Google is out there to make money. So Google is obviously going to make you ads that are going to make it money rather than necessarily being on your side. Just because Google says your ads excellent doesn't actually mean your ads. Excellent.
There's a real danger that you see, you know, the kind of the green light or the score like seventy out of one hundred score. And you think, oh, I'm doing a good job. It's like, no, you're doing a job that the machine thinks is good. And that's not always the same thing as doing an actual good job.
The machine doesn't know what level your audience is at, you know.
And sometimes, like you say, it's messaging that you know, that the machine thinks is a bad idea that's going to get you a low quality score. But you know that when your audience see it, they're gonna think, oh, that's me. Do you think there are places where it is genuinely useful to start looking at AI?
I think it can help, like I mentioned, chatbots earlier, done well, actually some of these chatbots are quite good.
Hello and welcome back to Digital Marketing From The Coalface. David is on holiday, so me and Julie have decided to steal the podcast. Yeah.
We do what we like.
Yeah. The cat's away and all that. Um, and today we're going to be talking about artificial intelligence. Artificial intelligence in marketing specifically.
Yeah. It's a bit of a big topic.
Is.
Artificial intelligence generally.
Self-driving cars? No. Um, yeah. So I guess we're going to talk about artificial intelligence, um, how it's becoming kind of steadily more popular in marketing. Um, some of the weaknesses of it, why you probably shouldn't rely on it too much. And I guess also some of the places where it is actually really useful, right?
Yeah, yeah. It's just, um, discuss the pros and cons really. I think it has its place, but not all the places.
Yeah. I think that's exactly it isn't it? It's one of those things that I think is becoming kind of not trendy, not a gimmick per se, but everybody's talking about it. Everybody likes to.
Use it is talked about a lot. It's Increasing in sort of just availability, I guess.
Yeah. And I think a lot of people have been kind of pushed towards it. I'm thinking specifically of things like Google ads and stuff like that. It's steadily becoming more kind of, you know, machine learning or AI driven. Yeah. Um, and I think there is a kind of real risk, I guess, of as, as marketers, isn't there, of becoming kind of overreliant on it and depending on it too much.
Yeah. There's a fine line between using it to make things easier and using it in place of like actual intelligence.
Yeah. Especially with like going back to the Google ads example, right? It's like they push you constantly to kind of like allow their, you know, machine learning algorithm to make decisions about when you bid and how much you bid on clicks and, you know, kind of what's in your ads and stuff like that.
And it's the thing, it's, it's like you put a load of lines of text in and it makes up the ad and yeah, it, it does worry me.
It's a world away from like when we first started using Google ads, right? And it's like, you have to do everything yourself. You have to go.
To like the crafting of it. It's like writing a haiku in school when you had to write a haiku and you had to, like, had to be exactly the right number of like characters, syllables or whatever it was. It was a little challenge just trying to get it exactly right and to fit into the space and make sense. Whereas now it feels like you just throw a load of words together and Google decides what your ad is going to say, which kind of scares me because sometimes there's going to be times where your ad just doesn't make sense.
Yeah. And it's back to that old thing, isn't it? Like, well, this is exactly it. Like sometimes it's just not that good. Right. So it's like the intelligence bit of artificial intelligence lags behind a little bit. And it's like you're seeding all this. Yeah, seeding all this control to Google ads and you're saying, okay, so here are fifteen headlines. You arrange them how you see fit, and you learn from how people interact with them. And then you come up with the best combination. And it sounds really good on paper, but when you actually let it loose, what you see is your ad displaying with, you know, just, just, you know, gobbledygook as headlines that don't even make sense together.
And I mean, it goes further than that because if you have at the moment it's in the process of swapping between sort of normal text ads and changing them.
All the responsive search.
Ads, stopping running the extended search text ads. So what it's saying is you need to now create responsive ads. So we've started one for you and you go in and it's basically made it up. And sometimes sometimes it's actually not bad. Sometimes it's just not quite right.
Yeah. I mean this is the thing isn't it? I mean, yeah, for anyone listening, that's a really important point. Right? So the old school, if you're using Google ads, the old school kind of expanded text ads are being retired on the thirtieth thirtieth of June.
Well, you the if you've got them, you can still run them, but you won't be able to edit them anymore.
Yeah.
And you can't make any new ones.
You can't make new ones. And not being able to edit them.
Once you need, if you need to edit them at all, you're just gonna have to delete them.
Yeah.
Remove them, which doesn't seem to delete them.
And replace them. Yeah, exactly. But then yeah, you'll be replacing them with responsive search ads where you have very little control over the messaging and very little control over kind of what displays. And yeah, like you say, sometimes.
You can't control what displays, but then they mark you down for it?
Yeah.
It's all a little bit of a game where you want that circle to go around and say, excellent. If you don't use Google ads, this is going to make no sense at all. But basically it Google kind of rates how good it thinks your ad is. So you think your ad is great. And for some reason Google goes, that's absolutely not what we want at all. And you have to tweak it so that you get a kind of good or excellent score with a little and it's a little sort of circle that fills in and you're waiting for that circle to fill in by tweaking all the words and trying to hit the perfect score.
Yeah. And I think this is the thing, isn't it? I mean, this just goes back to what I was saying earlier about overreliance. I think a little bit of machine learning is fine. I think saying like, you know, if you put in four different headlines, we're going to try out some different variations of them and see what works. It's fine. Where it becomes problematic is where Google is saying, well, no, you know, you can't do anything. The machine is going to do it all and we're going to make up headlines for you. We're going to add headlines and new ads to your account without telling you, you know, the machine is going to decide when dynamic.
The dynamic ads pretty much do that. You just put your eye doesn't even fix its own URL.
It does.
It basically goes, we're going to make up some ads.
Yeah. And this is the thing, like, I think it, you know, I think it's probably fair to say that it works, what, maybe three or four times out of ten that it comes up with something good, maybe. Yeah.
And I think also it probably works better for a fairly standard e-commerce shop than the types of things we're usually doing, which are business to business and not, not as standardised by now. Special offer type, you know, type offers.
Well, it's really interesting that you should say that actually, because one of the other bits of AI that we've had quite a lot of experience with semi recently was do you remember, um, the Jasper or Jarvis as it used to be called AI Jarvis? Because unsurprisingly, Disney and Marvel weren't very pleased about them calling it Jarvis. Ah, right. Um, so yes.
Of course I have another. Yeah. We've got teddy bear at home called Jarvis. And the the next teddy bear is now called Friday, which if you're into Marvel apparently makes sense. I it means nothing to me.
But yeah, I'm sure somebody listening will be sorry.
Going way off topic there.
But we, you know, we, we started using this, um, AI writing platform and the idea is that it's trained on this, this algorithm called GTP three, which is like the, the newest kind of AI writing algorithm. And it's trained on thousands of articles across the web, and it reads them and digests them. And then the idea is when you start typing something, it looks at all the things it's digested and learned from, and it picks what you know, what your sentence should say. And it uses machine learning to do this. And again, you know, it's fantastic if you're writing product reviews for shoes or, you know, something, lots.
Of stuff out there that it can pull, pull on and kind of understand.
Yeah. But as soon as we started using it for like B2B stuff that was fairly complex and fairly technical, it was just completely clueless. You know, you'd, you'd ask it to type about, you know, kind of, um, oil wells and it would start talking about, you know, just, just, you know, wells in your garden or, you know.
Yeah, it just wasn't, there wasn't enough out there for, I guess, for it to learn from.
Right. It wasn't really intelligent. I think that's the problem, isn't it? Ultimately, is a lot of this.
Not making it up? It can only use what's input into it. Yeah. So it's like, was it um garbage in, garbage out.
Yeah. Well, yeah.
So yeah, if there's nothing in there for it to pull in, it can't put up the right output out the other end.
Yeah. And I think that's what we see with Google ads too, isn't it? Like you say when you, you know, when you're doing something very simple and very straightforward, if you're running a shoe shop, if you're running a, you know, a kind of, I don't know, an automotive retailer or whatever, it's probably fairly simple and fairly straightforward to let Google do all of that for you. But if you're selling something complex, if you're selling something that you know, where the name or the keywords you're targeting could be confused with keywords from a different industry.
Well that's right.
It's it all falls apart.
Basically have to sanity check these things.
Yeah. I mean, and that's yeah. And that's another really good point, isn't it? Because you do spend a lot of your time just going through and kind of holding its hand and kind of correcting its mistakes. And, you know, it's something we were talking about just before we started recording was that you've got to make this kind of clear distinction between artificial intelligence and automation. Right?
Yeah. I think because automation, you know, automation is great and automation really does make your life easier. I mean, you know, we obviously we use HubSpot and the automation, HubSpot is brilliant that, you know, if somebody is setting up an email, the, the workflows are great. You go in and you basically say, right, if somebody fills in that form, send them this email, notify this person, remind this person three days later to do this. And if somebody opens the next email, do this. If they don't open the email, do that. If they watch the video, do this. And it all works because you're telling them based on an actual fact. If they do a thing, do this next thing and all these things actually exist. You're not saying if they do this thing, go away and and invent something else.
Yeah. Go away and work out what they should be given.
So automation, you have total control over what happens. You know what the input is. And you have total control of the output. You're just not having to do it manually. You're just telling it to do it by itself.
Yeah.
But yeah, AI is an element of.
Giving the machine the control to go and make up those rules for you. And when you can't set the rules anymore, it does sometimes feel a little bit like it's just kind of randomly deciding what to do. I mean.
We've, we've all had these, you know, these experiences with chatbots.
Yes.
So, you know, you go to the bank and you say, I ask you a question and you answer a question and then it goes. So which of these things do you want to do? And you're like, I don't want to do any of these things. So you type something into the box and it just gives you the same question over and over again. And you can't get any further because it it doesn't allow for anything that isn't on the list. And it doesn't even even have a no, my thing isn't on this list option. Please let me speak to an actual person. Um, it just, it just prevents you from going any further. So those things, I think they give AI a bad name.
Yeah, absolutely. But I think a lot of that is just people kind of misusing it isn't it? Well, yeah, this is the thing isn't it? Because, I mean, anyone that's listening to this, that kind of works in marketing will probably have been sold or will have found themselves the kind of subject to marketing about AI. And it is often pitched as this fantastic solution. You know, it's going to come in and solve all your problems. You don't need to do anything. You can just let the machine.
There are people out there. I think that, you know, there are websites out there that have like massive blog sections that are entirely written by AI. And, um, sometimes you can tell, you know, but like you say, if it, if it's something fairly standard, like buying shoes or whatever, there probably are quite a lot of blog posts out there that are written by AI and probably are reasonable. We have tried it. Um, we haven't used any of the outputs where, where I have used it is for a certain topic going in and, and sort of typing the topic in seeing what it comes out with. And well, well, what's come out has been kind of fairly nonsensical. It's given me ideas about the sort of things that I need to go away and write about.
So it works by not working.
As a starting point, just to get the sort of creative juices flowing. It kind of works because it gives you ideas of what other people might be writing about and what's happening out there, and roughly what it thinks you should be covering. So while the sort of actual things it's writing don't really make sense, that what it's, it's helping you do your research, I guess about your blog post.
Mhm. Yeah. Yeah. I guess that's the thing, isn't it? Maybe if they were pitched as kind of research tools or whatever, it would be kind of, you know, more useful than what it's currently being pitched at. It's interesting because when we signed up to use Jarvis, Jasper, whatever, I joined a Facebook group that I'm still in that's supposed to teach you how to use it very well. And occasionally when I'm like scrolling through Facebook, I'll see posts on my feed being like, you know, I've been trying, I've paid hundreds of pounds to use this, and I'm trying desperately to make it work for me. And it's just not working. And you'll see in all the comments, all the, all the people that love it and use it all the time saying, well, you just have to babysit it. You just have to go in and, and edit everything it says. You just have to, you know?
Yeah, that's kind of what they're saying, isn't it? You have to edit everything it says because, um, yeah, it's not doing the job.
What it all boils down to is that it's not really kind of actually very helpful in its, in its current incarnation.
But it's not doing the job of a person and it's nowhere close to doing the job of a person, I would say.
But I can see the lure. I can see why people, uh, you know, kind of.
Writing blog posts is hard. And it takes time. And yeah, it's and, you know, even if you're really good at it, it takes time. But if you're not very good at it, it's, you know, people do find it really difficult.
An uphill battle.
Blank piece of paper. So anything that is taking you from a blank piece of paper to a piece of paper with some words on, even if those words aren't very good, I guess is a starting point.
Yeah. No, absolutely. Absolutely. It's a really interesting point actually, isn't it? It's like, it's really easy to kind of sit there and say, well, these things just don't work. But I guess for some people out there, maybe it is useful. I mean, maybe even with AdWords, you know, if they're not going to sit and come up with fifteen headlines, yeah, maybe there is something to be said for just letting the machine do it. But I think, yeah, I guess the point is really, it's not the ideal way to do things, is it?
No. And there's a danger that everything out there is sort of not very good. And you're I, you know, Google is out there to make money. So Google is obviously going to make you ads that are going to make it money rather than necessarily being on your side. So I think you have to make sure that, yeah, you're, you're doing a, you, a person's going to be able to do a better job regardless.
Yeah.
Just because Google says your ads excellent doesn't actually mean your ads are excellent. I think you have to go and make sure that it actually works for your business, even if it's not getting an excellent score. I tend to go for goods because.
Yeah.
It makes more sense than going for the excellence that actually don't necessarily make that much sense.
Yeah, I think that's a really good point. Actually. There's a big move to kind of put, you know, push this idea that that the machines know better than you. And one of the other platforms I see recommended all the time. I don't know if you've ever come across. It's called surfer SEO, and it's like a browser based program that you drop your copy in and it tells you how good your copy is for SEO and for readability and all of those things. But again, you know, it's using a very basic machine algorithm to do.
A few of them do that. Yeah, they read it and they give you scores. I mean, WordPress has a lot of plugins for that and things like that. Well, and it says, you know, your sentences are too long or whatever. You're like, this is the sentence, you know, this is the only way to explain this. I can't make the sentence shorter because it needs all these words in it.
Yeah. And there's a real danger that you see, you know, the kind of the green light or the score like seventy out of one hundred score. And you think, oh, I'm doing a good job. It's like, no, you're doing a job that the machine thinks is good. And that's not always the same thing as doing an actually.
Good job because the machine doesn't know what you're, what level your audience is at, you know? Yes. You know, you know, make it easy enough for a ten year old to read it or whatever. But some in certain cases, your audience doesn't want to be spoken to like a ten year old.
And sometimes, like you say, it's messaging that you know, that the machine thinks is a bad idea. That's going to get you a low quality score. But you know that when your audience see it, they're going to think, oh, that's me. Like that appeals to me. I want to click that.
Yes. Um, yeah. And certain clients just, you know, what they're doing is quite complex. And, and it needs to be explained using certain words. And just because the machine doesn't like those words doesn't mean they're wrong.
So yeah, I guess what we're saying is sometimes it can be a little bit helpful to use these tools, but relying on them is, is a very dangerous thing. You have to be able to think for yourself.
Yeah.
You've got to be able to instincts.
There's, I guess there's two things. There's the inputting in the first place. So giving them the machine, what it has to work on, which has to be sensible. And then when the machine sticks sort of something out the other end, you then have to check it and edit it and make it into human.
Yeah.
Sort of able for human consumption.
Actually suitable for your audience. And I think there's an interesting point that you made a little bit earlier as well. Just to go back to that is the idea that they're kind of they're becoming more and more prevalent, right? They're kind of cropping up all over the place now. And like you, you gave the example of like Google and Facebook and things like that. And one of the things I think from a kind of strategic top level perspective, when we're, you know, doing a marketing campaign, we want the messaging to kind of sync up and we want some kind of synchronicity between all the different platforms.
Yeah. We spoke about this the other day, didn't we?
Yeah. And one of the things I think that really kind of concerns me is that, you know, if you, if we all become more and more reliant on these kind of machine learning algorithms and we let these kind of artificial intelligences take over, there's going to be no way of doing that anymore. Right. We're not going to be able to say, well, you know, this is my headline.
Because ideally, I mean, in some sort of campaign, you would have, you know, maybe people don't really know about the thing that you do. So you would put it on Facebook first or LinkedIn to target a specific audience who aren't actively looking for it. Put it in front of them so that they, they've heard of it. They kind of understand it. And then assuming they maybe think about it, go away, and then a few days later come back and Google it.
Yeah. You want Google to say yeah.
Like the thing that they're looking for. And if Google's making it up and Google doesn't know what you're putting in Facebook or you don't, you don't even know what Facebook is churning out. And then you don't know what Google's going to churn out. You can't make them match. So you can't do. Yeah. The sort of integrated campaign thing doesn't really work. Every machine is off doing its own thing.
Yeah. And it's interesting too, because sometimes on Google, especially where you let their AI take over completely and create dynamic search ads, you can't even see what the content of the, you know, no idea scares me.
I'm like, no, I can't deal with this because I can't not know what your ads are going to look like. I mean, I can't see how that could be a good idea.
Especially not when you're spending thousands of pounds, like you say, on like a demand generation campaign where you're like, this is our software. This is what it does. And then you're basically saying to one of your advertising platforms, okay, you go away and you make up, you decide what we say that it does.
Imagine you're spending millions on like a TV advertising campaign saying, you know, with very specific messaging in it. And then you, you can't.
Yeah.
Make sure that when people go straight to Google after seeing your ad on telly and look for it, they might, they might not find it because Google's sort of made it up and.
Just decided that that's not yeah, that's the wrong language to use. We'll use this language instead. And yeah, no.
It's yeah.
It's a nightmare I think isn't it. I think it's fair to say. Yeah. So then do you think there are places where it is genuinely useful to start looking at AI?
I think it can help. I mean, I see like I mentioned chatbots earlier done well, actually some of these chatbots are quite good. And it can take you to a point where it can answer your question and, and then put you through to an actual person.
Yeah.
If it can't answer your question, but I think it has to have that option. I don't, I just don't think it can totally take the place of a person because, you know, human beings just don't fit into boxes.
Boxes or like maybe they do occasionally, but the chances of there's.
Always.
And I think the negative.
Plenty that just don't fit into that, that thing that you think they were going to do.
Yeah. And I think the negative, like that situation you were describing before, where you end up on a like for me, it's always energy company websites. Whenever I'm trying to use the chat bots on energy company websites and it's like, you end up in that, that trap of like, you know, I'm trying to talk to somebody about a specific issue and it's like, do you want to pay your bill? Do you want to speak? I know, and it's like, no, just like, do you know what I mean? If I wanted.
To pay my bill, I'd probably have managed that. You know, it does seem to almost seems to give you options of things that you could have done without going onto the chat. But absolutely, the reason you're on the chat is because you want to do something that isn't possible to do on the website, and you want to do something else. And it, it can't pick that up and it can't work out from what you have typed in the box, what it is you want to do next because it's, it's not complex enough yet. Yeah, I think it will get better and I have seen some that have done quite well. But um, yeah, I think their role is really to take out the really tedious sort of um processes and yeah, involve people in the things that people are good at, which is the things that are a bit more unusual.
Because I think the negative of having those experiences, like if you, uh, sat here thinking, oh, well, you know, I've implemented an AI chatbot on my website and it saves me hours every day of replying to people's messages. I think the point is the negatives of somebody having a bad experience with a chatbot and getting trapped in some kind of weird loop is so bad that I think personally, it's just not worth the risk of of kind of doing that. Do you know what I mean? Like if you're going to use them, you have to be using them as a kind of pre-screening tool almost.
Absolutely. That's, that's exactly it. It, it can do a little bit and it can point people in the right direction so that if they do want to do something standard, it can show them where to go.
But as soon as you run against.
Against.
An issue.
A person that you can't sort of just take the people out of it completely. It just doesn't work.
Yeah. I'm trying to think of any other instances where artificial intelligence is becoming more useful in marketing, but it's a struggle.
Yeah. Um, I'm not sure that there are. I mean, the, yeah, the, the, I mean, the main things that we've come across are, are the chatbots and then the sort of writing tools.
Yeah. And a lot of the places where it's kind of like very rapidly kind of, it seems to be expanding very rapidly. Like I was reading a really interesting thing about, you know, the AI generated images recently. Oh, right. And it's getting to the point where, you know, you can tell it, oh, I want a picture of a squirrel surfing and it'll go away and using a really draw you a picture of a squirrel surfing all on its own. Yeah. And they're quite, they can't.
Quite see the uses of that, but.
Well, no, I mean it's not for marketing is it. But I think stuff like that's quite interesting. Like I can see a place in the future where like on I stock, for example, if you want a very specific photo of something for a landing page, and it doesn't exist potentially in artificial intelligence could make it.
It's almost like, yeah, photo automatic photoshopping.
Yeah. But I think certainly kind of.
Even with that, you're gonna then need a somebody who can use Photoshop to go in and just tidy it up and make it. It's never going to be, you know, it'll probably do most of the job. I still think you're gonna need somebody who knows, you know, about design and, and imagery to actually go and just smooth the edges and make it work. I suspect there's still going to be a human needed at some point.
It's a bit like really, I guess, isn't it? What we're saying? Artificial intelligence is a bit like having a toddler or a bad intern or a new apprentice, in that you can give them stuff to do work. Yeah, but you're going to be sitting over its shoulder the entire time being like, let's, let's fix that. Let's correct.
At the moment, sometimes it's like, you know, you know, when the, um, you're like doing some cooking and the kids help. So then it takes twice as long. Good job. At the moment, it's probably at that point where actually by helping you, it's making you take longer.
But yeah.
It could get to the stage where it's sort of a, yeah, an an intern and it does a decent job, but you just have to check it over and, and tweak it a little bit later. But yeah, I think, I think it probably will. I mean, almost certainly will.
Yeah, I think certainly on the writing side, I can see a situation where, you know, you could you could probably use artificial intelligence, like you were saying now to kind of like block out the framework of an article or tell you what topics to write on and what headings you need and that kind of thing. I can see it getting there.
Yeah. The, the sorts of things you have to cover and it's, it's close to that now. And it probably could get, you know, it's going to get better. It's bound to. Um, so yeah, that, that sort of thing, it probably will help say that the whole chatbot thing. So some of them are quite good. So I mean, again, the, the bad ones will probably get better.
Yeah. And I think it all does go back with chatbots especially. It goes back to the rules you set doesn't it? Because certainly like with ours, for example, I think ours, the one on our website set up in such a way that if you ask anything even remotely complex, it just kind of tells you to get in touch, doesn't it?
Doesn't think so. Yeah.
Um, and I think that kind of thing where, you know, you have to kind of anticipate use cases and work out how people are going to use it and try and kind of program it in such a way that it's not going to go horrendously wrong. But yeah, I mean, it's interesting to think about, I guess, isn't it? We may one day live in a world where we can comfortably say, yeah, use artificial intelligence for everything.
But yeah.
Yeah.
Yeah, we're, we're just not quite there yet.
Not quite there yet. No. Which is good for us, I guess.
Yeah. Keeps us on job security a little bit longer.
Yeah. The machines aren't coming for our jobs yet. I guess the other place it's probably worth kind of talking about this is in Google search itself, right? Because that is a form of artificial intelligence actually.
Yes. It's probably one of the most most sophisticated ones out there, isn't it?
Yeah. And that's I guess. Well, yeah. I mean, that's a really good point, isn't it. I mean it is kind of increasing in sophistication.
Yeah. I mean it used to be that if you typed one thing in, it would only give you the exact thing you'd typed. You know, the exact word you typed in would come out the other end, whereas now it's all the semantic stuff, so it knows if you type something in. That means the same as something else and it'll show you.
You know, sometimes sometimes it gets it very wrong.
Wrong. But I mean it. It's head. And again, it's getting better all the time as well. But, um, which makes writing content actually a lot easier. Yes. You know, the whole right for humans, not for search engines. It is that that kind of works more now. So instead of having to say the exact words seventeen times in your document, you can just talk about the subject.
And it did actually used to be like that, didn't it? At the beginning, it was like, if you didn't have the keyword in there fifteen times, Google wouldn't understand that. That's what you're trying to target. So yeah, I guess that's a place where AI advancements have made our lives an awful lot easier because now we don't have to kind of jump through those hoops.
And we also means, yeah, yeah, for people and, and people reading it don't have to read through things that are really quite painful to read. You know, you're actually reading something that that can be enjoyable to read and Google will understand what it's about without you having to make it blindingly obvious.
It did get a bit awful, didn't it? For a while. Mid mid kind of twenty ten. That sort of period where it's like people were almost writing like robots, just to kind of tick the boxes and get the keywords.
You had to create a sentence that kind of used the.
The.
The phrase in the right order with exactly the same words in it, even though it didn't quite make sense in the sentence. But, um, fortunately, we, you know, Google has moved on and, and you can write proper English now.
Although one of the things I find really interesting actually thinking about that is that the advancements they've made in kind of Google search with the like, you know, kind of semantic intelligence hasn't passed through to Google ads because this is something we see quite a lot. You can target a keyword that's very kind of specific. And rather than finding lots of things that are closely related to it, Google will just go and show your ad for completely unrelated keywords that it's decided a kind of semantically linked.
Yeah. It's yeah, there are occasions where you're like, yeah, because you can look at the actual keywords that came up and then you can use it, look at the, um, what people actually search for and you can look at it and go, well, that's not what I said. It's not even near.
We were writing ads for, um, you know, how to get planning permission for camping sites. And they were showing up for all kinds of keywords like, um, can I put a log cabin in my garden? And it's like, I'm not. Yeah, I know.
Well.
I mean, that's not, I mean, the.
Classic one was we had a client that did sort of, um, house extensions. Yeah. And it showed up for eyelash extensions, hair extensions, extensions.
Right.
Yeah. But it just didn't, it didn't know what kind of extensions we were talking about. And it would have been really quite helpful if it had been able to figure that out because they're very different.
Yeah. Artificial intelligence in inverted commas. Yeah. Yeah. Yes.
Yeah. Not quite intelligent enough in that situation.
So yeah. I guess just to kind of.
one thing, actually one thing. Have you seen the thing on Facebook where basically it knows who the people are in your photos?
Mhm.
Which is really scary.
Yeah. Stuff like that's kind of like that verges on creepy.
I think I, that kind of freaks me out a little bit because is this so-and-so like, uh, yes it is. How did you know?
Yeah, I saw there's a couple of companies, um, that I've seen presentations from before that, um, they're really big into this whole idea of user generated content. Yeah. And they're trying to push to a point where like, if you use their services, your products could be auto tagged in other people's like Instagram posts, as in it would recognise that those were your shoes.
Really?
Yeah.
Oh my.
Goodness. And then you could like bring it into your website on a feed and, you know, stuff like that. And I think stuff like that's like really quite invasive. That's the sort of stuff that gives marketers like a.
Yeah, pretty much.
Yeah.
Yes. We're just about wiped out retargeting, which people call the creepy ads that follow you around the internet. Um, and we're finding new ways of creeping people out.
Yeah. We we do have to be quite careful, I guess, don't we, about over leveraging these tools because there is no way of kind of turning them off or turning them down or saying, you know, don't be too invasive. Yeah. They don't understand people or the rules, do they?
There's just no, they haven't got any sort of subtlety built in.
Yeah. So yeah, I guess that's a, a semi optimistic, semi pessimistic take on I think the future.
Yeah. Yes. Currently it's um it has its uses. In future it will have more uses. But um as with anything the marketers will probably ruin it and yeah, make it completely useless for everybody.
We have to be careful not to overuse it. Yeah for sure. And also, I guess just be careful, I guess, not to over rely on it because the skills that it's trying to replace writing good ad headlines, working out how to target your ads, working out how to get your messaging right and appeal to audiences and, and how to write. Yes, all of this stuff is really hard. And sometimes when you kind of sat there at nine o'clock on a Monday morning and you think, if only a computer could do this for me, but at the end of the day, those skills are what makes marketing campaigns actually work. At the end of the day.
So yeah, it's knowing, knowing your topic and knowing your audience and actually figuring out what's going to work. And also, you know, trial and error and how AI does trial and error, but in a different way.
In a.
Way, the way humans do. So, you know, by all means, use it. But, but sanity check everything.
Yeah, yeah, yeah, I guess that's the thing. I guess that's pretty much everything, isn't it?
Yeah. It's it's interesting. It might be when we revisit in future and see where it's going, where it's where it's got to.
I'm sure David will have some thoughts as well.
But no doubt, no doubt we may revisit this.
But yeah, if you're listening and you have any thoughts about artificial intelligence and or want to ask any questions, you know where we are, how to get in touch. But for now, I think that's, that's goodbye from me.
And goodbye from him.
Excellent.