answer. I said, “Make it shorter.” Here’s the shorter version ChatGPT can help log or streamline and speed up the workflows by one, generating content ideas, outlines in articles. Two, performing research and answering questions. Three, proofreading and edit editing articles for grammar, spelling structure.
Four, optimizing content for search engines. This can save time and improve the quality of content, allowing bloggers to focus on the creative aspect of their work.
Casey Markee (21:11):
That is great, but let’s talk about the real world applications of this incredible tool. Last night, when I forgot to empty the dishwasher and I’m like, “Crap, I need a really good excuse.” I just go to ChatGPT and it was able to generate me seven really good excuses. I used the one about temporary loss of power and totally saved my ass. I mean that was really good.
Temporary loss of power is the way to go. ChatGPT saving marriages every day. I’m telling you right now, folks,
Andrew Wilder (21:43):
The more you use it, the more you think of applications for it.
Casey Markee (21:45):
Andrew Wilder (21:47):
Wait, hang on, why didn’t I go there sooner?
Brett Tabke (21:51):
I had to write a book for my kid.
Casey Markee (21:51):
Melissa Rice (21:54):
That’s amazing, Brett.
Casey Markee (21:55):
Quiet, keep it between us, Amy. Goodness, tattletale.
Melissa Rice (22:00):
A lot of people are going to be prompting it tomorrow to, “I don’t want to go to work tomorrow.”
Casey Markee (22:05):
You give me some excuses? Better yet, can you generate me a valid doctor’s note?
Brett Tabke (22:10):
Well, legitimately, my daughter’s car died last Friday night at nine o’clock at school. I raced downtown 20 minutes to get there and a nice kid stayed with her before I got there, sat with her in the car. It was middle of downtown, so it was great. A nice kid from school stayed with her until I got there. I needed to write this email to the parents and say, “Thank you so much for the kids staying there.”
I don’t know, it was just like writer’s block or something. I couldn’t write it. I asked ChatGPT and it came back with about four sentences that were just absolutely perfect. Tweaked it a little bit and sent it to him.
Arsen Rabinovich (22:44):
One of my friends had it help him with writing a toast for his buddy’s wedding, finding the perfect gift for his fiance, all kinds of things.
Brett Tabke (22:53):
We’re getting good at this really fast. Everybody is getting so good at this really fast.
Casey Markee (23:01):
Melissa Rice (23:04):
I wanted to say, Arsen, I know we spoke about this yesterday, but Zapier is integrated now with OpenAI.
Arsen Rabinovich (23:09):
Melissa Rice (23:11):
I saw somebody explaining how they’re using it and it’s just like, “Do you have an idea that comes to mind and you don’t have time to jot it down or work on it?” Create a zap for that and it’s amazing, I love that. Tool sets, there’s a lot. Brett, which tools would you recommend? I know there’s Jasper, etc.
Brett Tabke (23:35):
Zapier is awesome. What’s so awesome about is if you have an idea, you can use a WordPress plugin that couriers ChatGPT to write you a blog post and then couriers Zapier to take that blog post and drop it into QuillBot and have it tweaked and then have it take it out of a QuillBot, and throw it into Grammarly to have it grammar checked and put it back into your WordPress, and it’ll pass most AI content detectors right there.
Melissa Rice (24:02):
Brett Tabke (24:03):
It’s just amazing. The thing we got to remember about all these tools is most of them like Zapier are just skins over an OpenAI API call. Most of them are just calling it on the backside. Be careful with stuff. I believe Jasper is using OpenAI.
Casey Markee (24:24):
Now, they are. They decided to just switch over it and not tell anybody. We’ll use OpenAI because it’s better. We’ll just use ChatGPT we’re supposed to be offering something unique. I’m like, “No, thanks.”
Brett Tabke (24:37):
It also does is optimize prompts is all they’re doing is optimized prompts. If you study the reverse prompt engineering and in fact somebody on Twitter today, there’s this massive thread, they’ve already jailbroke Bing ChatGPT they’ve already got it out there what Bing is saying to keep it in the so-called guardrails to keep it legit. There’s that. I use ChatGPT for everything like everybody was saying, titles, all your metadata, all your keywords, all your content.
Casey Markee (25:12):
Robots files, all that.
Brett Tabke (25:14):
I just use it for everything. I would clean it up a little bit before, if you’re going to do this in bulk before you throw it into Google, give it a little bit of a scrub, put your own spin on it.
Arsen Rabinovich (25:27):
It’s a little spiny spin.
Casey Markee (25:30):
Unless there’s excuses for the dishwasher, and then boom verbatim. No need, don’t mess with what works.
Brett Tabke (25:37):
Well, your wife isn’t Google.
Casey Markee (25:41):
It’s all good.
Melissa Rice (25:43):
Speaking of what it can and can’t do, I think we had this talk a little bit earlier, but is it good at writing recipes, Arsen?
Arsen Rabinovich (25:52):
Yes, it is good at writing at recipes. From the bloggers that I’ve talked to who’ve used it to write recipes and they did a really good job of writing the recipes, but when they actually cooked the recipes they tasted like crap. Until AI has taste buds, you guys should be good as content creators.
Casey Markee (26:15):
Arsen Rabinovich (26:18):
It is really good. If you give it the proper instructions, it will do the job. I think Andrew’s been playing around with that.
Casey Markee (26:25):
Well, we did this, Amy’s on the call, we did this live with Amy where she had a really simple Amy, you’ll have to let me, can’t remember what it was. Was it overnight Oats or something? When we just kept refining it down, the first one came through live on the call and said, “This is pretty good.” Then, we refined it down. Now, rewrite it I was a 10-year-old and was like, “Oh my gosh, it’s even simpler.”
It can save you a lot of time, especially if we need to dumb anything down. Maybe there’s some extraneous steps in your recipe that you don’t necessarily need. That’d be a good testing bed to take a look at that.
Melissa Rice (26:57):
Well, Casey, can it actually generate images in graphics for blog posts?
Casey Markee (27:02):
Well, Chat can’t right now, but there’s plenty of other options out there that can. When we’re talking about, I think I said, “Wait, here’s an article. I’m going to go ahead and paste over an article where it has multiple best AI image generators.” There’s two really good ones that I would want to make a note of for everyone on the call. The first one is DALL-E, which I know that Andrew has been using for quite a while with extreme hilarity.
Including with his featured images that he’s been using lately. The second, and the way that DALL-E works is it’s put out by OpenAI and it’s an extremely advanced image generator. It can generate a variety of digital art and illustrations from text. Whatever you can describe it can put together. Again, sometimes with extremely hilarious results. Another great option is DreamStudio.
It’s also known as Stable Diffusion and it’s an open source model that converts text prompts, empty images, and it does it in seconds. It’s super fast. It can produce these photorealistic artworks by combining an uploaded photo and a written description. That’s really been an emerging trend all over the internet because people are trying to sell these unique works of art and people are buying them.
It’s like virtual NFTs so to speak in a way. That people are trying to use this, “I always wanted to be an artist, so I’m just going to use Stable Diffusion to write all these masterpieces and see what we see.” Following your dream. The world is your oyster.
Melissa Rice (28:32):
We’ll touch on it later, but I know that people have concerns about copyright and things of that nature.
Casey Markee (28:44):
Well, what is that copyright?
Melissa Rice (28:44):
Before we get there, hold on.
Casey Markee (28:44):
I’ve never heard that.
Melissa Rice (28:45):
Andrew, what are some of the best prompts to use for SEO in particular?
Andrew Wilder (28:49):
I think we’ve already talked on a bunch of them. SEO’s so broad, don’t, not quite sure how to answer that to be honest. It’s like, “Well, what problem are you trying to solve?” Ask that question and then you have to start refining. We’ve talked about keyword ideas. I think you could do it for keyword research, start there and then start refining. I’m more interested in using it for idea generation.
Casey Markee (29:19):
Andrew Wilder (29:20):
Thank you. If you want to write a new recipe but you’re not sure exactly what you want to do, you can start experimenting with it to see what strikes something with you and then you can run with it from there. Title tags, heading tags, you can just ask it to write that stuff. The nice thing about ChatGPT is because it’s that conversation, you don’t have to keep giving an individual prompt. It iterates throughout the conversation.
Writing prompts is definitely a skill. I think we’re probably going to start seeing people who are specialists in prompting AI.
Arsen Rabinovich (29:53):
There’s already job listings for prompt engineers or whatever they’re calling it.
Andrew Wilder (30:02):
It’s hard to get the thing to do exactly what you want. If you have an end of something in mind, you have to course correct and finesse it there. I don’t think you can say, “What would Arsen tell me right about my blog post.” It’s not that kind of prompt I guess.
Melissa Rice (30:22):
Where you play with it, I think we’ll figure it out. We reset the rabbit hole, you got to go down it. It’ll be fun. I had somebody ask if the DALL-E was free like ChatGPT.
Andrew Wilder (30:34):
It gives you 15 credits a month and you can pay for it. I think it was at 13 cents per generation, which if you think about a 13 cents that’s a lot computing power. That’s how much it’s taking to do this, but you can definitely tinker it. I think your first month you got 50 credits so you can experiment with it a little bit more and then you can buy 100 credits at a time or something.
Brett Tabke (30:59):
Do they have a community feed over there where you can see what other people have generated on DALL-E like Midjourney does?
Andrew Wilder (31:07):
I haven’t seen one. You can share individual ones, but I haven’t seen it full time.
Melissa Rice (31:14):
Arsen, is it going to get crazy competitive now because everybody’s using AI?
Arsen Rabinovich (31:20):
I don’t think so. I’d love to hear what Brett thinks about this because he’s definitely has more better vision on this than I do. Look, I think it’s going to make it easier. It’s going to make the job, the work easier, especially for bloggers, for content creators. Do I think that there will be other ways for Google outside of the water marketing? Do I think Google’s going to implement other technologies to really gauge how the usefulness of the content and how it’s being generated and all of that? Absolutely.
I think the expertise, experience, authority, and trust signaling is going to become even more important. I have a feeling we might see a resurrection of authorship as a strong signal. It was a big thing not even more than a few things back and just went away. Definitely, I think Google and other search engines will become much better at identifying AI generated content, and it will slowly level the playing field.
Look, the crappy content’s been around. We used to spin content, we used to rewrite content, we used to do phrase rewrites. We used to do spin index code where you would do, did take all say different words, say meaning we used to do all this stuff back in like 2008, 2009, 2010. Google cut onto that really quickly. I think right now I wouldn’t be too worried about it. Yes, it might seem overwhelming.
Yes, you will see an increase in content being produced using AI. I think it’s going to get cleaned up fairly quickly, but I’d love to hear what Brett has to say about this.
Brett Tabke (33:10):
You’re not going to be replaced by an AI, but I think you’re going to be replaced by somebody using ai. They’re going to just take it to the next level. I think we’re definitely going to get more competitive and we’re already seeing it. I think because the guys who are using AI, their titles are better, their tags are better, their URLs are better, their content’s actually better because they’re eliminating all their blind spots.
Some people say we don’t have voice in ChatGPT, it doesn’t sound like us. I think you can get there. The thing is, this is the worst it’s ever going to be. Here we go from here on out, it’s going to get better, it’s going to get constantly better. I think it’s going to get more competitive, but it’s also going to get more creative.
Arsen Rabinovich (34:00):
I think people are just freaked out by this is because it’s a new thing they have to learn and that’s why it feels weird and that’s why it’s scary and it’s strange. It’s a new set of skills that you have to acquire in order to be on the same playing field with everybody else. Once you actually dive in and create processes and get comfortable with it, you’ll see that you’re working smarter. You’re definitely helping yourself out with this.
Melissa Rice (34:27):
Brett, just to keep it going, if a topic is going to be asked by more than one person over and over again, is there likelihood of this repetition happening, a plagiarism, should we be concerned about copyright and intellectual property, and things like that?
Brett Tabke (34:43):
Getting the same thing out of ChatGPT is actually challenging. To get two people to get the same output is a bit of a challenge. You got to wipe everything you got to say new and you got to ask it in the exact same way at the, there’s something to do with time of day as well. It’s hard to get the exact same content out of ChatGPT.
We tried the other day to get to 10 same lists of what are the top 10 ChatGPT prompts and there were four of us on the call and we couldn’t get the same thing four times in a row. We got a lot of the same things, but none of us got the same list four different times. I don’t think plagiarism is a huge concern, especially if you’re going to spend it a little bit before you put it out there, put your own mark on it.
There’s personalization, we’ve talked about Google having being personalized based on your search history and what you’ve clicked on history. ChatGPT takes that, amps it up to a thousand because they know what you’ve asked it, where you are. It really opens up personalization. I’m not worried too much about plagiarism. What I am worried about though is the copyright aspect, especially the images.
Getty has been hyper aggressive at suing anybody and everyone one related to copyright. In the end, I think they’ll get turned away in the courts because it’s going to come back to the same thing of derivative works that we went through with the thumbnail suits in 2000 through 2005. Google was sued mercilessly and they never lost. The only time they ever lost was against Getty here when they switched Google images.
What was that four years ago I think when they had to switch Google images. I’m not that worried about copyright because you get derivative works on the images and I’m definitely not worried about it on text it’s just not a thing on text. You’re not going [inaudible 00:36:38]
Casey Markee (36:37):
Again I know that’s small solace to those on the collar we’re thinking, “Well, geez, what are we going to do about recipes? Do we stop people from just going straight to ChatGPT typing in a recipe for spaghetti carbonara and just taking that and doing whatever they want.
Brett Tabke (36:52):
Casey Markee (36:53):
People will, there’s nothing you can do about it there. Number one, we can’t copyright recipes. Number two, even if we put some ad, we can redo those. I can put in the recipe, I can change some of the prompts considerably and it’s a completely different recipe. If you think that even in the instructions, I’ve been able to write a complete recipe post using ChatGPT and it’s only requiring me to do a couple prompts.
I think that that is absolutely something that will help some content creators but will absolutely cause a lot of sleepless nights for others. I think the big question here, Brett, and I don’t think we have it as a question here, so let’s go ahead and get into it, is that what is going to happen if we’re moving forward with Google and Google’s moving forward Bard and they’re just going to overlay Bard right on top of the search results, and they’re not going to provide any references, won’t that lower traffic to content partners considerably?
Where’s that traffic? The traffic has to go somewhere. We going to take the traffic there? It’s going to have to come taken from somewhere else.
Andrew Wilder (37:59):
Casey, you need to back up. We haven’t talked about what Bard is yet.
Casey Markee (38:01):
No, we haven’t? Let’s skip to that right now. Let’s skip to it right now.
Melissa Rice (38:07):
How about with ChatGPT for a quick explanation.
Casey Markee (38:13):
Just type in Bard to ChatGPT. You guys are good. It’s all good.
Melissa Rice (38:17):
We don’t have a particular question addressing it.
Casey Markee (38:18):
I didn’t think we did. I don’t think we mentioned that at Bard. Basically, Google introduced their approach to ChatGPT is Bard. Unfortunately, they introduced it yesterday and their stock dropped 9% and they lost $100 million. Was it $100 million or $100 billion? It was $100 million because the generated answer was incorrect. It wasn’t a big deal, but it was also not a big deal considering that’s what they used in their ad.
We only have limited information and even in the test data that I’ve been able to see, it looks like they’re going to layer that on top of the search results. That’s going to be at the top. It’s going to be very visible unlike with ChatGPT and being where they put it off to the side, correct, Brett, so it’s not obtrusive.
Melissa Rice (39:05):
Casey Markee (39:06):
Now, Google’s going all in they’re going to put that on the top and if that’s a layer on top of the search results, that’s going to siphon off traffic considerably. That traffic has to come from somewhere and it’s going to come from these content sites. You’re asking on the call if you’re a content creator, yeah you have cause to be concerned because this means this might be one of the first years where you have less Google referrals in general.
Brett Tabke (39:32):
It’s time to focus more on content marketing than SEO almost. You implement your SEO strategies, but content marketing, you got to build an audience to keep it because that audience is not going to come from Google indefinitely.
Casey Markee (39:46):
Let’s be clear, they have not said they were going to give references yet. Annmarie has said, I thought Bard was giving references. No, they did not give any references in their initial demo. Even though I and others have been surveyed for feedback on it, that’s one of the things we’ve submitted feedback on.
Yes, please provide references just like with a featured snippet, but we have no idea if Bard’s going to do that. You’re at their mercy now. We hope they do. We hope it functions similar to a featured recipe. Here is the information on this and by the way, I took it from this result, so you can go down and click on it.
Brett Tabke (40:23):
More than likely it’s going to be like Google local where you got to hunt through an entire screen to try and find that one little icon for the website and they keep moving that little soccer around.
Andrew Wilder (40:38):
I think it’s in Google’s best interest financially to get people to click to websites because that’s where the ads are.
Casey Markee (40:43):
Well, and that’s the thing, that’s the rub because they make their money off of display content ads. If they’re doing something to reduce that traffic to those sites and they also are going to take a hit with reduced display advertising. I don’t know how they’re going to do it.
Arsen Rabinovich (41:00):
Amy’s absolutely right, don’t neglect your email list. We’ve been saying this for many episodes now. Futureproof your audience, build that email list, focus on those subscribers because that’s algorithm proof. That’s AI proof. You can always email your audiences and build traffic that way.
Andrew Wilder (41:22):
The tone from everybody right now is, “Oh my gosh, all the content’s going to be the same because it’s all creative from AI.” I don’t think that’s what humans want.
Casey Markee (41:31):
It’s not true as Brett has already said, and we’ve tested it ourselves. We’ve put the same prompt in and very rare do we get something that’s coming back that is exactly the same. It’s just just not happening. We would want to rewrite that content, but it is getting better guys. Just the prompts we use today is better than the information we got back seven days ago. It’s unbelievable.
Andrew Wilder (41:55):
I just want to say like Joanne was asking in the chat, do we even bother to humanize our posts anymore? My answer is absolutely yes.
Casey Markee (42:03):
Andrew Wilder (42:05):
As I always say, write for your readers not for Google. You don’t want to have the whole long random story like people used to do 20 years ago on their blogs that are off-topic. Yes, they want to redo and that’s I think the way to combat this is to not have everything be the same standard generic stuff.
Melissa Rice (42:27):
ChatGPT is not quite the expert. We still have people who are experts in so many things. I think Google even said that they’re still interested in that E-E-A-T, that original quality.
Casey Markee (42:38):
E-E-A-T, yes very much.
Arsen Rabinovich (42:45):
I definitely think it also will depend on the query type and what the intent behind and what the user is looking for. I think there’s still a lot of unknown about how Google’s going to handle things. I think before we went live, we were talking about, and I might be wrong, and Brett, Casey, correct me if I’m wrong, when featured snippets came out, they didn’t attribute those to people at first also.
When they started pushing them out and then everybody made enough noise and now they get read more at the bottom there. It might come out and not be what we want it to be, but like Casey said, like Andrew said, they still need to make money at the end of the day.
Casey Markee (43:25):
That’s right. We’re not going to cut their nose off in spite of their face. They’re going to do something to make sure that this display advertising is still worthwhile for all of you on the call [inaudible 00:43:34]
Brett Tabke (43:36):
I think the fear is it’s going to end up mobile where they have one big chunk of an answer right there on top and then the rest of the first page is going to be nothing but ads. You want to get to organic, do you have to go to page two? They haven’t said yet.
Casey Markee (43:47):
There’s no page two any more, Brett. I don’t know if you heard that. We have this thing called infinite scrolling. Everyone on this call has page one results.
Arsen Rabinovich (43:55):
It’s still hard to not seeing page one. I’m constantly catching myself on our calls.
Casey Markee (44:00):
All this is all page one results here, folks. People don’t get the humor in that, but it’s funny.
Melissa Rice (44:09):
Casey, for people who are not into food bugging and different niches, is the prompts going to be the same when going to ChatGPT? Should they be maybe asking different questions? Maybe they’re travel bloggers, maybe they’re medical, anything. We lost you for a second.
Arsen Rabinovich (44:28):
You’re muted, Casey.
Casey Markee (44:32):
What happened was that was the long arm of AI reaching out to prevent me from pacing over all of the really awesome prompts that I had ready to go here for you. We’re going to try that again and see if we can get them over, but we’re having some problems today.
Melissa Rice (44:47):
We’ll save all [inaudible 00:44:49] for the recap for everybody who can’t catch it in the chat.
Casey Markee (44:51):
Now, let’s go with everyone here and see if this will work.
Melissa Rice (44:59):
I’m in suspense, we’re all waiting.
Andrew Wilder (45:05):
Well, now Casey’s frozen.
Melissa Rice (45:07):
Oh no, he is frozen. It really is coming after him.
Arsen Rabinovich (45:17):
His wife got home and discovered it wasn’t in there. I don’t want to be associated with him.
Melissa Rice (45:22):
It’s ironic because the power actually went out on him and wasn’t that his excuse about the dishwasher? [inaudible 00:45:29] There he is. We’ve got two of them.
Arsen Rabinovich (45:33):
Whoa, two Caseys.
Casey Markee (45:34):
You guys, you won’t believe this, but I actually had a robot come on my screen and tell me to cut it out so I’m not allowed to paste anything over into the chat window anymore. He showed me some pictures from my college day, so I can’t. Clearly, that’s not good. I’m just telling you right now that it clearly has something against you.
Arsen Rabinovich (45:54):
You got a tie by Russian AI. [inaudible 00:45:57]
Melissa Rice (45:57):
Do you have any you can just tell us about and then we’ll share this in the recap.
Casey Markee (46:00):
Absolutely. The really cool thing about this is that if you go over to ChatGPT, all of you can literally just type in, “Can you please generate some prompts for the recipe niche?” If I’m a recipe blogger, it will generate a huge amount of a wealth of recipe ideation and topics. Things like how to generate a recipe for healthy vegetarian dish that can made under 30 minutes.
How can I create a recipe for classic dessert that is gluten-free and dairy free? I have these three ingredients, can you please generate a list of dishes that I can make within 30 minutes? Can you create me a recipe for a hearty soup that is perfect for cold winter days? It’s endless. It’s not even remotely close. It’s crazy. Like I said, I’m going to try to pace them in, but I expect that the robot overlord will pop up again here any minute.
We’ll see what happens here, looks like that worked. We got them in. In that case, there’s a couple that you can take a look at, but it’s pretty crazy there is a lot you can do. This is a way for you to supercharge your ideation. This is a way for you to increase seasonality. If Valentine’s Day is almost here, it’s next week. Really cool, go in type in. Could you generate me a list of vegan recipes for couples for Valentine’s Day?
It is going to blow your mind and then look at that for content gaps. Well, wow, this is cool there’s 10 here. I don’t have these six on my website. Maybe I can do a roundup and I can seek out some other content gaps and share this pretty quickly. I’m going to have nachos at my Super Bowl party. Can you generate me six nacho recipes that I can go ahead and take a look at and choose which one you want.
You can never have too many nachos. I’m going to have at least three types of nachos in my Super Bowl party. I mean, come on.
Melissa Rice (47:49):
We’ve got somebody who does. Casey always use please and thank you when prompting ChatGPT, and I think you should always be polite to AI.
Casey Markee (47:56):
Brett Tabke (47:58):
Just in case, you never know robot overlords might be touchy about.
Casey Markee (48:05):
No, guys, I do it all the time even because I even do it to my Alexa. I even say please and thank you. If they do take over the world, I want them to remember that I was nice to them.
Melissa Rice (48:15):
Alexa will actually prompt back with something saying about a giving tree and adding leaves to it. I’m like, “That scares me a little.” She’s literally counting the thank yous. As far as I’m concerned, being polite, it doesn’t hurt.
Casey Markee (48:29):
It doesn’t hurt, just an extra step there. Let me tell you, I really did appreciate those excuses to get out of me for getting to load the dishwasher. I’m just telling you always be polite. I’m all good.
Melissa Rice (48:43):
Andrew, do you have any favorite ChatGPT-based Google Chrome extensions that you’ve been using?
Andrew Wilder (48:49):
I actually haven’t been using any, to be totally honest. Casey was kind enough to provide me with a link to an article that has six good extensions. There’s so many different things you can do with this. I’m not sure what that question was asking for. Right, because you could use ChatGPT for helping you tweak better. You write the prompt where you are.
I think the idea of the extension though is that you don’t have to go over to the ChatGPT interface and use it and then come back, you can use it where you are. You can take a look at some of the extensions in there and depending on which browser you use, you’ll have different options.
Melissa Rice (49:28):
Awesome, thank you. Brett, are there any other uses for AI in SEO? For example, we’ve seen people create full webpages and I’ve even seen people create apps.
Brett Tabke (49:43):
ChatGPT as a programmer is the scary one. When they did ChatGPT-2, it didn’t take off last year like they wanted it to, it just didn’t take off hardly at all. They went down this niche of trying to do individual transformers and one of the first ones they did was programming. They fed it all PHP and then they fed it all Perl and then they fed it all this. Every major programming language and it’s very, very good at programming.
There’s a long blog post out there about a guy who made a Twitter clone using ChatGPT in about eight hours and it’s start to finish a Twitter clone. It’s just absolutely mind-blowing how good it is. I use it all the time, I actually do, I’m a programmer by trade. For WordPress plugins, it’s outstanding at WordPress plugins. The less complicated the better of course, keep it simple, keep it stupid. You got that as a programmer, then you got things like let me run down my list here.
Translating language, it’s great at translating through a whole bunch of stuff. Correcting grammar, simplifying or expanding text, summarizing text. It’s awesome at summarizing stuff. I use it all the time for that. Brainstorming an idea generation, in fact, summarizing text, there’s a word, a Chrome plugin that, an extension that allows you to go to any YouTube video, and then it pulls in the text from the caption, the closed captioning.
Then, summarizes the video for you so you don’t have to go watch the entire video. Key keyword analysis, detecting correlations, brainstorming, we’ve talked about ideation nonstop here. Analyze the content of a webpage for keywords and classification metadata, social media, obviously ads and posts. Have it right, see what it comes up with for a Facebook ad. [inaudible 00:51:39] all the time.
Casey Markee (51:41):
It’ll definitely do that. It’ll write all your ads for you.
Brett Tabke (51:42):
It’s great. Somebody posted it on a meme on Twitter about, “Google’s AI already writes your ads.” I go, I responded. Yeah, Google’s AI writes your ads so that they can make more money. I want ChatGPT to write an ad so that I make more money and there is a definite difference there.
I got it baked into everything I do now. If I’m having problems coming up with content at all, I immediately go to ChatGPT and ask it a question and it kicks off a good creative storm for me. I’m using it all over the place.
Melissa Rice (52:18):
I think I put a lot of my fear stress too, when I first started hearing about how much people are integrating it into their work. I was like, “Will there be a need for any of us after this, but everything I read makes me feel a lot better.” I know we already said this earlier, but it does need a human touch at the end of the day. I love that.
Casey, some people are inaccurately talking about it being the death of SEO, but do you agree at all or what do you think?
Casey Markee (52:50):
Crystal ball says no. The thing is that it’s a great language model. It’s great at creating content. It doesn’t take the place of search engine optimization. It doesn’t take the place of the many decades of experience that are on this call. It doesn’t take the place of our learned processes and what we know will work and what won’t. SEO is complex and continually changing.
We have Google alone goes through dozens if not hundreds of changes a day, thousands of algorithmic adjustments a year. That makes what we do compelling and continuously relevant because we have to adjust to change that. Any SEO worth their weight knows that SEO involves everything from keyword research to content creation, to schema, to link building, to UX justification, to auditing, to whatever.
We have to wear many hats and there isn’t one tool that’s going to replace all those hats. I think that especially Arsen and I think we provide extremely competent and very professional audits, those are going to be impossible to replace by one piece of software. Not worried.
Melissa Rice (53:57):
Awesome, we’re short on time here, so I just want to get to a couple questions before we go, and it ties into what I had for the panel to wrap up with, it’s the future. We’ve got a question from Chula and does anyone have any idea when ChatGPT four will be released? Is it moving fast? Should we be worried about how quickly it’s going?
Casey Markee (54:20):
Well, they said it’s this year.
Brett Tabke (54:24):
That’s it. The last thing Sam Altman said was, “Don’t look for it anytime soon.” He said, “You’re going to be disappointed.” They’re going to set on it. They’re working with Bard right now. That’s all they’re worried about their focus is right now and satisfying their commitments to Microsoft.
Melissa Rice (54:40):
No Terminator sale Cyborg’s coming to get us?
Brett Tabke (54:42):
No, no. It’s going to be a while. I don’t think we’ll see ChatGPT-4 until probably the fall.
Melissa Rice (54:50):
Very cool. Another question from Janelle. If Bard is taking multiple sources to spit out the best one answer, how and will, or can it give reference?
Casey Markee (55:02):
We have no idea and neither does. I don’t think there’s anyone that can tell us that. We have no idea. I assume if it’s taking the majority of the answer from one source, that’s why we’ll see the citation and we’ll see what we see. Maybe we won’t, I don’t know.
Melissa Rice (55:22):
I know we talked about this earlier too about Google seeming to not really care, but do you think that there will be a digital watermark for content created using ChatGPT eventually?
Arsen Rabinovich (55:35):
Any kind of AI.
Melissa Rice (55:36):
I think yeah, exactly.
Casey Markee (55:39):
Wasn’t ChatGPT saying that they were going to work on that and that that’s going to be available soon?
Brett Tabke (55:43):
Yeah, but it’s going to be easily circumvented. Honestly, the only way they could come up with a watermark is if they kept every bit of output and give it a hash code and then compared the hash code to something somebody submitted. It’s going to be just as easily beatable as what we have now. Arsen, quiet, quiet, I think that’s for text image imagery is a little different.
Yes, that’s true, very true. If you haven’t tried Midjourney yet, go try Midjourney. It’s like DALL-E on steroids and then find the community feed on Midjourney and you’ll get this constant feed of high quality images that is just stunning. I can sit there and watch it like a movie. I caught myself, I was about 30 minutes into it the other day.
Melissa Rice (56:35):
Same, it’s beautiful. It’s also really inspiring. Thank you everybody really, this was such a hot topic, and I feel like we covered a lot of bases here today. Brett, thank you again for joining us.
Brett Tabke (56:47):
Thank you for having me. I appreciate it, this was fun.
Melissa Rice (56:49):
It’s been lovely. We are going to Tastemakers next month. TopHatRank and NerdPress will be sponsoring the conference in Chicago, so be sure to come up to the third floor, and say hi to us if you’re going to be attending. We’ll be excited to see some of our clients and maybe some potential new clients in the future in person.
Brett Tabke (57:09):
Where’s that at in Chicago? Where’s that at?
Melissa Rice (57:11):
It’s in Franklin Park.
Arsen Rabinovich (57:16):
Brett’s conference, which is one of our favorite Pubcon is end of this month. Casey and I will be in Austin speaking at this conference. Those of you who are in the area definitely check it out, pubcon.com.
Casey Markee (57:31):
Arsen and I will be sober on Tuesday. We will not be sober on Monday. I want to make sure that you understand that if you’re coming to our session on Tuesday, I will be alert not so much on Sunday night and Monday. I just want to make sure we preface that right now.
Melissa Rice (57:49):
Wear dark clothes, Arsen because if you’re going to be eating all that barbecue, I know how you can get.
Arsen Rabinovich (57:59):
If you’re in Austin, definitely check it out or in surrounding area. Come and meet us, come and check it out. There’s going to be a lot of SEO information, a lot of SEO brains at this show. Always a lot of fun.
Brett Tabke (58:10):
Google and Microsoft will both be there and they’re both talking AI.
Casey Markee (58:14):
Fantastic, it’s going to be awesome.
Melissa Rice (58:17):
Well, thank you everybody, thank you for joining us today. Thank you so much again as always, and we will see you next time.
Arsen Rabinovich (58:24):
Casey Markee (58:24):
Take care, everybody.
Brett Tabke (58:25):
Melissa Rice (58:25):
Brett Tabke (58:25):