Transcription
If you're thinking about using mass AI-generated text for search engine optimization, this episode will be for you. Even before the execution, this is one to listen to, to watch, to listen to. This is one to think about even before starting the execution with the mass AI-generated content.
Generating tons of pages of SEO content using AI, this person asked on the SEO subreddit: "What's your experience with mass-generated AI content?" The person said, "I'd like to know about the real-world experience with mass-generated SEO-optimized AI content." That's a mouthful. I'm really looking at experiments that have been running for more than 6 to 8 months, so I'm going to share what the people in this post said. The replies, then I'm going to share my own take on this question. This post actually had zero—it got a lot of responses, and its karma, its ratio of upvotes to downvotes, was negative because more people downvoted this question than upvoted it. People really hate this topic.
And so the top comment is: "Why would you do that to your readers?" A follow-up was: "Very good question." And the original poster said, "Do what? Do readers care if the content is generated by AI if it solves a problem for them or offers value?" That's a great question. Do readers care if the content is generated by AI if it solves a problem for them or offers value? Someone said, "It's not new information you're providing to your readers. All information used by AI is already on the internet." And so the original poster responds to that and says, "Of course, but AI is great at compiling that information and presenting it in unique and interesting ways." Then the person arguing with the poster says, "Treat your readers however you want to treat them. No matter how it's presented, it's not new; it's not new information." And an argument ensues.
Someone says, "To play devil's advocate here, how much of the information you tell other people is new information? It's much more likely that your knowledge has been influenced by someone outside of yourself. Isn't that the same concept but on a much smaller and more biased scale?" People are like, "Exactly. Copywriters were regurgitating the same topics time and time again before we had large language models." Another person said, "That's a fair point, but not all teachers teach the same way, for example, even though they may have attended the same classes. AI doesn't have that option; AI teaches the same way." One of my businesses is in-home tutoring students with ADHD because I have ADHD. I've written articles based on my personal experiences tutoring. So everyone may write about a way to study for a spelling test, but I would have a different spin on it or even a different method based on personal experience.
And people are really hating on the poster—the person who made the post in the first place, by the way, if you're wondering. If you're watching on YouTube and you're wondering why I have a band-aid on my finger, it's because I cut my finger like an idiot while cutting lemons. When life gives me lemons, I stab my finger. That's what I do. So I have a band-aid, and it'll be off tomorrow. Yeah, people are hating on the original poster. This person goes, "You ask for people's experiences with mass-produced AI content, and when somebody shares their negative experience, you are trying to convince them otherwise." What's the catch here? Somebody else: "There's nothing new under the sun."
All right, here's another direct response that was actually all part of a thread within—that was all part of a comment thread within this post. Somebody said—and this was a really good take—somebody said, "Mass AI-generated SEO content looks good until Google shadowbans the website and its traffic drops drastically." It happened with Farm Easy last year. They were fetching around 1 to 2 million traffic per month around June to July 2024. Then, by the start of August 2024, they were penalized, and by December they hardly could pull traffic of 8 to 9,000 per month. First, it's unethical for millions of content writers out there who are under constant pressure of delivering quality projects. Second, Google would never endorse AI content. And so the person who posted the overall post said, "Wow, didn't know about Farm Easy using AI-generated content." Google says they don't care about AI content as long as it's valuable. It's true; Google does say that. There were several experiments launched to test the traffic-pulling abilities of AI content. I can't find any of them on YouTube or on other websites. Wondered—wonder what happened to those experiments.
And the last two posts: someone said, "Depends on the content and the prompt; garbage in, garbage out." And then somebody said, "Google is against mass AI-generated content," and that actually is true, and I'll tell you why in a moment. But the original poster falsely said, "I think Google is not against mass AI-generated content; it's against the low-value content." I'm experimenting with AI content, and I'm getting content that's way better than human writers. However, I'm hesitant to put it on a live site. Google might just notice "perfect content," quote-unquote perfect, and flag it. I remember reading about some experiments being done by YouTubers and independent SEO consultants in this area. I wonder what's the progress on their experiment. I can't find them now. Someone responds and goes, "You need better human writers."
And then the last comment on this is a comment to the comment that I just read, and the person says, "Funny reading the different experiences here. I manage copywriters and have spotted the ones who are relying on AI because the quality of their work fell off a cliff, and now the time I take to edit their work doubled. I think something to keep an eye on if you are running the experiment is duplication from other websites. I have seen AI pieces that steal sentences word for word from content written by humans. Google will not flag your content for being perfect; it will flag it for duplication. You will have a different definition to what perfect is, by the sounds of it."
All right, so here's my take. Edward Stern, here's my take on mass AI-generated content. I've seen it work; I've seen it work for a few months, and then I've seen sites get super penalized. Their traffic drops off a cliff, or they're like, "This is too good." We've seen other sites get penalized, and then they stop doing it. But with that said, I've seen it work in the short term. I've never really seen it work in the long term, for a long amount of time. Bill Hartzer, a very experienced and sophisticated SEO, came on this podcast—that's episode 608, "SEO shortcuts from a 25-year veteran"—and Bill said, "Google will look at the before and after velocity of new content on your site. So if you suddenly start putting up way more content than normal, there's a good chance that's going to be seen as spam." So if you actually are going to do the mass AI-generated text, you need to work your way up to it. That's what he said. However, we did not talk about—I don't think we talked about—how long it will last for, but we both agreed it's possible that Google might put in AI detection in the future. They don't have that right now; they do not have that right now.
So what I have seen: I've never seen a site getting away with mass AI-generated text for a long amount of time. I have seen it work in the short amount of time, and if you go from zero to 100 really fast, there's a good chance that that will get flagged. So here are—here's my direct advice on how to handle mass AI-generated text for SEO: I would not do it on a site that you've been developing for a while. It's actually pretty easy to detect, and I'm going to give an example, because even if some AI detection tools are not able to detect it right now, they might in the future. But even so, for like—okay, here's a great example: ChatGPT-40. It's actually getting past a lot of AI detection tools, but I recently tried putting up a question on the search engine optimization subreddit, and that question was synthesized for me based off of my prompt. It was synthesized for me using GPT-40, which gets past a lot of these top AI detection tools for now. I put this question into the SEO subreddit, and the post was immediately rejected; immediately rejected; didn't work. And then I removed all the punctuation and just rewrote the punctuation myself—the periods, the apostrophes, the everything in it, commas—and then the same exact content got through, but the punctuation—some of them were different Unicode characters than what is normal—and these different Unicode characters are common with AI. And so that said to the SEO subreddit, this is AI content, and it was automatically flagged and didn't work. And so even if it looks like it's not AI to you, it could still easily be picked up as AI.
I personally wouldn't be surprised if a special identical alphabet is developed, and you know that any of the letters that are used—it was developed by AI—that actually would be a really good solution to detecting AI-written text. There's also already just patterns with words put into AI-generated content that makes it more easily identifiable as AI-generated, even with GPT-40-generated images. There are literally patterns encoded in the pixels—at the pixel level—that makes the images recognized as AI. You might not be able to tell; it might get past the human eye, but AI detection tools can tell with the patterns encoded in the pixels with GPT-40. I will say—I'll play—I will play devil's advocate on this and I'll say it's possible that special AI copywriting tools might get past all of this, but it could still become a cat-and-mouse game, and I just personally would not recommend trying this on a site that you have been developing for a long time. If you've been developing a website for a while, it could hurt—or even honestly—or even on a website that is connected to the rest of your websites using Google Search Console. One year ago, Google went after all of these websites that were using mass AI-generated text, penalized all of them. But the scary thing—and they did this—they did this for a reason; they did this to scare people—and you know what worked. The scary thing is then they penalized non-AI sites that were on the same Google Search Console as the sites that were penalized. Google Search Console—if you don't know—it's Google's own tool for accessing Google Search Index. You submit your pages to it; you get your SEO data from it. So all your sites are connected to Google Search Console, and so Google would see one site was using mass AI-generated spam text; it would penalize that, and then it would penalize all the websites connected to that Google Search Console. That's scary.
But if you do decide to do AI and you still want to use it on your website, my advice: don't go hard at the beginning; start slow. I would really—and you can't really do this when it's mass-generated, when you're putting up tons of pages—but even just in general, if you're using AI-generated text, just make sure that it doesn't read like AI. Make sure that it's not verbose; make sure that it's clear; it's concise; it's not using special characters that AI always uses, like the em dash, or words that AI always uses, like "delve." The em dash—em dash—look it up; AI always uses the em dash. I even one time told AI not to use it, and it's like, "It is better to use it; it is better for writing." In fact, I think maybe it was instructed to use the em dash to make it more recognizable as AI. I think that's possible. But yeah, I agree with the people who said, "Don't do that to your readers; don't give your readers AI-generated slop. Try to give them some good stuff." Make sure that it contains good information. AI often will get key facts mixed up when it's summarizing—summarizing and synthesizing—it will get key facts mixed up with one another, or it will combine them. So make sure the information is good and it's reliable; it's not hallucinating. I've been getting so many hallucinations recently.
And then if you want to look—if you want to try AI, I really would caution against using it a lot on your personal websites or your money sites—websites that you have been developing and putting time into for a long time. A better place to try would be parasite search engine optimization on Medium or LinkedIn Pulse. So, for example, if you Google "crypto growth hacking"—and I talked about this on the podcast recently—it is the number one result is entirely written by AI using an older model of ChatGPT, and it's on an aged account on Medium that used to write human-written articles. Then when ChatGPT came out, it switched to AI-written articles, and all of its articles are now AI-written, and it's an aged account coming out as well—on an aged publication—as in like these accounts and publications have existed for years, and they're using AI-generated text that is targeting keywords, and the text is not good, but it still ranks. I think it still ranks because it hasn't been caught. I think when it is caught, it will be taken down, but for now, it is not caught. It's a cat-and-mouse game, but even if it is caught, it's unlikely that it's going to hurt the person who was posting it because it was posted on Medium. Same goes for LinkedIn Pulse. This isn't connected to your Google Search Console, and so because it's not connected to the rest of your properties, you keep the rest of your property safe. So it's a safer way to do it with parasite SEO. Parasite SEO is using the authority that another platform has or the prioritization that Google is giving another platform to rank really fast and easily on Google. It is also black hat, and it is against the rules, so you should know that. But so is mass—mass—and so mass AI-generated content for SEO—it's not against the rules in its—in it—in and of itself, because if it—if there was a way for it all to be human-reviewed, that would be fine. But because there is not a way for it—or it's just extremely unlikely that it is all going to be human-reviewed—it's likely most of it is going to be slop, and Google doesn't want slop; it doesn't want to—it doesn't want to index slop, and it doesn't want to serve slop to people who are using Google. And so Google is very staunchly against this kind of AI-generated spam. Google does not want to give spam to its users. That's the principle behind why this is a bad idea—why to address the poster's question on Reddit—why mass AI-generated content is not a good idea.
And that is all that I got for you on this episode of the show. This is episode 600—what is this? Is it 650? No. Oh my god, it's not. That's tomorrow. This is episode 649 of my daily digital marketing show. Wow, 649 days in a row doing this podcast. I do it every day, no matter what—even when I cut my finger open while slicing a lemon this morning—no matter what—or at least I try. I'm sure there will be days that I have to miss in the future, but I hope not. Knock on wood. Thank you so much for watching; thank you so much for listening, and I will talk to you all again tomorrow. Bye now.