My interview Joost De Valk about SEO, Ai and WordPress.
Joost Interview summary
- Joost De Valk, founder of the Yoast SEO plugin, discusses the impact of AI on search engine optimization (SEO).
- Joost is surprised by Google’s response to AI, both amusing and concerning, as it poses challenges in distinguishing between real and fake information.
- Google’s experimentation with AI-generated search interface raises questions about its reliability and the impact on content creators and the SEO business community.
- Joost believes that AI-generated text is not built to generate truth, and fact-checking becomes crucial.
- The future of AI in content creation and productivity is promising, but governments need to implement regulations and increase public knowledge about the models used.
Interview transcript
[00:00:00] Good morning. A few days ago I was at Word Camp Athens and I got the chance to interview Yos. Yost is the founder of the Yost SEO plugin, which is the world’s most popular SEO plugin with now over 11 million downloads. He knows more than anyone else about seo, WordPress, and now ai. And I got the chance to chat to him about what he thinks is gonna happen with AI and search engine optimization.So if search engine optimization is important to you and content is important to you, and AI interests you, This interview has some real gold dust in it. Just one more quick thing to add. I do seem to say, yeah, quite a lot in the interview, which irritated me a little bit, but hopefully it won’t irritate you too much and it’s worth persevering because there’s some brilliant content in here.
Let’s get to it. So I guess the big topic at the moment everyone’s talking about is ai. It, there is a lot of chat about, there’s a lot of chat. Yeah. Um, and you obviously a SEO jury, so I guess broadly where. What’s your, have you been taken by surprise of how quickly this has hit us, I guess? Well, [00:01:00] I’ve been mostly surprised by Google’s response to it.
I, it, it, so chat BT came and suddenly everyone was in panic. Yeah. And it was hilarious to see that. Um, and also a bit frightening in, in that they really seem to have been caught a bit, a bit by surprise. Yeah. Um, It’s interesting. I think it, it, it has many, many implications of things that we can do with it that are awesome.
It also has many implications of things that we’re gonna do with it that are gonna be fairly bad for everyone involved. Yeah. So, yeah. So, uh, fake news was a problem already. It’s not gonna get any smaller with this. Yeah. Um, and I think. That, that’s gonna be hard to bustle for Google for, for a while, because suddenly it’s not just fake news, it’s fake everything.
Yeah. So recipes, uh, reviews, whatever you want, suddenly all becoming [00:02:00] well, a lot harder to, to realize like what’s real, what’s not. And I, I, I saw that Google is starting to experiment with generated, generated, changing the search interface. So the top bit’s gonna, um, potentially. Start to use chat two vp like answers yes.
How, how do, I guess a few questions. One is, do you view that as a good thing? Two, how do you, how do you view that’s gonna affect content creators? And actually three, how do you think that will affect the SEO business community? So I think these, the answers are actually all connected because I view, view that not necessarily as a good thing.
Okay. Um, mostly because, These models are not built to generate truth. Yeah. They’re built to generate text. Yeah. Which is something entirely different. And I think that we have, we have been so trained to think that what we find on [00:03:00] Google is true. Yeah. Uh, and that is a very scary thought. Um, and those, those large language models just generate the next word.
I mean, they, it’s, it’s like, what’s the most logical word is next? It’s not like, What is the right answer to this question? Yeah. Um, I hope that they’ll iterate and make it actually good, but so far and what we’re seeing from the early people that have access, there’s a lot of nonsense in the results and, um, well, I hope we get rid of that before, um, it becomes really adopted mainstream, then the problem is that.
Once they start generating their text, um, the question is how did they send traffic back to the websites that they’ve based their answers on? And so far the answers that I’ve seen to that question are not very positive. So I think I’ve seen little featured snippets maybe in the book. Is that right? Yeah.
And, and some of them have some links, et cetera. But yeah, I think [00:04:00] overall traffic will go down. Yeah. And uh, so Google will become less of a referring source. Yeah. Which is not good for anyone. It’s certainly not good for sites to rely on that traffic to make money to be able to generate more good content.
The problem then becomes if, if more of those models don’t work, then there’s less new good content to train the AI on and it sort of becomes a self-defeating thing. Yeah. And do you think, do you think, uh, Google’s business model is gonna hold it back from. For example, the, the Google search search page has been kind of broken for a while.
Really. I mean, in terms of the, I see a lot of searches where actually the, almost like the whole page is almost like, Paid dad’s, right. Paid ads. Yeah. It feels like it’s kind of been broken for a while. Right. The, the thing is, is that broken though? I think it’s actually working quite well in terms of their business.
Okay. But from the, from the user’s sort of search intent experience, that’s, that feels great. So for a [00:05:00] normal user, if they get what they want from those paid ads, they’re perfectly happy. Right. Okay. So yeah. If Google’s succeed at actually giving results that are good and, and get paid for it. Yeah. And they’re actually doing a quite good job.
Yeah. Um, I don’t think they’re always doing that. Yeah. Uh, and I think that if they struggle with that too much, then they leave room for competition. Yeah. Which I sort of like. Yeah. Bing. Bing was the first to have chat G B T in her interface. That’s also why Google is now trying to push it so hard because suddenly they get beaten.
Yeah. And I love that because we’ve had Google dominance for far too long. It’s not good for a company to be that dominant that long. Um. I think that will bring improvements. I think there’s a lot that we could do with AI in all sorts of ways that can help all of us. I think we ha we have to realize also that we need to maybe train people on what they can [00:06:00] believe and can’t believe on the web.
Yeah. And, and that with that it also becomes more important. Who wrote a piece of content? Yeah. A, a generic post about seo. Is worth a whole lot less than a post written about SEO by John o elison a com. A colleague, uh, from Yoa in the past, the head of SEO there, he, I mean, if he writes something and you want to read that, you, you knowed you’re reading his opinion.
If you’re, if you’re reading a generic piece of content, whose opinion is that? And, and what is it based on? And is it actually true? So, The author becomes more important than that, in a way is very good for us in the WordPress ecosystem. Yeah. Because suddenly blogging becomes a thing again. Yeah, absolutely.
And so do you have any like discreet advice for content creators who are starting to leverage AI in terms of the content they’re producing? Like would you say, yep, go for it? Or would you say, hold off and don’t use it [00:07:00] or, so I think it’s a good tool. Yeah. Gen if you then make it generate too much text.
Then you need to check every fact that’s in there. So that’s a bit hard. Yeah. Um, because the, the things that you won’t check because they seem so logical will be false. Yeah. I mean, the first time I had chat, g b t generate a bio for me, it’s said that I was an investor in rank last. And while I.
Appreciate that, that is a plugin company as well. I certainly did not invest in rank math. So it it is like it’s confidently wrong a lot. Yeah. And, and it, it is very confident in how it works those things. And that me, and that means, means that you have to check everything. Yeah. But at the same time, having it help you create summaries could be very worthwhile.
Yeah. So do you have a, if you were gonna predict like a year, two years out of what? This, all this all looks like for [00:08:00] us. What, do you have a vision of what that might, can you start to piece what that might look like? I think that’s hard actually. It’s moving so fast. Um, I think that what we’re seeing so far in the general world, the like generic text creation Yeah.
Is only a tiny bit of what’s to come. If you look at the, uh, the AI agents space a bit. Yeah. That is super exciting. So you can create an agent, you can ask it to do something like research a topic. Yeah. And it’ll come back to you with stuff. I think that using stuff like that will become a major driver of, of innovation, but also of, of making people more productive.
Yeah. It’s already like, there’s large companies out there that have. Build their own coding models to make their developers more, uh, work faster and work better. Yeah. [00:09:00] And, and you see people saying things like, Hey, 30 to 40% faster, more code being committed by the same amount of developers, I think that can actually re resolve some of our scarcity issues.
Yeah. So I think there’s a lot of good. Yeah, I mean, I was watching this, um, Ted talk about, um, the Khan Academy and he was talking about the fact that every kid could potentially have their own tutor, personalized tutor, which is phenomenal, right? Yeah, I, so I think there’s great things that we will be able to do with it.
It’ll, it’ll take some time. So just a couple of final questions. What, um, this is a negative question. Uh, there’s been a lot of catastrophizing around AI and governments are getting involved and trying to put guardrails. So do you, do you worry about some of that stuff or do you think that’s overblown? Um, I, I think the guard rails are needed.
Yeah. And I, my worry is that governments are not moving fast enough. Okay. Um, yeah. The guard rails are needed. We, we need to also, [00:10:00] well train people a bit more in what is this actually? Yeah. And, And which things are they doing and which things are they? Um, um, which, which things are they, are we telling people that that AI is good for and is not good for?
Yeah. And uh, so there, yeah, there’s, um, there’s a bit of a, a worry there. It’s interesting cuz it’s just hit my kids at school. They’re just, they’re cause they, um, Snapchat have introducing AI on Snapchat. Yeah. Yeah. So suddenly all the kids in all the schools in England now. No, there’s this thing that, well, I, that I dunno, I, I, I think that there’s already quite a few, uh, Macbeth, uh, type reviews and stuff being written by chat GBT for June.
So I think, uh, organizationally it’s interesting how school’s gonna be, uh, forced to change really quickly and probably won’t be able to change that quickly. No, same, same with government. So they’re gonna try and stop it, which I think is weird. Yeah. Because not, that’s not how this works. Yeah. [00:11:00] But use it right.
Is is also hard. Yeah. So, yeah, it’s gonna be an interesting thing. I, I’m not necessarily very like thinking that the world’s gonna blow, but, but it, I do think it needs to go outta reels. And do you think Final question. Um, the big worries for me has always been around the single. So, uh, like with trading, like information systems used to be distributed and then we start, started to centralize them with things like Google.
This, this feels like a, a further consolidation of like, we’re creating this super brain that we go to for our single source of truth. Um, but then I kind of start to investigate the fact that we might have different LLMs that we actually, I I think we’ll end up with lots and lots of different LLMs. Yeah.
And that’s actually better than, than having one bit Google. Yeah. Um, I think, we’ll, I mean, if, what’s, one of the things I think should be regulated is like, When you work with an L l M, you need to know what’s in it. Yeah. And that should be like [00:12:00] public knowledge. It should be public knowledge. Yeah. And if you know what’s in it, then you know what you can use it for.
Yeah. And I think that that is something that, um, there’s no visibility on that at all at the moment. No. No. Right now there’s not. And I, I desperately hope that that will get laws very quickly that say, okay, if you’re. Uh, if you’re a an open AI type company, you have to tell us what’s in your model. Yeah, yeah.
And, and then some publishers will, will start complaining like, Hey, but why are we not getting paid for it? So, a big thank you to Yos for his time and his insights. Some really valuable stuff in there. I hope you enjoyed that and I hope you found it useful. This actually forms a series of videos that I’ve been doing around AI and WordPress.
Last week I interviewed Matt Mullenweg, the co-founder of WordPress, and a few weeks before that, I interviewed Chris Lemo with the fantastic interview and insights around AI as well. Those links are in the description below. If you enjoyed this video, if you can hit the like button now, it would be amazing because it makes a huge difference in terms of how this video is shared [00:13:00] around YouTube.
And also every time you do hit that like button, our cats get a little treat.
If you want to keep up to date with all things AI and WordPress, hit that subscribe button down below now, and you’ll be notified every time I release a new video. Keep well, and I’ll see you soon. Bye for now.
Leave a Reply