How Will AI Effect Being A WordPress Freelancer or Agency Owner
With Special Guest Max Ziebell
AI’s impact on WordPress freelancers and agency owners is profound. Discover how to adapt and thrive in a changing digital landscape.
Are you curious about how artificial intelligence will influence your role as a WordPress freelancer or agency owner? In this engaging show, we break down key trends and emerging technologies that are set to redefine digital services. Learn about both challenges and opportunities AI presents, ensuring you’re equipped for success in a rapidly evolving market.
This Week Show’s Sponsors
LifterLMS: LifterLMS
Convesio: Convesio
Omnisend: Omnisend
The Show’s Main Transcript
[00:00:01.410] – Jonathan Denwood
Welcome back to the WP tonic show in WordPress and SAS this week. In this episode, we will discuss AI with our guest, Max Ziebell. I’ve butchered his surname, and I’ve managed to insult another guest, which I do every week, it seems, but Max seems to laugh it off because he looks like a good soul. As I said, we will discuss all things AI in this episode. What it is. I’ve never known a subject in tech that gives different answers from different experts, supposedly experts in the field. I’m baffled. I’m sure you’re confused. Hopefully, Max can clear up the confusion a little bit. He seems up to it. So, Max. Yes. There you go. Max, would you like to give a quick 2030 2nd intro to the tribe?
[00:01:27.110] – Jonathan Denwood
Yeah.
[00:01:27.880] – Max Ziebell
Hi, I’m Max Ziebell. I’m from Germany. I’m based in Berlin, so I’m in the capital now. And yeah, I’ve been into WordPress since the beginning and I will get into that a little later and have been working as a web developer for a long time. Also, I have done animation and some adjacent topics on my life. So, I’m a little multidisciplinary on some topics. As we discussed it, I will also talk about AI. But that’s more recent.
[00:01:56.610] – Jonathan Denwood
Yes, he’s been. He’s done a few presentations on the subject, which I’ve watched. He’s done a fantastic job there. So, I thought he would be a great person to have on because he’s a web developer, he’s into WordPress, and he’s into AI. What more could you ask for? Beloved tribe. So, Kurt, would you briefly introduce the new listeners and views?
[00:02:20.220] – Kurt von Ahnen
Jonathan. My name is Kurt von Ahnene. I own an agency called MananaNoMas. We focus a lot on eLearning and membership-type websites, and I also work with the great folks at WP-Tonic on the team and LFTR LMS.
[00:02:34.420] – Jonathan Denwood
That’s fantastic. So before we go into the meat and potatoes of this great show, I’ve got a couple of messages from our major sponsors. We will be back in a few moments. Folks. Three to one, we’re coming back, folks. I would also like to point out we’ve got a sponsor for the next couple of weeks called Motopress. They make a fantastic appointment booking plugin. You can find details about Motopress and the other sponsors, plus a list of the best WordPress plugins and services for WordPress professionals or power users. You can get all these fabulous goodies, these free goodies by going over to wphantonic.com slash deals, wp tonic.com deals. What more could you ask for? Beloved tribe? Probably a lot more, but that’s all you’ll get from that page. So, um. Smile, Kirk. You don’t smile at that. So there we go. So, let’s delve in. So, Max, maybe you can give the listeners and viewers how you got into the world. You seem a very nice person. I get a nice vibe from you, Max. You’re laughing at my jokes, my terrible jokes. But how did you get into the crazy world of web development, and how did your interest in AI start to develop, Max?
[00:04:05.150] – Max Ziebell
Oh yeah. It’s a fun story for me because I started somewhere completely different. I made a century of journeys because I began on the island of Formentera in Spain. It’s like a tiny island in the Mediterranean Sea. My parents were hippies. They stepped out of society. My dad, my stepdad, was a wood carver. So we had no electricity, no technology, no running water. We had a well, fireplace, beautiful beaches, and all that. That was my childhood. And so when I came to Germany because my mother moved there because of employment, we went straight to Berlin because she’s German. I was in my teens and got into computers; it fascinated me because I never had anything to do with computers. It was only the machine occasionally in the main town of Formentera. So when I came to Berlin, I lost all my friends from Spain. I got into a nerdy little circle, Sci-Fi, and everything else. I had my first computer, CPC 600 4646 Amstrad. It’s really like it had a tape deck and everything. I got into programming, and I was still a team member, and that’s the background of where I’m coming from.
[00:05:24.520] – Max Ziebell
So I had like this really big jump from no technology into the midst of having this early goes at programming. And then I switched back to being living in a commune in Berlin. And that was like cultural industry. I worked there. I did set design and all that kind of stuff. Stuff. But that was like the mid end of nineties. And then it came up id software, and we were doing like game mods. We were hacking around all the youth there in the commune, and I did the first web design for our commune. We were like a cultural thing. We had like stages and a lot of variety and different kind of sets, programs. And then over that kind of connection, I really got into the whole.com era because that was also a thing here in Berlin. I mean, America kind of pushed it, but then suddenly it was also here in Germany. And that was tease. Yeah. And I got to meet some really fun people. They did like, graphic design, paste up. So they were like the old school people still doing, because all the people were working on little machines, the designers, because they couldn’t afford big ones.
[00:06:32.760] – Max Ziebell
And then we had these big machines and we would rebuild designs for them. So I got into that kind of work, agency work. Flash, web design, HTML, all that kind of stuff happened to me. So, yeah, I did that until after the.com. and I was always working for agency, but I didn’t have my own agency. So in 2001.
[00:06:59.770] – Jonathan Denwood
Oh, dear, he’s frozen, folks. Hopefully he’s gonna come back. We might have to start afresh.
[00:07:09.240] – Kurt von Ahnen
Everybody raise your hand if you miss Flash.
[00:07:12.960] – Jonathan Denwood
Oh, I say, I got into it, actually. It was really influenced. It was all the rage. And I started doing some work for agency and it. The program. Can you send a message to Max through chat saying he’s totally frozen? Because.
[00:07:31.740] – Kurt von Ahnen
Yeah, we can do that.
[00:07:33.560] – Jonathan Denwood
Because I don’t. I’m not sure if he knows that he’s. You. You totally froze Max, actually. So maybe you can tell us how you got into AI as well. Can you go?
[00:07:46.600] – Max Ziebell
Yeah, sure. I didn’t want to get too far. I just want to. Where did you.
[00:07:49.950] – Jonathan Denwood
Where did I stop the flashbit?
[00:07:54.890] – Max Ziebell
Okay. I just wanted to say that from that flash times, I kind of was working for agencies. I got my own agency is 2001. Kind of got into their whole world. Optimizer, we called ourselves, and we did our own cmss at the time for Flash and for HTML. And how I got into WordPress, just that little story. And then I’m going to go over to AI, was that I was actually doing my own cmss, but I also was side eyeing text pattern. I don’t know if you guys know that that was like a thing slightly before WordPress from gene. Ellen. I was also in that vc 200 fund. We were kind of getting that thing going. I think Matt Bollenberg was also connected to Dean Allen at the time, and then WordPress ten came out. So I’m really always in this, in the whole crowd because I was watching how WordPress always developed because I saw it from the get go. So it was very interesting to me all the time. Yeah, but we can fast forward now. There’s more that happened.
[00:08:56.080] – Jonathan Denwood
Yeah, we can go over you, FedEx. We’ve got very similarities, got similarities in our background. So how did, how did you get into. Because I, I got into it through flash and action scripting and. But how did you, how did your interest in AI start to develop.
[00:09:19.800] – Max Ziebell
Early on? Actually also nineties, because I was also in computer science. I was into science fiction, as I said, and I really loved all that. And I kind of was a library child and I was going through the books about neural networks. And then there was a long AI winter, and I wasn’t in research or anything. I was doing my life, as I explained. But then I don’t know how I got into it, but I was on the waiting list for openaiden. And when they opened up their first API, I think that was 2021 around there with GPT-3 base model, I got an invitation. So that’s when it started for me. And I was totally blown away that you had now a voice interface or not voice language interface to computers. It was still pretty basic, and it wasn’t that what we know now. It wasn’t a chat interface or nothing, but that’s how I started.
[00:10:10.830] – Jonathan Denwood
That’s fantastic. Over to you, Kurt.
[00:10:14.150] – Kurt von Ahnen
I really like the stories about how people got started or what their background was, because it seems like WordPress or, you know, let’s just say working on the Internet, it just welcomes people from all walks of life. Right. It’s just such a cool thing to see that everybody can get into that. If we stick on the AI subject, though, can you kind of describe, and I don’t want to put words in your mouth, but like there’s language models, there’s, people are talking about generative AI, general intelligence. Like, could you give us kind of a maybe segment with the types of AI? Are you familiar enough in those terms to fill us in?
[00:10:57.950] – Max Ziebell
Yeah, well, I started also with the lmMs. We all used. There was technology before the LNMs that was basically called convolutional networks and recurring neural networks, but they were very narrow AI, you would call them. They were one task, one solution, probably like pattern recognition of faces or text, like those kinds of things that existed before. But the breakthrough then came with LMM’s. A transformer technology that with backpropagation and all that, that kind of came into the. Into the GPT-3 for open access. They had it internally before already, but then first time we could see the recurrence of neural networks again, and then having a big symbolic space mapping out everything language related. And that was like a little big breakthrough at the time. And your question was how that differs from. How was the question again, kind of.
[00:12:03.390] – Kurt von Ahnen
Like the different classifications of AI. And I’ll tell you where I’m kind of coming from with my question. The way I see it is there’s this group of people in WordPress in the Internet that work like us. We’re professionals, and we talk about AI kind of at that level, right. And we know that there’s some things we can do and we know that can make some tasks quicker, but you shouldn’t be fully dependent on all those things. But I feel like there’s a whole segment of our audience that imagine AI to be this weird, magical thing that does things that it just doesn’t do yet, too. So if you could kind of describe those different elements of AI and kind of what they do, that’d be great.
[00:12:47.880] – Max Ziebell
Yeah, well, as I said, we have now large language models, and they basically have mapped out a huge amounts of data and text. They started with text, now they went also multimodal, but that’s a little later. They mapped out the symbolic connections between a lot of things in the world, and they kind of found also rationals. There’s no personification I want to do because I said that they kind of found the algorithm they’re using kind of uncovered those connections between a lot of concepts. So it’s called a latent space. It’s more symbolic than actual language, and you can extrapolate text completion from those. That was the main goal. You can get the text completion with the really base models. They came out with, like GPT-3 and when you use it back then, you had an area was called the playground. You could go in there and then you just had a. A single text box, and you basically could enter a sentence and then send it to the language model, and it would try to complete it, but it could also get into complete gibberish or into a loop. And so it wasn’t really good at the time, but open.
[00:14:08.860] – Max Ziebell
I kind of made it better with a new technique that’s called reinforcement by reinforcement technique, and that made the networks behave more like in a chat interface. They could complete sentences in a more useful way to people using them. And that’s when kind of was when the public was invited to the entire thing. It’s around end of 22 with chat GPT, and then it blew up. But it was the funniest thing. It was middle or beginning 2021 for me, and I was already totally blown away by that simple text box. It was in the beginning and I was telling everybody about it and people didn’t really understand the moment they kind of came out with this chat interface. It totally blew up. Everybody got it and saw the potential in it.
[00:15:04.460] – Jonathan Denwood
Well, that’s got similarities. I just want to see similarities to the web itself because you had web, you know, the universities had access to the web much earlier or very basic kind of, and then you had Firefox and, and they built the first browser and then the, then the Internet kind of just blew up with that interface. And that seems very similar to what you’ve just said about the chat interface. Do you think I’m on the right track there, Max?
[00:15:48.990] – Max Ziebell
Yeah, it’s totally, basically the browser made it popular for everyone. And I think the chat interface did the same thing here. Funny enough, I think it was half a year before there was this guy on the Internet called Doctor Alan D. Thompson, and he basically used that really simple API and created a chatbot. And then he had a conversation with it, but he prompted the whole conversation. There was no real chat interface, but he simulated the whole thing and then he fed it into one of these talking avatars you had at the time already, and he called the whole thing a lita. That really blew me away because there you really saw already the conversational nature. And I tried myself. You could initialize kind of chat interface where you would simulate a chat by writing speaker and then answer, question answer, question answer, and then it would complete always just a question in a long text block. There was no interface. And the funny thing, I actually simulated a conversation with Einstein at the time. It got into a loop after a while, but it was really funny because you could feed it all this information like everybody’s used to now, but it was like a role play now.
[00:17:01.400] – Max Ziebell
I was into role play for a while now. I did role play with my son based on that kind of concept. And that was before church came out. It was, but this interface made it accessible to everybody. And one thing I really want to drive home from that era is when you try to simulate a chat by yourself and you just put, you have this one text box and you just put like question answer, question answer. And if you wouldn’t have a stop signal, it was called a stop signal, a certain string of text that would be passed just like a regular expression. If it would find that string, it would stop generating tokens or words. And if you wouldn’t put it in there, the AI would just talk to itself. It would pose a question, it would pose an answer, it would pose a question. You really had the feeling, I’m not relevant here, it’s just completing a virtual discussion with itself. And I think that drove it for me home, that systems like chat GPT are simulating a conversation and always just breaking off at the moment where they say, now you have the chance to say something, but the completion algorithm would actually be glad to talk to itself.
[00:18:16.050] – Jonathan Denwood
Yeah, bit like me, really. So the language models. So with OpenAI, was it a gradual improvement of the basic mathematics and the technology, or did OpenAI develop something new which really expanded the capacity of these language models? Because my understanding that the theory and the basic knowledge has been around quite a while in the AI communities from the earl, from the early mid fifties, and it’s coming waves and been over built up and then it’s died down again. But did AI develop something, or was it they built this interface and they had the money and resources to offer it to the general public, or did they add something unique to the mix?
[00:19:27.800] – Max Ziebell
A little bit of both, because, yeah, OpenAI was mainly a research organization at the time and rolling out something like chat GPT. That’s why there was probably also all this delay that also needed a big infrastructure partner, and that they got that with Microsoft. And before that, I think that it wouldn’t have been possible to roll out chat GPT like it’s available now to millions of people, and it’s the same as with WordPress, if you take the parallels here, back then, when WordPress zero came out, it was a little crew of people using it and it blew up. And the same thing happened here. I mean, when I got first invited to the API, the few channels covering it, and I was totally hyped about it and nobody else was talking about it on the web, and I was like, that’s revolutionary what’s happening here? And it was, I think it was a few or 10,000 people, I guess, or something like that, and now we’re talking about millions. And sure, they added a lot of infrastructure. They also added, like I said, after the base GPT model, they added a technique called reinforcement learning from human feedback.
[00:20:39.510] – Max Ziebell
And that’s like tacked on to make it more human friendly and useful. And then they also, with that technique. They brought out an instruct model that could program better or understand code better, and then they brought out this chat GPT interface. So there are little parts that went together, interface back end and technology refinement. The base technology is still the same that made it successful.
[00:21:08.830] – Jonathan Denwood
So in your opinion, it’s got very similar, what they’ve done is very similar in obviously different, but has similar flavor to what Netscape did with the bright, with the browser. There are kind of similarities, obviously. So they didn’t fundamentally add a to the background technology, but what they did around interface and usability is, can’t be diminished. It’s still quite important. Do you think I’m on the right track?
[00:21:47.260] – Max Ziebell
Yeah. And even if I’m totally, I remember my Netscape three days fondly. And even if Openaid is like overtaken by some other big company or some new technology breakthrough comes, I think they have laid the groundwork and will go into the history books for being the company, even if like in the case of Netscape, they go away. But I don’t think they’re going to go away that soon. And because there are.
[00:22:18.460] – Jonathan Denwood
He’s frozen. He’s frozen right in middle. Interesting phrase. Obviously Berlin doesn’t have a good connection. Kirkland. Oh, he’s back again.
[00:22:29.390] – Max Ziebell
Okay, I’m back, I’m back.
[00:22:30.860] – Jonathan Denwood
Is there something wrong with the Internet in Berlin? So always frozen again a little bit. He’s back again. He’s back again. So, yeah, you’re back again. I like to cover this, this bit. Are you there, Max, or you’re frozen?
[00:22:51.340] – Max Ziebell
Yeah, I’m there.
[00:22:52.010] – Jonathan Denwood
I’m here to, so this thing about the language model and general AI, I’ve listened, obviously I haven’t got a high formal education in mathematics and I’m not an expert, but I’ve listened to a lot of resources. I’ve watched your latest presentations on the subject as well, and I think you’ve done an excellent job, Max. That’s why I thought you would be a good guest to invite to the show. But I’ve never observed such contradictory informational views from a set of experts in my life. Actually only, and not, not normally when certain experts have paid a lot of money to have a certain view, but these people seem to be extremely honest. But I’ve never observed such a different view. So I’ve observed people that have worked for OpenAI and who are experts or scientists in the field saying that in the next five years, in the next three to five years, they actually think that General AI will be available and it will be the biggest change in human society ever observed. And then I’ve observed other scientists, AI scientists, saying that there’s a total inflation between what a language model can do, general AI.
[00:24:42.950] – Jonathan Denwood
And it’s, they say that it’s been totally inflated. So what’s, what’s your general view and why do you think, why do you think there’s such a mismatch in what the experts in the field are saying? Is he frozen? He’s frozen.
[00:25:07.770] – Kurt von Ahnen
I do believe he’s frozen.
[00:25:10.420] – Max Ziebell
Okay, I’m back. This is, I don’t know what’s going on today. There were some construction works in my house. Okay. So, yeah, your last question was, what are my views about those glaring optimistic predictions?
[00:25:29.620] – Jonathan Denwood
Well, such a, would you agree, it’s really quiet. I, different, what these different experts in the same field are saying. Would you agree with that element of what I’ve outlined to you?
[00:25:43.970] – Max Ziebell
Yeah, totally. Because we also have to keep in mind, if you have, we talked about it before. If you haven’t, the growth from 2021 to now, with all that hype cycle was after 2022 with ChedJPt. There’s been such an influx of money into that sector, you can’t take everybody serious because if they wouldn’t be tooting in their own horn, they wouldn’t maybe get the funds they’re getting. And you have to be pretty optimistic. But there is, if you stay on the ground level of optimism, there is a lot of potential going forward with this technology. And just the latest releases have shown that there is still gains to be had. I, and there doesn’t seem to be immediate ceiling, but there is also, and there are a lot of doomers that actually see that as well, and they go in the other direction. They are people that kind of see the dangers more than the benefits and are talking about the end of humanity. Based on that quick change, I’m a little more grounded and I’m a user, as most are. I use AI for my daily work and I kind of see the benefits in my day to day, and that’s where I’m judging AI from.
[00:27:14.480] – Max Ziebell
I also watch all the news coming out, but I judge it by what’s available.
[00:27:24.990] – Jonathan Denwood
Before we go to our break, I know we’ve only covered one question, but it’s the biggest two questions, but the second question on our list is a very big question. Before we go for our break, is there a real clear difference between these language models in general AI, or is there a lot of overlap between a language model and the concept of general AI, general intelligence? Because is it so overlapped that you can’t really put the two in two different buckets? Some people seem to be suggesting what you can do with a language model really doesn’t affect the concept of general AI, general artificial intelligence. How would you respond to what I’ve just said?
[00:28:28.090] – Max Ziebell
Yeah, it goes a little bit into a philosophical discussion and it’s also just from a technology perspective, there’s already a split in the community. There are many people, especially on the OpenAI side, that believe that the scale paradigma to just get bigger and bigger. And also now that doing a little bit of architecture changes in their system will lead to something like AGI. And then there is also more, another sort of school that there has to be a step change, some, some breakthrough to have actual AGI. But that brings us to the term AGI, and that is a very undefined term. So we can go into that in a moment if you want.
[00:29:18.440] – Jonathan Denwood
Well, we won’t do that at the end, but. Because we need to get through the article. But this, this is one of the biggest questions of this interview and I’ve been thinking about this myself. Like I say, it’s extremely interesting, but also extremely frustrating because there’s so many different contradictory views and experts saying complex.
[00:29:43.030] – Max Ziebell
Well, it’s an open question. So there has to be that kind of mindset because there is no clear answer yet because the goal hasn’t been achieved and there are different ways to get there different people believe in. But as you said, that’s one thing that certainly is true. It’s a gray area and it’s certainly.
[00:30:06.050] – Jonathan Denwood
But Max, Max, I’m a simple soul at heart. I just want to. We’re going to go for. We’re going to go for our midbreak. Hopefully you have got something from that conversation. We’ve had some technical problems. Max from Berlin hasn’t been paying his Internet bill, so there we go. But we’ve struggled a few and hopefully you got something from the first half. We’ve got some fabulous questions for Max in the second half. We will be back in a few moments, folks. Three, two, one. We’re coming back, folks, with. We’ve had a feast, a feast of AI. I think I’ve lost Kurt and I think I’ve lost Max most of the way. But. But I think we’ve had a good discussion. It is a hot topic, but before we go into the second half of the show, I want to point out if, if you’re looking for a great WordPress partner hosting, but much more for your larger membership or community focused websites or any large project. Why don’t you look at becoming a partner with Wptonic? We’ve got over ten years experience in helping freelancers with their bigger projects. It’s fantastic to have a reliable partner.
[00:31:36.210] – Jonathan Denwood
The background to help you. We offer some great packages to our freelancer partners. You can find more by going over to wphantonic.com partners, wphantonic.com partners and why don’t we build something amazing together? So I’m going to throw it over to Kirk again for question three.
[00:32:07.080] – Kurt von Ahnen
Jonathan’s going to get mad at me because I have a curiosity that I need to ask you, Max.
[00:32:12.880] – Jonathan Denwood
So you’re not going to keep to the outline of the questions? That’s what you said, really, I’m so.
[00:32:18.210] – Kurt von Ahnen
Curious about this and it’s an extension of question two, but it’s not one of the staff questions. When I have the unbelievable pleasure to travel outside of the United States and interact with people in other countries, I’m always amazed at how many languages people know. So you say you’re in Germany, I assume you speak German. You probably, I don’t know, speak Swiss, French, French, Italian, who knows how many languages. But when you use AI, when you use AI, because Americans are so arrogant and because we build a lot of these tools, do you find yourself normally working in English or do you get different results if you put in different languages for your questions and prompts?
[00:33:04.510] – Max Ziebell
That’s, that’s, that’s a great question. I usually work in English. I’ve been actually, as I kind of outlined in my, in the beginning, I’ve brought, been brought up by an American in Spain. So I was learning Spanish and English before I learned German, but I still think in English. And for me it’s English. Also, the biggest chunk of data the AI’s have trained on is English, but they kind of looked into it. Certainly OpenAI has done it and we have also really big chunks now of German and all these kind of european languages. And it’s pretty fluent in German. I can also talk with it in German and when I go, when I use it with the kids, certainly German. And yeah, so yeah, I was really.
[00:33:52.910] – Kurt von Ahnen
Fascinated by that one little chunk because I didn’t know how that would work. If I stick to the outline, the next question is how do you think AI is going to affect web development? Or even more fascinating to me, design? I’m graphically handicapped. I’m great at taking assets that customers give me and making awesome websites or using designers on my team that are graphical but I am not personally a graphic designer. So how do you think AI will affect web development or design over the next couple of years?
[00:34:25.610] – Max Ziebell
Okay, when we look into the future, we already, I mean, many of the listeners will probably know about that. There are AI’s that can create images. There are a eyes that can write text. Now lecturegbt. So you have all these disciplines and the models are actually getting multimodal. So the latest models of OpenAI actually can produce images. We are still getting images from a separate model because they haven’t rolled it out yet. But actually the model can itself dream and create images. All those kinds of capabilities can be used and I already use them. Yeah, in many cases you still see that they’re AI generated, but it’s really evolving really quick. And I use, for example, I use a lot of the image capabilities for placeholder images or text placeholders in my design work. And lately you can even request websites, designs from AI’s or also image AI’s or specialized AI’s, and you can use them for quick blue blueprints or to just present something quickly to the client if you want to have options. So where I see, was that the question, or where do I see it going?
[00:35:44.030] – Kurt von Ahnen
That’s kind of it. I mean, thinking about it from an agency perspective, some agencies are actually like almost living in fear that a year and a half, two years from now, a customer is going to be able to walk up to a microphone and say, create me a website with this name that can make appointments and is about landscaping, you know, and like, poof, here’s your website. To a certain extent, some of that’s already true, but, you know, how simple will it be? Like, I don’t know. Do you think it would get to the point where it would be literally that simple to just tell it, to make it and put it on this URL and snap your fingers and it’s done.
[00:36:25.050] – Max Ziebell
Yeah, I think it will get there. But then again, having somebody create a website for you is not only the task of creating a website, it’s positioning you in the market, keeping you holding your hand and so forth. So there’s a big market still going to be left over. And even if the client could do it, I mean, if you’re honest, I was always fascinated with CMS’s, even though I was also writing direct code, because I always thought, okay, I can hand it off to my client or my client can do something in that system. That’s why I actually started with all those kind of systems like text pattern and WordPress. Afterwards. But there were so many times the client didn’t even want to go in there. And I think for a long time all the systems are going to be really helpful. But to the developers and to the agency owners and sure there’s always going to be a client that says I don’t have the budget, I’m going to just talk into a mic and have my website done. And maybe those don’t need an agency because they’re too small. Maybe that market shifts a little bit.
[00:37:30.800] – Max Ziebell
But overall I think the agency market’s not going to go away anytime soon.
[00:37:36.670] – Kurt von Ahnen
Yeah, I think there’s a lot of insight in your answer there. It’s very similar to the answer we got with Kevin Geary. I don’t know if he can hear me or not because he’s frozen. But Kevin Geary operated under the pretense that customers generally wouldn’t.
[00:37:50.410] – Max Ziebell
I’m back.
[00:37:51.730] – Kurt von Ahnen
Good, good. We interviewed Kevin Geary and he was very resolute in the idea that when he works with a website from an agency perspective, you know, customers wouldn’t be making edits or doing things. And to your point, you know, a couple hundred websites out of the shoot and everyone says they want access to edit it but then very few actually jump in and edit anything. So. Yeah that’s a really good point. Jonathan, over to you.
[00:38:16.530] – Jonathan Denwood
Yeah, comment. You know, at the present moment you’ve got code pilot and you know to me it seems like a UI, a coding UI interface on steroids. Will the tools in the next two to three years get to a level wherever I. It replaces a lower to intermediate coder. You know, would it get to the level where you would need to know what you’re doing? You would need some education in the field. But a lot of the present work that a lot of developers are involved with has been removed. Do you think in the next two to three years we’ll get to that level or is that something that’s out of the scope of the foreseeable future? Max?
[00:39:16.390] – Max Ziebell
No, totally. We will get there and we’re already having. I mean you mentioned Microsoft Copilot. It’s like a text completion on steroids. We already have systems also now much newer like cursor or. And that are more integrated ides and have much more capabilities already. You can chat and do a whole multi file projects and stuff like that. It’s already happening. And next is coming Devon, I think it’s called. And also systems like the just rolled out zero one from OpenAI that can actually mull over our idea. With reasoning, a little bit like it’s fake reasoning, but it’s still reasoning. And we are getting a system called Orion next year. I guess it’s somewhere around there, but that’s still in the tea teapot or how you say, in the tea leaves. And I’m totally convinced that certainly beginner and like mid tier developer will be not replaced but enhanced and will be at the level that they have coworkers, agentic co workers that are based on AI that could help them or work with them together, even working on multi step tasks and come back with the results.
[00:40:34.260] – Jonathan Denwood
So kind of based on your reply, let’s go on to the next question. Let’s say you had a young relative or a friend, a relative of a friend, a young person that’s thinking of going to university or getting into web development or that side of the business, and they’re asking you, should I really focus on this or is it really not a field that I should spend any time in? How would you, what advice would you, general advice would you give them? Is it something that they should still focus and they have the prospects of having a reasonably long term career?
[00:41:20.310] – Max Ziebell
Yeah, I would with some caveats, but yes, I learned a lot by going into the fundamentals and the web fundamentals. And even with systems producing at the level of a developer, it’s like if you were working in a company and you get that code handed, handed to you, you still have to check it into your company workflow or into your repository. And I think for the, for the future, for the, there is still going to be human oversight in either way, that that requires also mastering fundamentals to validate the AI outputs. Also, I would tell the young person, come on, embrace AI. Well, no, that’s something I would tell older persons. I would say, tell them, embrace AI as a tool and not as a threat because it’s kind of something you have to live with. That’s our new world we’re in. But the young person, I would probably tell, learn the fundamentals so you can actually validate the outputs. So. And work with it together.
[00:42:18.280] – Jonathan Denwood
All right, over to you, Kurt.
[00:42:21.160] – Kurt von Ahnen
Well, now we’re all kind of curious. You know, the listeners and viewers have heard you talk about AI. So what are the tools? What are the main tools that you use in your daily work? Like, if you choose a tool, I am not just, you know, I’m not a prompt expert, so I use tools that help me cheat. So what tools do you like to use?
[00:42:45.040] – Max Ziebell
Yeah, in the beginning I was very fascinated about exploding market because there was so much influx of money and startups and I kind of booked some of them. But in the end I, I got like a tool fatigue, fatigue. And also I realized because I was programming myself with the OpenAI API, that many of those tools were like OpenAI wrappers with little prompt engineering behind them sprinkled on. And then they wanted to have a subscription from me. If it’s really useful, I would pay it. But usually after a while I found out how to do it myself with the base model or with jet GPT and the tools I use. That’s why I kind of reverted to, I settled on the bottom up approach. The tools are at a current time in history, they are most effective in generating code still, because that’s a really big chunk of information they have ingested. So I really like to use them on a code editor level. I use them as we discussed with Microsoft’s copilot. I went in into also tools like Cursor. Now recently I just tried it out, but it’s pretty new and I use the interfaces from chat GPT from also recently claude.
[00:44:03.950] – Max Ziebell
And Claude artifacts are really great. They give you a visual feedback for boilerplating stuff out or doing a little code explorations. And those are the kind of tools, and one big tool we could also maybe mention I’m using a lot is perplexity and that’s kind of search engine based on AI that for me, it has replaced Google and there’s only not that many kind of moments when I go back to Google here and there I go. But that will be going forward because OpenAI is also going to probably release something along those lines called Search GPT that’s going to be changing the whole SEO market and it’s going to be interesting times coming ahead because a lot of people in the WordPress ecosystem work in that segment.
[00:44:51.100] – Jonathan Denwood
Yeah, well, how do you see that chat? You know, because I think it’s dependent on your ability to verbalize or write prompts that can actually in, because you’ve got that experience. But you can, but you can get into a technical bubble where the majority of population, if you don’t really know the how to present the question, the prompt, you’re not going to get you. I would imagine that you’re going to get very, you spend as much time messing around with the prompts as you or the verbal commands if you’re doing it verbally, as you would do in doing a general Google search. But I’m only making that assumption. How do you respond to what I’ve just outlined.
[00:45:49.290] – Max Ziebell
Yeah, in the beginning I witnessed that as well. As I said, I was very optimistic before chat GPD came out and I was running around telling everybody around about it, and then I just set them in front of my computer in the playground. There was like one text box. I said type something. And they were like completely shell shocked. I didn’t know what to do. And I think a lot of people are having that experience as well. Or they just type in one question, they don’t get the completely right answer or the answer they were expecting and they kind of give up a little. Yeah, in the beginning you needed prompt engineering, but that is a thing that’s going away slowly, especially with those search engines. You can just take natural language, ask youre question and you get into a discussion about your topic and it’s individualized to you. So if you, and I really love that OpenAI has offered web search inside of their interface as well. So in chat GPT, you can actually, if you, if you want to have in context learning, that means that inside of a conversation you’re having with the AI, you can always prompt it to research something and it goes out into the web, gets some information back and puts it into your conversation.
[00:47:00.480] – Max Ziebell
Yeah, it’s a skill, but it’s going away that there is a need to learn it because the latest models that just came out, like, was that like a week or two weeks ago or one week, I guess zero one. Those models get a little hidden space to prompt themselves. So you kind of put in your question, then the model, the latest models mull over the, the intent you’re actually putting into the question and they come back with a much more relevant answer or even with a question with follow up?
[00:47:30.240] – Jonathan Denwood
So could you envision this army of SEO agencies and experts that their purpose will be to manipulate the AI models to, that’s, they, they won’t be interested in manipulating the end user. Their interest will be to manipulate the AI agent that produces the result to the end user.
[00:48:03.170] – Max Ziebell
Yeah. So they still want to have the user to buy something, but yeah, they will have to play like in billiard over the border. They will have to try to get a good ranking. I mean, I still search the web, even if it’s like a semantic search, more symbolic space, latent space, but it’s.
[00:48:20.290] – Jonathan Denwood
Still, I’m trying to interrupt. Can you speak? Can you see where I’m going that at the present moment you could use the AI agent to give you a recommendation and you can to some degree feel that that’s going, that agent that AI agent going to give you what you consider an honest reply. But if the agent is being totally manipulated by outside forces, you no longer have that confidence that you’re being given the answer that meets your honest requirements. Can you see where I’m coming from, Max?
[00:49:06.380] – Max Ziebell
I totally see that and I guess that’s also why we have different product lines for those tasks. And we are seeing that even though the OpenAI can actually search the web when you’re using chat GPT and you can say search GPT and so they’re coming out with a new product search GPT and I, that’s kind of copying the product that’s also very popular currently, perplexity. And they are in a way, there are conversation, but they also pull in all those web references. So it’s not such a big shock and not such a big change at first. And you probably could also still sell ads at the side, maybe not injected directly into the responses, but they’re creating a space where they could probably be an intermediary solution between direct chat interface and a search engine. So I think there is still going to be the diversity for a user to also see alternative results for a while at least.
[00:50:18.140] – Jonathan Denwood
Yeah, I’m a bit more cynical because of this. I can see it being totally manipulated and people. But that’s the.
[00:50:26.510] – Max Ziebell
Yeah, but let’s be honest, I mean, Google is a dumpster fire. It’s, I mean, it’s a sewer.
[00:50:32.250] – Jonathan Denwood
It’s a sewer, Max, it’s a sewer.
[00:50:34.580] – Max Ziebell
So it can only get better. And that’s why I’m using perplexity because I really hate using Google these days. I have to go through all those.
[00:50:42.450] – Jonathan Denwood
Unfortunately, I think unfortunately, the cynical side of me, Max, feels that what’s happened to Google will happen to your beloved perplexity, unfortunately, over to you again, Kirk, you have the likes I know you’re going to disappear in the next four or five minutes.
[00:51:02.180] – Kurt von Ahnen
Yeah, yeah. So I think it’s our last question, Max, and it is if you had a time machine like doctor who or think about HG wells, you travel back in time and you go to the beginning of your career, right, hanging at the commune and you go back to the beginning of your career.
[00:51:20.840] – Jonathan Denwood
What you call me backs you combo you copybacks, marxist backs.
[00:51:27.890] – Max Ziebell
What cultural common?
[00:51:30.600] – Kurt von Ahnen
What advice would you give yourself knowing what you know now?
[00:51:38.520] – Max Ziebell
Well, maybe invest in Nvidia a little earlier. No, I’m just joking. No, I probably, when I would really, I would probably with all this technology background and all, but we’re talking about. I go back to the humanities. I had a time. I was so invested in all this that I bought my MacBook Titan at the time. I was in Spain back there. I should have gone to my stepfather, who was a wood sculptor, and done a little more work with him because he passed away after that. So I’m sad I didn’t have my priorities right then. And after all this talk about technology, for me, that, that’s the thing where I don’t want to end on a sad note, but it’s kind of, after all, humans are more important than machines, and I would like to end on something like that.
[00:52:26.400] – Kurt von Ahnen
Yeah, I don’t know that I could. I don’t know if I could agree more with you on that. It’s. I’m in a phase of life where I am working tirelessly, knowing that I’ll be able to unplug just a little bit and spend more time with kids and family and community and stuff. It’s super, super important for me, personally. The more prevalent AI becomes, and the more focused people get on technology, the more I want to have a fire pit and, you know, a bottle of red wine and enjoy the company of others.
[00:52:56.380] – Max Ziebell
That’s also the big problem of AI, in a way, and we have to see how that plays out because all those big investors have put down so much money, and they want a return on investment. But in the long run, as I said in the beginning, I’m a science fiction fan. I would hope for something like a Star Trek in a universe where we have more time, leisure, or work as a passion. I see that you can define, oh, more time.
[00:53:21.300] – Jonathan Denwood
More time to get into mischief, Max. Oh, he’s right. Frozen. But that, you know, is he going to come back?
[00:53:32.750] – Kurt von Ahnen
I don’t know. I don’t know, but you.
[00:53:36.660] – Max Ziebell
Why not? I’m not mischief, but I got a lot often. There’s also.
[00:53:40.990] – Jonathan Denwood
Yeah, well, I think it’s time, Max. Max, you’ve been fantastic. We’ve solved our technical problems, but I think it’s time to end our discussion before your meter runs out, Max. But you have to come back in the new year. You must return in the new year because I think this has been a fun discussion, Max, and you’ve done a fantastic job. How can people learn more about you and what you’re up to? Max.
[00:54:12.540] – Max Ziebell
So you can find me on my website in Germany, and you can find me at Max Siebel on Twitter and with my agency. It’s called worldoptimizer.com. I’m happy to hear from you. And we certainly left the philosophical topic of AGI on the table.
[00:54:36.310] – Jonathan Denwood
Well, you have to come back in. You must return in the new year, and we’ll discuss that. I can get into Wittgenstein with you and Turling. So, Kurt, what’s the best way for people to learn more about you?
[00:54:51.820] – Max Ziebell
Yeah, do that. Let’s do some Daniel for business.
[00:54:55.620] – Kurt von Ahnen
The best way is manana, nomas.com, or anything on the Internet. Manyana Nomas generally points to me, but if it’s personal and you just want to make a connection, I am on LinkedIn almost every day, and I am the owner of Kurt von Ahnen on LinkedIn. Easy to find, easy to connect.
[00:55:11.710] – Jonathan Denwood
Please share this show on your social media platforms. I honestly feel this is the only WordPress podcast where you find these types of discussions. Hopefully, you’re enjoying them. They are a bit different from the standard fare that you will find in the WordPress podcast world. Please give us some feedback and share it. Our audience is growing. Thank you so much, and we’ll see you next week.
[00:55:41.500] – Max Ziebell
Bye.
WP-Tonic & The Membership Machine Facebook Group
Why don’t you sign up and be part of the Membership Machine Show & WP-Tonic Facebook group, where you can get all the best advice and support connected to building your membership or community website on WordPress?
Facebook Group
