[00:00:00.000] - Nathan Wrigley
Welcome to the WP Builds podcast, bringing you the latest news from the WordPress community.Now, welcome your hosts.David Waumsley and Nathan Wrigley. Hello there and welcome once again to the WP Builds podcast. You have reached episode number 315 entitled Thinking the Unthinkable, Episode 2, AI Will Kill Us. It was published on Thursday, the 16th of February 2023. My name's Nathan Wrigley and in a few moments I'll be joined by my good friend, David Woaumsley. But before that, some very short housekeeping.
[00:00:48.240] - Nathan Wrigley
The Page Builder Summit version 5 is coming around. It is happening next week. So that's the 20th to the 24th of February 2023. If you haven't already signed up to be a member of the summit, do so now. Go to pagebuildersummit. Com. We've got heaps of presentations all surrounding WordPress and building your WordPress website. It is going to be a fabulous event. It's completely free. If you would like to watch, you can see the videos for the first 48 hours after they have been released. But there is an upsell should you be unavailable or you want to get the bonuses that many of the speakers have provided. There's quite a lot of value in there. You can do that by signing up and purchasing what we're calling the power pack. Go to pagebuildersummit. Com, sign up, and I hope to see you there.
[00:01:40.920] - Nathan Wrigley
The WP Builds podcast is brought to you today by Go Daddy Pro. Go Daddy Pro, the home of managed WordPress hosting that includes free domain, SSL, and 24 7 support. Bundle that with The Hub by Go Daddy Pro to unlock more free benefits to manage multiple sites in one place, invoice clients and get 30 % off new purchases. Find out more by heading to go.Me/WPBuilds. That's go.me.WPBuilds. And truly, sincerely, we thank Go Daddy Pro for their continuing support of the WP Builds podcast.
[00:02:21.080] - Nathan Wrigley
Okay, this is episode two of the Thinking the Unthinkable. The acronym is TUT, TTUT. It's perfect. It's episode two of that series, and we're exploring the fact that AI will kill us. It's interesting because David and I have really different approaches. David is much more sanguine than I am. I'm a bit afraid, in all honesty, of what's going to go on. So we chat through all of that today. What better than to get ChatGPT to kick the conversation off by telling us whether or not AI will kill the web design industry?
[00:02:51.760] - Nathan Wrigley
I think it's fair to say that a lot of the things that we assumed AI would never be able to do are now definitely possible. Creating images, creating music, writing text. What will be left for us, the web developers, the WordPress website builders, will we be completely replaced? Find out what David and I think. I hope you enjoy the podcast.
[00:03:13.430] - David Waumsley
Hello, welcome to the second episode of our Think In The Unthinkable series. Here we attempt to rationalize negative views of WordPress and web design, and it is most definitely not only a ploy to create click bait titles for the show. Today's topic is AI will Killers, and that's artificial intelligence and not artificial dissemination. Oh, good Lord.
[00:03:35.060] - David Waumsley
Which might kill you.I think we should stop right now. Can I just say at this point, there's no point in recording this podcast. It's definitely true. Ai will kill us and that's all I have to say on the matter.
[00:03:52.240] - Nathan Wrigley
I'm going to love this chat because I think we've reversed our roles a bit. We're both generally positive, but I think you're probably more positive than I am. I tend to look for the negatives on this topic. I think we're entirely the other way around. I see it's a great tool and I think... We'll get into it anyway. Shall we see what chat GPT says? Because we started the first episode asking it. So I asked it, could artificial intelligence kill the web design industry? And I won't read it all out because it's.
[00:04:24.610] - David Waumsley
Pretty boring. No, I think it's actually worth reading out because when I read it, I thought the nuance in there was quite remarkable. So yeah, do read it out and I'll put it in the show notes.
[00:04:35.720] - Nathan Wrigley
Okay. It's unlikely that artificial intelligence will completely kill the web design industry while AI and automation tools may...
[00:04:49.940] - David Waumsley
Gosh, should I read it for you? Here we go. It is unlikely that artificial intelligence, AI, will completely kill the web design industry. While AI and automation tools may be able to handle certain aspects of web design. It is unlikely that they will be able to fully replace the creativity and problem solving skills of human web designers. Web design involves more than just technical skills. It also requires an understanding of user experience, branding, and visual design principles. These are areas where AI may not be able to fully replace human designers. Do you notice how the AI is couching the language with words like not to completely kill the web design industry, may not be able to fully replace, and all of this stuff. In other words, it's basically saying, you know what, we're just giving you a bit of time.
[00:05:37.780] - Nathan Wrigley
That's all. Yeah. Actually, it goes on further than that as well. Yeah. there's a good example. I couldn't read that out because I find the way that it talks trips me up. It doesn't already sounds a little bit artificial to me. I don't know. Maybe it's just today. Maybe I can't read.
[00:05:57.200] - David Waumsley
Well, you know what? When I read that, I really did put some thought into whether or not I thought there were any AI trip wires in there, and I genuinely couldn't see them. When I read it, I thought a human could totally have written that. Okay, some caveats there. A human who was told to write something very informational, there's no creativity or anything, but there's no clunky use of grammar, there's correct punctuation, it's capitalised correctly, the order of the sentence flows perfectly well to my mind. In other words, I think it's a pretty breathtakingly good job, given that you just asked a totally random question. And five years ago, if you just said that any of this stuff was on the horizon, the answer would have absolutely been no. Nobody saw this stuff coming. And yet now we're trying to figure out, did a human write that or did an AI write that? It's almost like it's passed the Turing test in many respects. It's been able to convince us that it's real. And the answer seems perfectly legitimate. Everything in there seems like the right thing to say. It's totally on message.
[00:07:06.230] - Nathan Wrigley
I've been working with it quite a bit to try and make some blog posts for clients. And the real skill in using this is using prompts to guide it to do other things. And even with this one, I tried it, but it didn't succeed that much. I said, Can you give me this in the style of Shakespeare? And it said, No fair sir, it is unlikely that artificial intelligence shall completely slay the web design industry. And then it went on and carried on just repeating what it said before without the Shakespeare.
[00:07:37.640] - David Waumsley
But you wrote.
[00:07:38.160] - Nathan Wrigley
It, it must be tired.
[00:07:41.110] - David Waumsley
Yeah, it is quite funny and you can play with it and it can tell you jokes which are quite funny. So it is truly impressive, no doubt about that. And I think maybe we will all agree, I think that at the moment, AI is disrupting web design.
[00:07:57.780] - Nathan Wrigley
In some form or another. Yes. And I think at the moment, it's almost like it's comedy a little bit. It's a little bit engaging just because it's new and interesting. So in the same way that anything that's new gathers some interest. I don't know, if a new brand of chocolate comes out, you want to try it because you've never had it before. Your favorite restaurant adds some things in on the menu. You want to give that a go because you've not done it before. And it's a bit like that. So it's interesting. But also when you pour your thought into it, it's also fairly alarming because it genuinely can do things that are a complete replacement for a human being. In other words, that text that you just wrote, obviously, neither of us was going to write that down. So it's not like we've replaced a human. But you can see that humans are going to be replaced in certainly copywriting in this case with AI generated content simply because it is quicker, it's cheaper. You can have a version of the text written in half of one second and then you can read through it and ask it to have another go at Shakespeare if you like, and another half a second later, you can have a second try, and so on and so forth.
[00:09:18.880] - David Waumsley
And the same is true of images. And I just can't see a scenario where people aren't going to be tempted en masse to say, I'm not paying a human, I'm going to get an AI to do it. So for images, for text, and increasingly, although we haven't seen any real evidence of it yet, I would imagine the whole process of building a website will come under the purview of an AI. So the design, the layout, the branding, the choice of the colors, everything, including where things pop up and how should the forms be structured, and SEO and all of that will just be a line item on a tick box on a form for a web design company. And you go to AI website. Com and say, I need a plumber's website and I'm going to tick these boxes. I'll pay a little bit extra to have a second round of SEO done. I'll pay a little bit extra to get some bespoke AI images done. And three seconds later, you'll have a website with text, images, various different pages, forms. And then you'll say, Actually, no, I don't really like that one. Give me 20 other alternatives.
[00:10:24.490] - David Waumsley
Press a button. Five seconds later, you got 20 more to look at. Pick the best one. Can we iterate on that one? 20 seconds later, you've got another hundred to choose from. And where do we stand then?
[00:10:39.600] - Nathan Wrigley
Well, then we end up with choice overload, which can be a problem itself. You know, interesting... Let's talk about that. I found this recently. It's produced me so much stuff because I have jumped on this and thought, Wow, this is really exciting. And this is a bit of a window of opportunity, I think, for me to produce content for clients because that's what a couple of clients are looking after are missing, really. They really need that for their SEO. And I thought this is a great way of being able to do that. Maybe we should talk about that first because it's not strictly web design, but copy and SEO using AI for that. So chat GPT as it stands alone on it, it can pretty much do all that you need. And there are, of course, other tools for doing that as well. There's a problem, isn't there? First off with that, because it is a violation to use AI at the moment in terms of Google's rules. It's spam, plus the spam.
[00:11:35.470] - David Waumsley
I guess the issue there is, so my understanding is that Google have said we would really rather that people didn't go for AI based content. Have I got that right? There was some proclamation somewhere saying that if everything goes to AI, then we'll be in a bit of trouble. So how would it know, though? How would it know that this was created by AI? I was trying to think of examples, and the only thing I could think of was the rate at which fresh content came out. So let's say, for example, that on your website, you're producing 500 pieces of 10,000 word paces of content, you're producing 500 of those a week. Clearly, something weird is going on there unless you've got a massive company and the endeavours, you're like the BBC or something like that. So that's maybe a trigger. But reading the individual pieces of content, how would it know? How would it be able to judge that's AI and that's not? So although there seems to be this rule in place, I don't know, maybe Google has got some incredible engine that can spot this stuff, but I'm really struggling to imagine how they're going to do that.
[00:12:48.840] - David Waumsley
Then, of course, you get into this cycle of you create content in the hope that the search engine will spot it, or rather pick it up and not spot it as AI content. That will then make the Google algorithm push you to the top. So you create more of that, you go further to the top, and it's just this cycle of AI creating content for Google's AI to read and index so that you create more. And you can see there's this cyclical pattern going on, and it's just more and more and more.
[00:13:20.740] - Nathan Wrigley
I've watched a few videos from SEO people, and one of them mentioned something, but I've seen nothing on this. But I'm certain because they are very serious about their SEO. But apparently there is to be in chat GPT some watermarketing that will be available to Google to be able to pick out whether this is unchanged content. So there may be, I don't know how they would do that. It's obviously the formula for their words. So I think they will be working. I mean, at the end of the day, I don't know if a trillion dollar company like Google are going to be too worried about this. If they are, chances are they going to try and buy it, aren't they?
[00:14:02.580] - David Waumsley
Yeah. Well, I guess the other thing to be said about that is that there are open source. So stable diffusion, which is by Open AI, is an open source variant. So I think chat GPT might in fact be based upon that, showing my ignorance here, what would stop somebody forking the software to strip out the watermarking? And wouldn't everybody then immediately flock over to the non watermarked variant of the AI because it wasn't it wasn't getting flagged. I would imagine that it's going to be... In the same way that you have black cat SEO, this seems like some race condition going on here where the people creating the AI are going to be trying to hoodwink Google and Google... It'll just be this towing and throwing, trying to beat each other. And I don't know who would win, but it does seem at the minute the output of this stuff, if it's just pure text copied and pasted onto a website, I can't imagine how in the future it'll know.
[00:15:03.780] - Nathan Wrigley
You're right. I think it is this always working out what Google knows and chancing you. The same goes on another thing, which is cluster spam. And I think this is a more recent update when they've been more categorical about buying links, but particularly also offering guest posting. That's also a violation. And I bet you get, I do, just for my rubbish blogs, emails almost daily for someone offering to put a guest post on my blog. Obviously, what they know is that SEOs know that that works at the moment and Google haven't got a chance of being able to spot it. And I guess you'll always be able to outsmart them with this. But they've not been very clear about what they mean in terms of using it. At the moment, it's just generic. You shouldn't use it for your content where most of us are now seeing, well, it's a tool to help us generate ideas to create some of our content, and it might help us with bits of it rather than seeing it as an article. We're not article spinning as used to be done in the 2000s.
[00:16:11.900] - David Waumsley
I think it's such a compelling thing to do as well. So on the one hand, I'm a bit frightened by AI. But also, if you imagine that you need to... Let's say that you've got a client, you were mentioning a client that you've got at the minute that requires content based around solar panels. Now, I could literally write on a postage stamp, on the back of a postage stamp, what I know about solar panels. So if I was to be given the job of writing some technical article or some SEO based content about solar panels, I'm going to have to go and do several hours research. It's tedium, essentially for me. Whereas if I go to the AI and say, can I have a 10,000 word article about the, I don't know, some aspect of solar panels, it is going to do it for me. And that's the other irresistible bridge is that it's got the world's knowledge because it's scraped the internet at its fingertips. So maybe at the minute it's not very good at making us laugh, maybe it's not very good at being innovative, but it's very good at knowing things and writing knowledgeable things in the correct order.
[00:17:16.540] - David Waumsley
And I feel for SEO, mostly that's probably what people want. So I think it is going to be more and more compelling. I think people will use it more and more. And I do think that copywriter's jobs are going to be much more difficult to maintain in the future.
[00:17:33.920] - Nathan Wrigley
I think that's true. I mean, I used it and it's been successful actually in a short time. But in this case, it was used as a tool because the person behind it gave me what they thought was an article, but it was insufficient in the words or just the bullet points and they were able to use Chat GPT to be able to expand on it. And for me, for my purposes, just to rewrite it, it helped me answer some questions in a way that would be much quicker than trying to Google it. I think this is where Google is on the threat because of the fact that you ask it a question. Although there's a question over its reliability, and I've seen this over and over mistakes it's made. And also sometimes it tries to use the words, so it edges its bets. If you're to say the pros and cons of something, it tries to... And it can be misleading in the way that it does that. And then when you question it, which I have done, saying, you're in play with this, this is this, and then it goes, apologies, I'm just a learning language machine, blah, blah, blah, and tells you that, yes, please ignore what I said.
[00:18:36.760] - David Waumsley
But don't you think in a way that really what we're dealing with is like a primary school age child? We're literally talking to a three year old, and the expectation at this point that it's going to be getting things correct is unrealistic. I mean, it's remarkable, right? It's truly amazing. But it's only been in inverted commas, I'm going to say it alive for a small amount of time. But imagine in 10 years or 15 years or whatever it may be, the rate at which it's improving, I would imagine those mistakes will be far, far less. So that I think is simply a product of time and resources and more training in the model. But I don't see that being a problem in a few years' time. If you ask it to do the solar panel article, I think it'll get the fact of it pretty much straightened out on the first go. I could be wrong, but that's my feeling of the way it's going. Can I just ask a question as well about the work that you just did on the solar panels? Imagine you'd have that exact same conversation. So guy's got solar panels.
[00:19:44.540] - David Waumsley
That technology has been around for decades, so we can assume that you could have had a solar panel conversation 10 years ago. What would you have said 10 years ago if you said, Can you write me some content? Where would you have gone for that? Would you have written it yourself and spent the time reading, or would you have farmed it out to somebody? What I'm getting at, what I'm trying to crowbar into the conversation is, have you, in inverted commas, deprived a copywriter of work by going to the AI? Whereas 10 years ago, you would have gone and employed somebody to do it.
[00:20:14.770] - Nathan Wrigley
That's a good question. I actually don't know. In this case, I think if it's 10 years ago, I'd have probably just taken what they've got bullet points and would have put it up and had the best. I wouldn't have tried to expand on this because I thought there was something that would make me do it quicker. But here's the interesting thing. There's a lot of people who jumped on and said this will kill Google because it's so much more efficient to just say ask the question and then ask for further clarification and it gives it. But I don't think it will. And I think this is key to why it won't take over. It's because we'll still want to go to Google because we'll want opinions, we'll want to buy directly from the person, we'll want the thing that they're big on since last year is their new EAT thing, their expertise, authoritiveness. I don't know why they don't just call it authority and trustworthiness.
[00:21:07.740] - Nathan Wrigley
Yeah, that's hard. I know. But anyway, that's the thing that they're into. And I don't think this is ever going to make up for this because it's going to... It's simulating all the information that's out there. But it isn't one person's opinion where you can see it, their credentials and understand their voice. And I think that's it. And it's the same way as doing this solar panel because most of it is functional, the information. I only need to pad this thing out, and I just need a few more facts and I can get it quickly through that. But that person led on what information they felt needed to be shared with their type of person. And without that's the human element. And I think that's right. And when I've just tried to think about doing this for a whole different industry, therapists, counselors, I suddenly quickly realized that GPT just cannot do that. You cannot speak like a human. And at the end of the day, their business is all about selling a service where they sit as a human being talking to another human being about their genuine emotions. And you just can't replicate.
[00:22:08.510] - Nathan Wrigley
I tried it. I tried to make some articles and they're just pretty hideous. Right.
[00:22:13.230] - David Waumsley
So it couldn't seem to emulate something which really truly requires a human. So in this case, therapy.
[00:22:21.140] - Nathan Wrigley
Yeah. And I think that's where it always fails. I mean, it's going to do some of the functional stuff really well and speed up how you can get there. But the intent and the individual creativeness or personality of people is I think our job as web designers, to some extent, is about bringing that human aspect to technology via the web and making that connection with people. And I think for that reason only, it will only ever be a tool that may simulate of foolers at certain points. But I think we will always adjust to that and spot what is fake and see what is real. So that's why I'm more positive.
[00:23:05.400] - David Waumsley
Yeah, it's interesting. I do know what you mean. I'm not sure that I'm entirely on board with that because my assumption is always going to be that clients... Okay, so I'm going to broaden this out a little bit. I think there's going to be two types of clients. I mean, this is ridiculous to just classify them as two, but there you go. On the one hand, there's going to be the client that just wants to do everything as quickly as possible for the smallest possible budget and be as effective as they can be with the money in their pocket. So in that scenario, I'm imagining somebody coming to you and saying, I want a website, I've got this much money. Can we automate as much as possible to reduce the cost? So automated images, automated text, so long as it works and it gets me out there, that's fine. And I think that will be a possible majority of the client base that we'll be looking for in the future. And on the other hand, there'll be the people who want the artisanal stuff. They want the... Instead of having a Casio watch that cost £14, they want the £50,000 handmade timepiece that you're the only one in the world that's got it.
[00:24:15.710] - David Waumsley
And I think that will also continue to exist. And people will be saying, Who did your website? Did you get the AI to do it? Oh, no, no. I had a real human do it. So I see it bifurcating in that way. I'm not making a prediction about that, but I could just see that being a possible flow.
[00:24:36.420] - Nathan Wrigley
I'm going to counter that with, just when you were talking, it reminded me of some of the conversations I've heard about the people who've been testing it for coding purposes. So a lot of people are using Copilot at the moment, aren't they? And that's a way that I think it's a good name, I think, for the product. It's the Microsoft, isn't it? Github thing that allows you to code and it's using AI. And I think that's a great name for it. That's how I see AI working as that tool to help you. But I think what all the coders have done, and there's been various people testing things with serious code and then just things like CSS as well. And I think what they all learn from it is that it'll give you functional code that you can use, but it will probably be either more complex. I know what you're going to say. You're going to say it'll just get better at judging these things. But the thing is, I think what they know is that you need to know how it's put that code together if you're going to maintain it and change it all the time.
[00:25:42.220] - Nathan Wrigley
And I think this is the thing why it always needs... It's just a tool for them. They still need to understand how it's working and what the strategy is and what you're aiming to do with it. And it will never come up with that. And it won't know the real life things that are going on. It might churn you out some code, which will speed things up. You'll probably need to check it out to make sure it's not more Voboose than it needs to be. But you're always going to need to bring the strategy with it, I think.
[00:26:11.400] - David Waumsley
Interesting. So actually, I'm going to counter that. So this is going to be an episode of countering things. I would say that in the same way that page builders allow people to create websites with absolutely no idea what's going on with the code, I would imagine that AI will provide the same thing. You won't say, Excuse me, you won't say, Okay, I would like six images in circles, and I would like them to follow me down the page or whatever. You're not going to be saying, And show me the code. You're just saying, Great, there's six images. They're too big. Make them smaller. Great. That's all you need. You don't need the show me the code because I think the whole point of AI in this conversation is to remove the people, is to have it so that you don't need to go to a WordPress expert. You just like a Squarespace type thing. You just have a service. It's a website service. You don't want to know what the code is doing. Great. I mean, for people who want to use Copilot and all that's brilliant. But I'm imagining that the crowd of people who will be using AI to create the website, they won't want to look at the code.
[00:27:19.010] - David Waumsley
They've got no interest. It's just the output that matters. But also I imagine that the code, over time, as you said, will get better and better and better and leaner and leaner as the models get better as well.
[00:27:30.740] - Nathan Wrigley
I would say your counter with the page builders is part of why I think... Because spending all this time jumping on board with page builders, now spending some time with clients who start to do it themselves a little bit and learning a little bit more about it. I see there is the problem for AI. That is what I now see with page builders, because people do start with the page builders, and unless they stay within a template that's been optimized for accessibility, which the rules change on all of the time, and they know about performance, which has needed to improve recently, and they know about mobile responsiveness, and if they make any changes to the content they add, they're probably going to be fine. Most of the time, though, they're using it, think they're doing a good job and have no idea of all the stuff that's being automated, and they have no idea the problems. And I think that's... I see AI being exactly the same in the wrong hands. They won't see the problems they are making with it.
[00:28:35.900] - David Waumsley
Yeah. The difficult thing about this conversation is we are really just guessing, aren't we? And I'd see everything you've just said, it makes perfect sense. I just wonder if the whole point of the AI, though, is to reduce or possibly remove the reliance on an expert. And you might be right. But let's say, for example, that Google put out some documentation saying, Okay, any site that's created with AI is making these mistakes. I imagine it won't be long before the people behind the AI will get rid of those mistakes and say, Okay, right, whatever it is, I don't know who this too much bloat, this divception, there's divs within divs within divs. The AI for the website is from now on, we're just going to strip all of that out and make it better. So yeah, I don't know. I'm not sure.
[00:29:27.320] - Nathan Wrigley
But you're right. And I think that's always going to be our challenge. And in the same way that the people who want to DIY with a page builder and skippers entirely, and they will think they've had a success. But I'm not sure if they will ever succeed over. And the more people doing that, the more similar those things are going to be. And I think this is the thing. These things are always going to be language models which are going to bring to us in a much more efficient way the knowledge that already exists in the world. But they are not, I don't think, on their own, going to be creative enough to bring new knowledge into the world because it's us as living organisms who, if you like, act and behave in strange ways that need adjusting to and now problems are something that I think just comes out of being organisms. And they really only provide us with logic. I don't fear for it really for creativity because I think, although nothing to prove it, that it would be almost arrogant to think that we could be simulated entirely through a series of ones and zeros.
[00:30:34.210] - Nathan Wrigley
Presently, the computer model for understanding the world is the one because we've invented computers and it comes closer enough to anything else. But when it comes to human beings, for me, human consciousness, we can explain everything that allows us to see, think, hear, and everything, but we just cannot explain human consciousness, why it's there, why it disappears, the experience of it. We're still stuck with Des cartes' original dualism between mind and body. I think when it comes to AI, it's the same thing because at the end of the day, all we've got is this computer model which can simulate certain stuff, and it's very impressive. But I'm sure creativity comes out of something which organisms have. This is a.
[00:31:21.280] - David Waumsley
Really interesting... No, that's brilliant because I love all this stuff. You're right. At the minute, AI is based upon consuming information and regurgitating it. Essentially, what's the next best word? Or if it's creating pictures, what's the next best pixel? And it just radiates out from the beginning point. But the endeavor is to create what's called general artificial intelligence. And basically what that means is, and I probably got the term wrong there, I think there might be another word in there somewhere, but certainly general is in there. And that is when artificial intelligence can replicate basically all the functions of a human being. And that is the goal. So where we've got at the moment is nowhere near that. We've got a model which works and is interesting and puts out the stuff that we've got at the moment, and it's remarkable. But the endeavor is to create an intelligence that can train itself. So with almost no sets of guidelines, it figures stuff out. And in a sense, it's indistinguishable from a human. If you put it somewhere absolutely new, it'll start to figure things out and it will learn. And that is the goal.
[00:32:36.420] - David Waumsley
And it's called the hard problem of consciousness. And whether it's, they call it substrate dependent, whether or not it's literally bound to our brain, or whether it's some quality of atoms or whatever, we don't know. It's the hard problem. I'm not sure we'll ever figure that out. But if we do manage to get general artificial intelligence, can you imagine it? It will become incredibly good at things incredibly quickly. There's this lovely AI which was designed to play Alpha Go, I believe it's called, which is a really difficult game. It involves black and white tiles and you put them on the board and allegedly it's one of the most difficult games to create. So they created this artificial intelligence to play Go and within a very short space of time, it took three days for it to become better than any human had ever been. And then they showed it chess and it had never played chess before. And within moments, it... Well, moments, I don't know how long, but in a certain period of time, it could beat all of the AIs designed to play chess. So it really was an exponential rise. And I think this is the thing which I can't somehow sum it up when I talk about it.
[00:33:57.120] - David Waumsley
It's the exponential curve of it which has me worried is the fact that it will do things at such a prodigious rate that once the cat's out the bag, if you know what I mean, we won't be able to get it back in. Anyway, sorry, I went totally off the rails there.
[00:34:12.560] - Nathan Wrigley
No, I found it fascinating. And I agree with you. Almost everything you said there, I think, because it is the rise and how quick it is. And it does scare me the idea of robots acting on the world with their own autonomy, not being enough safeguards, being in place. I know we've had other chats where that's one of your concerns, and there's plenty of stuff online about that as well. I share all of that. I guess at the end of the day, it's the arrogance, I guess, of thinking that we might have sussed it. I find it interesting when I did studying of psychology and stuff, particularly brain and behavior, as I mentioned, this daycaut dualism, which we're trying to do. People who like to study the brain as a way of understanding human beings don't really get very far with this one. But it's interesting to see the history. Once, everything was explained because the technology of the day was telephone, so everything was a switchboard in the brain. Everything was connected. Then we got the computer analogy and suddenly everything could be explained in that way. And that's all I think these things are is analogies to understand it.
[00:35:28.100] - Nathan Wrigley
So I just think we're never there because there is something more about... And at the end of the day, unless we're all completely wiped out, of course, by these robots which have actually removed us all and they are in charge and they judge in. But the interesting thing is they make a nice picture, but they're not the ones who are going to be able to tell is that a nice picture? Well, I guess at.
[00:35:53.700] - David Waumsley
The point where you've got general AI, and I'm sure there's another word in there I keep forgetting anyway, this general purpose AI. I think at that point, you are straying into the territory of, yes, it can figure out that that's a nice picture. It will be able to give an opinion on, well, I like that because, and at the minute, we believe that that is almost the basis of being a human or alive. And once it does start to do that, whether it's contrived or not, we won't be able to tell. There's no way of staring into the AI and saying, No, you're lying to me. It will just come back with another lie. But I think that's the goal. Anyway, we've really strayed off the whole WordPress thing, but it's.
[00:36:44.950] - Nathan Wrigley
Fascinating. yeah. But I think how we view it is quite important, I guess, whether we see it as a tool or threat. It's definitely going to disrupt us. So we have to make some decisions, I think, about how we use it in our businesses and how we sell it to clients.
[00:37:00.200] - David Waumsley
Yeah. And I think that if you actually dissect the bits and pieces of a website, it is a jigsaw puzzle, isn't it? And because it's a jigsaw puzzle, I can see AI being able to get a foothold in it. So for example, a typical website, it's got menus at the top. Well, okay, tell the AI most websites have a menu, so offer a menu. Should it be left or right? Well, it's a binary choice. Then what comes next? Well, it's a homepage. It's probably got some hero and call to action. Can we put some text in there? Yes, of course we can. We can do that. And then what comes underneath that? And what comes underneath that? And because it's this jigsaw puzzle, which we believe is very clever, we're very good at selling our skills and designing courses for effective websites. I don't see anything in there that's not in the purview of the AI. I think give it enough examples of that, and it will be able to give you really good examples back, but it will be able to do them instantly and abundantly. And I think at that point, it's going to be very, very difficult to go to clients and say, Look, can you pay me a living wage, or go over there and spend $20?
[00:38:16.560] - David Waumsley
And the whole thing's finished. So my analogy would be, I guess, something like the robot in the car factory. If you'd have gone back 50 years ago and said to the car manufacturers, Okay, so you've got 100 people down on that production line. It's costing you several million a year. Would you like me to build you a robot that will mean that you have to employ five people and get the work done in a fifth of the time? It's like, Well, you know what? Now I'm going to keep the humans on. That's what I want. I think that's such a compelling... It's lean, it's going to be cheap, it's going to be affordable, it's going to be fast. It's everything that our economy tells us needs to be done. It's cheaper, it's faster, it's more efficient, it's abundant, all of that. It's going to be hard.
[00:39:09.780] - Nathan Wrigley
I completely agree with you. But here's how I would look at this. I don't need anything that makes building websites to me seems like a real joy because if that's the case, then there's probably still going to be some money for stuff that we never get the time to do, which I think is in essence, what is required for all websites. And that is a user experience design lead process where you actually test whether this is going to be working on the people it's designed for and measure all of that stuff. So yeah, the robot creates the thing. I test whether it's actually working in the desired way on the real humans who are using it.
[00:39:49.020] - David Waumsley
You see, I would say that's the job of an AI because I would imagine that if you build into the software the components to measure that stuff in real time. So let's say, for example, you give the website, the AI, the tool that's creating the website, the ability on the fly to change the colours of the buttons based upon the actions that it sees in real time. In other words, it's split testing all the time, but it's deciding in real time, Okay, I've now got enough data, I'm swapping the buttons out. Or that text will show 15 variations over here and 15 variations over here or whatever, and then we'll just immediately decide. So I think if the data can be analyzed by a human, it can be analyzed by an AI, and it can swap it in real time. So why would you need a human to go back in and change the buttons? Because the AI just did it seven weeks ago without telling anybody, and it's proven to be effective and we got the data to back it up.
[00:40:48.850] - Nathan Wrigley
No, I completely agree with that as well. But I guess what I'm thinking is that it's these things where you can't because no one knows the outcome. So let's say we're looking at something up, an installed clarity by Microsoft. It's free. I did this quite a bit on clients and I'm looking and I'm thinking, why are people not scrolling down to the bottom where this is really interesting, the call to actions down the bottom on that page. Why are they not going down the bottom? Now, there's a lot of things I've got to think about about how humans might behave, how they're going to react to that. There is something I think I need to understand about the humans and their particular type of human to be able to make that. So certainly, I agree. I mean, that's already automated things like A B split testing and stuff, definitely. But I think there is a level of interpretation where you need to understand humans. And I think humans are changing all the time. So the language machine is going to know up to a certain point. And it's language is the interesting thing. Back to psychology again, we really don't understand how we learn language.
[00:41:53.550] - Nathan Wrigley
We proved the behaviorists were wrong with its inputs and outputs when people picked up or use grammar in different ways, but still, we don't understand how language is and we don't understand why we keep changing our language constantly between us, agreeing this strange communication between us. We change language, we change the meaning of it and our subtle understanding of it. So I think there's always going to be this need to interpret that a machine will not do because it's always going to be fixed.
[00:42:25.280] - David Waumsley
At some point. I actually think that's the strongest argument, isn't it, for the AI not being suitable? I think if you can get that argument into the conversation, I think that's a bit of a slam dunk. At the end of the day, for now at least anyway, the audience is humans. When that changes, then we're screwed. But if the audience is humans and humans really do not function in anticipated ways all the time, do they? They have proclivities and each one of us has a different reaction to a website or whatever it may be. I like this song, but you don't. I like this color, but you don't. That font annoys me, but you love it. And I think if that argument can be made, which it certainly can, and I don't see any scenario where that argument won't be made in any time in the next few generations, then yes, that's the foothold, isn't it? That's the point at which you say, look, the AI is great, but it's not serving up things which humans are wanting, or at least it's not doing it all of the time. And a human needs to decide what a human is looking at.
[00:43:39.580] - David Waumsley
So yeah, I think that's a really strong argument.
[00:43:42.060] - Nathan Wrigley
I think AI would completely misunderstand you. A lot of the time, you do really excellent dry humour sometimes. It would literally miss it. You know? Because we.
[00:43:53.120] - David Waumsley
All know. Sarcasm and AI are probably enemies, really. Because if you say, Oh, that's really good. The AI think, What? Okay, that must be really good. No, it's not really good. Yeah, sarcasm and AI. I imagine. Yes, you're right. Good point.
[00:44:09.950] - Nathan Wrigley
And I think it's circumstantial and I think it keeps moving and I just don't think it will ever keep up. If it's there to serve us, I think it's going to be a great tool that's going to replace a lot of jobs, but I just think we'll still need to do something. We'll still be employed to do stuff. So yes, the certain industries are going to go. But I still think web design, generally, there's so much in web design that I would love to do, but the clients don't have the budget for it. So I'd love more automation so we could do some of the stuff that really is connected to the humans through the technology.
[00:44:40.770] - David Waumsley
This sounds very strikingly like the argument that technology will give us more leisure time. The idea back in the 60s that if we can automate all the things, humans will have loads more time. Where did that go? Who's got more time now than they had 100 years ago? It just isn't the case. We'll just end up inspecting the AI and being baffled by it, and maybe our jobs will be settled to keeping the AI within its wheelhouse and not straying outside. Good grief. The only real thing we can say is that we should revisit this exact subject in a few years' time and see how horrendously wrong we were.
[00:45:23.790] - Nathan Wrigley
Yeah, I know. Well, I don't think we can, but it's a great chat. I really enjoyed this. I bet everybody listening to this really hates it.
[00:45:29.950] - David Waumsley
Yeah, what's it got to do with WordPress now?
[00:45:32.040] - Nathan Wrigley
Yeah, I know.
[00:45:32.980] - David Waumsley
But very interesting and not typical for us. We've strayed into all sorts of areas, but do you think we're done with that? We are. I think it won't kill us, but it'll certainly make our lives more difficult and easy at the same time there. How's that for sitting on the fence?
[00:45:50.620] - Nathan Wrigley
I'll go with that.
[00:45:51.350] - David Waumsley
All right, what's coming up next?
[00:45:53.860] - Nathan Wrigley
Well, okay, dangerous one now, but I think we can do this. Is WordPress too woke?
[00:46:00.510] - David Waumsley
Oh, Lord.
[00:46:02.080] - Nathan Wrigley
I know. I've got views on.
[00:46:04.720] - David Waumsley
This one. Okay. Yeah. I mean, I need to go and do some research so that I fully understand all of the vocabulary surrounding this, but we'll be back in a couple of weeks. And unless both of us have chickened out, that's the subject that we'll be doing. So, yeah, all right. I'll see you in a couple of weeks. That was lovely.
[00:46:20.200] - Nathan Wrigley
Yeah, lovely. Thank you.
[00:46:21.680] - Nathan Wrigley
Bye. Well, I hope you enjoyed that. Certainly an interesting episode chatting with David and I. It's really off the scale for us. We normally talk about WordPress in particular, but this is all about the things that may happen because of AI. Maybe you disagree, maybe you think we were over emphasizing it. Maybe we've got things wrong. Give us a comment. Head over to WP Builds. Com. Search for Episode 315, and we'd love to receive your comments there. Alternatively, WP Builds. Com forward slash Facebook, that's going to get you to our Facebook group. You could post a comment there by finding the piece as well.
[00:46:58.360] - Nathan Wrigley
The WP Builds podcast is brought to you today by Go Daddy Pro. Go Daddy Pro, the home of managed WordPress hosting that includes free domain, SSL and 24 7 support. Bundle that with The Hub by Go Daddy Pro to unlock more free benefits, to manage multiple sites in one place, invoice clients, and get 30 % off new purchases. Find out more by heading to go.me/wpbuilds. And again, very honest, sincere thanks to Go Daddy Pro for their continuing support of the WP Builds podcast.
[00:47:35.450] - Nathan Wrigley
Okay, as I said at the top of the show, don't forget, the Page Builder Summit is happening next week, 20th to the 24th of February 2023.Head to pagebuildersummit. Com to sign up. You can see the list of speakers and all of the things that are going on that week. We'd love to have you join us. Page Builder Summit. Com. Okay, we will be back, of course, next week. We'll have a podcast. This time it will be an interview. I hope to see you there. Do stay safe. Do join us for the Page Builder Summit. Have a nice week. Cheesy music fading in. Bye bye for now.
