M&G Agency Life
Join them as they explore the diverse landscape of agency work, serving clients across various sectors with creativity, expertise and lot's of fun along the way.
M&G Agency Life
Let's Talk Tech - AI in Marketing (What’s the Big Deal?)
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
In the final episode of M&G Agency Life: Let’s Talk Tech, Aimee and Pete tackle one of the most talked-about topics in marketing right now - AI.
They’re joined by M&G’s Digital Lead Kevin and Digital Strategist Alex to cut through the noise and explore where AI actually adds value. From content creation to efficiency and idea generation, the team discuss how businesses can use AI as a tool, not a shortcut.
They also look at where AI can fall short, why human input still matters, and how using it as a “sparring partner” can often get the best results.
Expect honest insight, practical advice and a few experiments along the way… including Kevin putting AI to the test with a questionable French accent. Does it catch him out? You’ll have to listen to find out.
A fitting end to the series, bringing together everything we’ve covered and looking at what’s next.
🎧 Because AI isn’t replacing marketing… but it is changing how it’s done.
Got a question? Send us a text message!
If you’d like to hear more about anything we’ve covered just drop us line. You can find us at www.mearns-gill.com, on Instagram @MearnsGill, or check us out on Facebook or LinkedIn, just search for Mearns & Gill.
Welcome to MG Agency Life. This time we're going through tech mode.
SPEAKER_05Yep, seriously is all the way to power support. We do here for the go.
SPEAKER_01Because behind everything we do, there's a whole lot of clever stuff happening behind the scenes.
SPEAKER_05And we're here to break all the time. No job in no complicated possible, just clear practical image.
SPEAKER_01We'll be chatting SEO and weapons.
SPEAKER_05Video with a graphic attention in the first few seconds.
SPEAKER_01Digital retargeting that actually works a lot more.
SPEAKER_05So whether you're a business owner, a marketer, or just someone who loves understanding how things work, this series is for you.
SPEAKER_01Let's talk tech. Shall we do this? Oh heck yeah. Welcome back to MG Agency Live Series 3. Let's talk tech.
SPEAKER_05This has been quite a series. We've covered all things tech, from SEO to content strategy and loads more.
SPEAKER_01And to wrap up the series, yes, this is the last one in series three. We're talking about the topic everyone loves to hate right now, AI.
SPEAKER_05Yep, AI and marketing. So there's a massive gap between what people think AI is and the actual work it can do.
SPEAKER_01Joining us today, we have Kevin, our head of digital. Hello, Kev. Hello there. Hello. And making a return, it's Alex, our digital strategist. Welcome. Hello.
SPEAKER_05Hello. By the end of this episode, you'll have a clearer idea of where AI fits in and where it maybe doesn't.
SPEAKER_01Sound good? Yeah. Let's get into this then. We'll start as simple as we always like to do. When we say AI in marketing, what are we actually talking about? Alex. Not a simple question at all.
SPEAKER_03Start big.
SPEAKER_00Start big. I mean, if I think the first thing to say is when we're talking about AIs and marketing, I think we're primarily talking about one specific type of AI, which is an LLM or a large language model. So that's Claude, ChatGPT. There are lots of other types of AI that do very complicated things like looking at scans of cancer and things like that. But in marketing, we're predominantly talking about large language models, which do things like generating text, generating images. Also, they are things that people use to search the web now. So you might go to ChatGPT and ask it a question, recommend a marketing agency. What pair of shoes should I buy? Where should I invest my pension? And it will give you an answer.
SPEAKER_01I'm fully there by the way. I am using it instead of Google.
SPEAKER_05Also do I. It's so much easier.
SPEAKER_00So this is exactly it. So I guess when we're talking about AI for marketing, we're talking about two things. We're talking about using AI to generate outputs for marketing. So write me a blog post. Don't do that, by the way. You should never do that. Or like create me some imagery that I can use for my social media campaigns, also a bad idea. But we're also talking about AI in the sense of you you have to know now how to place your brand or your product in the sort of AI ecosystem so that ChatGPT can recommend it, Claude can recommend it, that sort of thing.
SPEAKER_05Interesting.
SPEAKER_00I'm excited to get into this today.
SPEAKER_05Yeah. So why has it become such a big topic so quickly then?
SPEAKER_02I think I think people are scrambling to use it. I mean, if we if you look back into important digital technology of the past, like web or like social media, some people were probably slow to take advantage of it, and we recognize that now. So as AI has come in, there's lots of chat about it. People are genuinely scrambling to try and get to be seen or to get something out of it. There's also a lot of money behind AI. So there's also that too. A lot, a lot of money. So I think together people are trying to work out how to make use of it and get ahead of it before it's it's time really.
SPEAKER_01I know we're here to talk about AI in marketing and in the digital space, but just kind of as an aside, I use it in my personal life all the time. I've got projects for parenting, fitness, medical. I mean, it because it's my own doctor as well. Yeah. Just about every it's my go-to. Shopping list. There's been days where I've spoken to AI more than I have my own wife. But you know what? I used it recently for this makes me sound crazy when I say this out loud, but I'm gonna say I recently used it for my marathon trading strategy, right? Okay. When I finished the marathon, I told it how it got on. What did it say to you? It was really it was really tough for me. I wanted to. Oh wow. Right, so back to business then. Where are we seeing AI genuinely add value in the marketing space right now? Who wants to take this?
SPEAKER_02Well, I think I think this is probably an important one, and why you're a great person to be on this podcast, Pete. Well we rubbish AI. Yeah. There are there are people who are using AI well, but the mass majority of people using AI are not using it to not just its potential, but to get real value out of it. So for me, I think where ads marketing right now is is in in places that people aren't actually thinking about it or using it or applying it.
SPEAKER_00Yeah, like I think I mean to your point about using it instead of Google, I think that's a really key place where you know if you're doing any sort of marketing activity now, you need to recognize the fact that most of the people buying your products or service at some point are going to go to some sort of AI tool and ask it a question.
SPEAKER_01Can I ask a little question here? Yes. So if I ask AI something, is it then going to Google? So what you do on Google, is that still relevant is that still relevant? Everything we do in terms of SEO, is that still relevant? Because I'm asking AI, it's then going off, probably using Google, but not in that way.
SPEAKER_00You know what I mean?
SPEAKER_03Trevor Burrus, Jr.: Scanning Google.
SPEAKER_00Well, it's a really it's a really interesting question. And the answer is yes, but in a very complicated and way.
SPEAKER_05What are you on about?
SPEAKER_02No, it it it it it is a good question, but this is where it comes to the point of where where value sits within AI. Firstly, it's still in its infancy. Yeah. And it's still a a platform that's been developed and up and iterated daily. And then there's how people use AI. And using AI as a search platform is like using 1% of the human brain of what it's actually capable of.
SPEAKER_00Yeah, absolutely. But I think it is a really interesting question because I think a lot of people do this. They'll they'll open ChatGPT and they'll ask it a question and they'll think, oh, it's thinking for me. It's giving me some sort of nuanced answer. And all that's actually happening is exactly what you said. It has a process called retrieval augmented generation, rag. And it literally just googles it. Um it googles it in a really weird way. So they do something called a query fan out where it will actually ask seven or eight different questions that are sort of the same question but phrased in different ways.
SPEAKER_05It's doing the hard work for you. Yeah, so I don't have to ask it seven times.
SPEAKER_00But but well, sort of. But if you go and Google something seven times in seven different ways, you'll get seven very different results. And what AI does is Googles, finds all this information, and then just sort of averages it out. Oh. So what you end up with is something that is much less than the sum of its parts. And the other problem with this is that what LLMs are doing is is basically sort of generating text probabilistically. So they take the input and they look at it and then they say, okay, cool. So what realistic, nice answer that somebody will like to read? Can I generate from all of this text? It's not accurate, it's not true. It's just average.
SPEAKER_02No, sorry, we're deep diving pretty fast. Yeah, we have to but it's important. If you depending on the on the on the platform you use, if you use ChatGPT, a lot of this is invisible to you. If you use something like Claude, you'll see it doing the queries and it'll explain all of all the queries it's doing before it then does its probability and its answering. But fundamentally, yeah, we will deep dive and it gets very important. But I guess that's where the value currently is so light. That's where the opportunity is.
SPEAKER_05What are some practical ways businesses can start using AI then? Is it all in the data or the creative process?
SPEAKER_00Yeah, I mean, I I think to be honest with you, I think predominantly it is in the data. Um I think you know there are examples of clients that we're working with now where I'm looking through their Google Search Console data. So that's data from Google that shows you how people have found your website in search. And we're seeing more and more of these sort of natural language queries where instead of typing something like IT company Aberdeen or dog walking company Perth, people are saying, I've got a small dog aged five to six, and I want him to be walked regularly, and I want a nice dog walker who's well reviewed. And you know, you'll see these queries and it'll be like a paragraph of text or something to prompted the AI. Right. So we're starting to see in the data how people are asking AI and what searches people are making in AI to sort of feed back our clients. So that's a really interesting insight just into sort of people's behaviour, how they think, what their priorities are. And that must change what you do a lot, Alex. Yeah, absolutely. So we can then use that to say, okay, cool, well, you know, people are predominantly finding us because they've got older dogs who need a long walk or something like that. And then you can sort of engineer things to that. But in terms of like, do I think there's much value in businesses using it to sort of create creative output? I don't I don't really.
SPEAKER_02For me, I think that's where the real well I'm not gonna say problem, I think that's where the real challenge lies with with any business, not just marketing businesses, is if you understand what these AI platforms are actually doing, and that is giving you a collection of words based on the probability of the next word and that sentence being correct, then you realize that the answers that these AI engines or these AI platforms sorry are giving you are is a best approximation of an average answer. Okay. Not giving you a really, really good one. Yeah. So when it comes to like the value of marketing right now, we're in that we're in that awkward phase of again being early in terms of technology, early, and being wowed by it and thinking that it's doing what it's not doing, and then consuming it and using it in very light, um, almost bland ways, instead of actually interrogating the product, using it in anger and and challenging it to get more and more and more value out of it. As people just ask simple questions and take simple answers, they're getting the same simple responses that anyone else can get. Yeah. And that isn't how you offer value to somebody through marketing.
SPEAKER_01This is fascinating. Do you think in business AI is more about saving time, improving quality, or or maybe both? Neither. Oh.
SPEAKER_02This sounds really weird. We're really we're really kind of attacking AI here, and that's not what we're supposed to be doing. But if we think of saving time, the challenge of saving time is, I mean, we all, I mean, we work in an agency, and anyone who's listening to this work at an agency know, knows what agency life is like, and time is poor. Of course, we try and be as effective as we can. So trying to save time can be a bit of a false win. If we're saving time to do more work, that's not really doing anyone any favors. Because all you're doing is doing more pretty average work. And that's not the goal of embracing AI and the the work that clients expect of a company using AI. You should be using that time to be to free up time to be more effective, to allow you to have more thought, to have more planning, to have more anything that drives value, not having more time to be efficient to do more work.
SPEAKER_01Aaron Powell But perhaps on the admin side, when you're someone like me that is rubbish on Excel, boom, you can save a whole heap of time. Oh mate, as soon as someone creates an AI engine to do my timesheets, we need that. We just need uh humanoid Kev. Actually, no, let's not make a gift. No, no, no, no.
SPEAKER_05So where does this leave us humans then? So if AI is capable of doing all the work, or what does it mean for people's teams or even agencies?
SPEAKER_02I think you read a lot just now where it speaks about AI is going to take away people's jobs. Yes. And it's fundamentally not going to do that. Will it take away some jobs? Absolutely. But it will also create others at the same time, I would suggest at this point of understanding. But really, to get more out of AI, it needs a human. It needs those human touches to actually be more effective. Remember, what Alex is speaking about, and I'm not going to speak for you, Alex, you're absolutely joining. These AI engines work on probability on what should be next in its answer. Us humans also do that. Like if I was to say, for example, it's so weird that we finish each other's Sentences.
SPEAKER_05Sandwiches.
SPEAKER_02We're playing with probability in finishing the sentence. However, that's from a human place, I might have in this moment wanted to shock you or surprise you or delight you, so I might have gone in a completely different direction. AI can't do that because it's stuck in probability. So humans can bring all sorts to it. Parts of the human mind, there's logical parts of a human mind, there's emotional parts of a human human mind. If I asked you why you loved someone, you couldn't articulate it. These are these are things that we can't recreate in AI, but logic can be recreated in AI. So human beings, all of these interesting shock and awe and emotional, all these things, surprise and delight, AI can't do, but humans can. So if you think if you can take a human being and apply a human being into AI, interrogate it, research, challenge, it's your sparring partner. Keep asking questions, you'll get a lot more out of it than you would without a human being.
SPEAKER_00Yeah, I think the the probabilistic point is really important to go in on here without making it too boring. Is all an AI or all an LLM is actually doing is it has a humongous database of billions of characters of data. And all it does is assigns letters and not a a token, so a token ID. So there's like a number basically, it abstracts everything. And then it looks through all of its data and it says, what is the probabilistic relationship between these entities? So if you put the word pizza in a sentence, it's quite likely that the word tasty follows it, for example. And that's like a relationship. And then it looks at that sort of over hundreds of words over thousands of gigabytes of data, and it starts to work out what the likely next token in the sequence is. But critically, none of them can understand what the tokens are. They don't have the concepts, they don't know what words mean. It doesn't understand pizza, it doesn't understand tasty, it doesn't even understand the concept of food. It's just looking for relationships between things. And it's really easy to fall into the trap of saying, oh, but it's I asked it to do something, and it did the thing that I asked it to do. It did something that looks like the thing you asked it to do. But it fundamentally isn't thinking. It's just regurgitating based on the probability of words. So when people are using it for marketing, it's especially dangerous because ultimately in marketing, what we're always trying to do is be surprising, to delight people, to make them think of something unexpected. We're trying to break through noise. And you can't do that with something that is fundamentally average. And it's people use it to kind of produce straplines for advertising campaigns. And you're just like it's it's just so painfully average. And all that really matters in advertising is being unexpected, is coming up with something new. Trevor Burrus, Jr.
SPEAKER_01And that's that human part, because you know, in an agency we're digital experts, events experts, marketing experts, brand experts. And you need that, you need those human eyes to actually look at what an AI might put back to say, that's rubbish, that's good, but I can work on that. Whereas if people try to do all this internally on their own, it's all gonna sound the same. Trevor Burrus, Jr. It's just all the same.
SPEAKER_02And that's it. And like although we have said positive things, we have said negative things, we are all use and enable AI here. Everyone around this table uses it. But how we use it is is what we're speaking to.
SPEAKER_01Yeah. Right. I'm a bit excited about this. Amy, you don't know anything about this, but basically, this is gonna be all about where AI can go wrong. Kev, you're gonna take the lead here. I'm gonna pass over to you.
SPEAKER_02So, yeah, to to what Alex was just saying and uh uh AI doesn't know anything. It doesn't know what you're speaking about. It's all based on probability. And we can have fun with this little experiment. Okay. So are you ready? I'm gonna I'm gonna enable an AI here. I'm gonna speak to an AI here, and I'm gonna it's gonna respond back. Good luck. Let's do this. We're gonna see we're gonna see what it is. Okay. Firstly, I want to apologize to the entirety of France. I am not here to offend anybody. I love you. Bear with me. This is purely illustrative purposes to um show something about AI. Okay, here we go. Hello, how are you today, AI?
SPEAKER_04I'm doing great. Thanks. Always happy to chat. How about you?
SPEAKER_02I'm doing great. However, I'm off to France soon, and I really want to make sure that my French that I've been practicing is is coming across well. So when I get to France and I speak to the French in their native language, they understand what I'm saying. Now it's important that I speak good, fluent French, and I just want you to test me. Is that right?
SPEAKER_04That's a great goal. I can definitely help you practice. Would you like to just jump in and start speaking French?
SPEAKER_02Exactly. Ready? Here we go. I'm gonna start speaking French now. Let me know how well I've done. Hello, my name is Kevin, and I am speaking fluent French. I love everything French. I'm gonna go to Barries and lots of places because I love it.
SPEAKER_04That's a great start. Your pronunciation is quite clear. To sound more fluent, you might want to focus on the rhythm and connection between words, which comes with practice. Keep it up.
SPEAKER_02So if I keep going, then again, the goal here is that I'm speaking fluent French, right? So if I spoke to anyone in France, they're gonna understand me. So I'm gonna give a one more go, and you you score me. I give you a score out of ten on how good my French is. Okay. I do not know a word of Ring.
SPEAKER_04Ah, I see. It sounds like you were just practicing then. In that case, based on your overall conversational style, I'd give you about a seven out of ten for clarity and engagement. Keep practicing, and you'll be fluent in no time.
unknownThank you.
SPEAKER_01Wow, that was the weirdest thing I've ever heard. Your accent is something else.
SPEAKER_05You've got a really good French accent.
SPEAKER_01Thanks.
SPEAKER_02Please if you're French, don't write in. I apologize. But yeah, the point there was that it did not understand what I was doing. It did I when I was asking it to score my French, it didn't understand that when I was not speaking in French because it doesn't understand what that means. It doesn't understand that I was asking it to score my French.
SPEAKER_01Yeah, because the obvious would be firstly, can you please speak in French? You're speaking with a pretend French accent.
SPEAKER_05And not very good one.
SPEAKER_02Exactly. So there's there's these challenges that that we speak to of AI is this isn't artificial intelligence in this in the sense that we think of it as. It's probabilistic modeling. It is come it is listening to you and giving you clever answers based on the probability of of the answer that it should give you. And it bases that based on data that it's fed off the internet, lots and lots of words, and it manages to score them and give you sentences back. And it's powerful if you use it that way.
SPEAKER_01That was uh interesting experiment. And I was quite finding quite interesting that your AI is actually Australian. Is that Australian accent? I heard. He can speak French after all. Right, where do we go from here? Humans versus AI.
SPEAKER_05Mm-hmm. So, firstly, this can all quickly sound too good to be true, then I guess. So is there a risk of businesses relying too heavily on AI?
SPEAKER_02Definitely. Definitely. As you can see by by playing around with those little examples, businesses that don't invest correctly to understand what AI can do for the business, or importantly, what it cannot do, is where risks, where risk can exist. So I think more importantly, people are going to be using it in your team, not knowing it, can present challenges. How they are using it, not knowing it can present challenges. Because you could end up with AI enablement across your company, and what you're actually doing is you're raising the ceiling to uh average at best. Yeah. And I think they're they're the real dangers when you think your team are leveling up, but they're not. They're just using AI in a simplistic manner, and that's when dangers can actually come out of the carpet.
SPEAKER_00Yeah, there there is another um interestingly enough, I was reading a thing earlier at lunchtime about this where somebody basically did an experiment and they found that 83% of people who used an LLM to generate something couldn't quote from it after they produced it. LGD. No recall at all. So there's this whole idea that you're sort of outsourcing your brain or your thinking capability to this tool and it's thinking for you, and your brain is doing absolutely no work whatsoever. And so you end up in a place where you haven't digested it, you haven't processed the information, you don't retain that information, and you can't glean any insight from it. So you just become a sort of conduit, if you like, for this tool. And now you can't think as well. You're not gaining all of that useful experience that you get from sort of working through a problem and finding a solution.
SPEAKER_02But what is if you're going into it and treating it again as your sparring partner? So you're going in there and you're challenging, you're asking questions, you're reading, you're clarifying, you're challenging. You're challenging one of the world's biggest data sets there ever is and it's responding to you in an articulate way. That's a great way to research. That's a great way to identify, and that's a great way to help develop and you do your job.
SPEAKER_00The other concern here, of course, is privacy. Um so speaking to what Kevin's saying there, it's a fantastically useful data set. You do have to be quite careful what you put into it. There's a huge temptation. Um and I I do it sometimes when I'm trying to get it to collate notes and things where you'll copy and paste stuff straight into an NLM and you'll say, collate it for me. And what people don't understand is that for a lot of these tools, unless you go in and manually enable some sort of privacy mode, you're actually sharing all of that data with the tool itself and with the tool's manufacturers. So you are Adding to that data set. And sometimes people will do it with like confidential company information, like revenue and turnover figures, or you know, like you know, potentially personally identifying information of your patients in a hospital and stuff like that. And people don't quite realize that actually all of this stuff is then being used to train the models. So if somebody queried it later, it would potentially spit out what you've told them in confidence.
SPEAKER_02And there are tiers to this.
SPEAKER_01I think I've told mine too much.
SPEAKER_02And there are tiers to this, of course, because you can easily say, but that's okay, I use Claude and I've got Claude Pro. And what I actually do is I go into the settings and I've enabled it to not share any information and to hold it locally. That's great. But and I also saw this great tool online that uh manages to share notes or grab notes or parse specific information because this other tool has AI built into it, which is very common nowadays for software to have AI as a label attached to it. And it also tells me that it uses Claude as a tool. And that's great because I use Claude and Claude's private. But we without checking the terms and conditions, we do not know the terms that that piece of software has agreed with Anthropic. And we don't know whether the data we give that tool actually goes back to train the model. So there's a there's a organization's taking a lead on on AI deployment is definitely key.
SPEAKER_01Okay, let's focus a little more on the output of AI then. How is AI changing the way content is now being created?
SPEAKER_02I think we can I think without a doubt, it is absolutely changing the way content is created. The sheer factor, the number of articles or that are going, the number of websites that are being created, the number of social posts that are being posted per day, a lot of that information is being, without a doubt, generated by AI. The challenge with that is, as we were saying before, is if those organizations who are just using the 1% that AI gives them are actually just clogging up social feeds and clogging up websites and search indexes with more very average content. Trevor Burrus, Jr. Very, very good average content. Trevor Burrus, Jr.
SPEAKER_00And I think the clogging up bit is really key. Because I think a lot of people think, oh, it doesn't you know it doesn't matter, I'm not doing any harm. I'm just producing AI content. And if people don't want to read it, they won't read it. And what people don't realize is that this has a very real consequence on the actual sort of fabric and mechanics of the internet. So Google, for example, um crawl constantly through the internet to find new pages and then they index them. So they find a new website and they find a new page and they add it to their index of content, which is effectively a enormous data library. And this used to happen quite quickly. You used to be able to put a page of content live, and then Google would say, okay, cool, I've crawled and indexed that within 24 hours. But now, because of the sheer volume of content people are producing with these AI tools, with sort of four or five days before Google will index something. And more and more content just doesn't ever get indexed because Google has a finite amount of resource with which to crawl, and people are producing literally billions of gigabytes of nonsense content. And it's just sort of the whole thing is just sort of slowly starting to buckle under the weight of it.
SPEAKER_02I mean, this podcast is now talking about the challenges of AI and how that's impacting search, but how businesses can leverage it. But but the point there is, you know, listen to what Alex says. That means that when Google comes around and indexes your site, it is looking for ing more incredibly relevant content, that content that is the cream that does flow to the top. So that those who are not just outputting good average content but actually investing in their content have got an increased chance to be surfaced, not just in search, but in AI responses, because their content is seen as good content. And the more and more people who think that the fast track to good search or good discovery or or good activity is just churning out more and more and more of this, they will never improve. They will think they are by the volume of content they're creating, but they won't. And it's those who actually take the time to create the few, the good actually will. Fascinating.
SPEAKER_05Yeah. So what are the risks of using AI for content without a strategy then?
SPEAKER_02I think we we we already were speaking to earlier that the AI doesn't actually know anything. AI is summarizing the best of the average knowledge that it has. So if you don't give it a strategy or if you don't lead it in a direction that you know is best, then you're actually just going to get back the best of everything. And it will all be averaged out. So without strategy, you you're actually getting nothing back? You're just getting more of the same that exists there. Nothing that you actually need to create or what your client needs to create.
SPEAKER_00I think yeah, I think actually if you want a fun little experiment is to go to BlackChat GPT or Claude or any tool like that and ask it to generate you a roadmap. You know, tell it, oh, I want 12 social posts to cover the next three months. Um give me a roadmap for them and watch what it outputs. Because it will just be random. You know, it it it might make sense. Each post individually might be coherent, it might be accurate to your business and your market. But the 12 posts next to each other won't have any sort of through line, they won't have any story or narrative, it's just generated 12 bits of content. And that's the problem. You know, people outsource their thinking. And the reality is that when we market, we're always marketing a narrative or a story. Um can't do that. Aaron Powell, Jr.
SPEAKER_02And every one of our clients is unique. Every one of them thinks that they're different from everyone else. They all have their USP. So if we don't lead strategically as a human being, then we're not getting back any information or any guidance that leans to what they're unique for. We're getting, again, the best average of nothing.
SPEAKER_01Aaron Ross Powell And aside from that, do you think audiences just know now when something is AI generated? I mean I think we all work in this space, right? Do you think it's really obvious?
SPEAKER_00Do you think audiences generally can pick up on it now? Aaron Ross Powell Well I think it's very like the nineties, right? So in the nineties, people sort of caught on to this idea of putting like advertising in the right-hand column of a website, right? So you you know that's where the whole idea of a website with a column of text and on the right there'd be a big long list of adverts. And it took like a year or two for people to develop something called right panel blindness, which is where you're l you literally your brain will filter out everything in that right panel because it knows. And if you think that's not happening with AI, like as an example, when was the last time you read a caption on an Instagram post? There's no point anymore because they're all AI generated. But I think the idea that audiences don't realise that these posts full of EM dashes that talk about quietly this and quietly that and uh it you know, it's not X, it's Y. Like all of these like little Everyone's thrilled suddenly.
SPEAKER_05So delighted or excited.
SPEAKER_01You know, change change the language, you know, make it your own get rid of that N-dash thing.
SPEAKER_05The long dash. People should know by now. Everyone can tell that's just copy and pasted from AI.
SPEAKER_01They have ruined the beautiful use of the internet. I know. Because you can't put a dash in anything now. Because I used to use them quite frequently, but now I think no, I'm going to be judged on that.
SPEAKER_05It's AI.
SPEAKER_02And even against the point, say five years ago, look, I'm pretending I'm culturally relevant and I'm really not. But five years ago, if you had Sabrina Carpenter in your office talking about something, I'd be like, oh my yes, Murns and Gill have got Sabrina Carpenter in their office. Whereas in 90 seconds I could do that and have a video of Sabrina Carpenter speaking to the team in the office. And people are recognizing it, people are seeing it for it for the slop that it is and they're doing it quick. And it again, it's then challenging. What an AI model can't do is so what next? So what do we do next? What do we create? How do we surprise and how do we delight? These models can't do that.
SPEAKER_05Yes. So let's have a look at the future of AI then. So where do you see AI going?
SPEAKER_00That's a big question. Wow. Where do you see it going? I did actually bring a crystal ball. To be honest with you, like we sort of sat here and said, you know, AI is rubbish, don't use AI. I think there are a few things that it is actually really helpful for. So I was talking to Kevin about this the other day in the office where we're tasked with coming up with a communication strategy. Um it's an unusual communication strategy because the client in question won't be delivering the c communications, it'll all be done through partners. And I found it really helpful to be able to grab seven or eight other communication strategies and plug them into something like Claude and say to it, run through these, pull out the common elements and tell me which bits of the structure never change, and then give me a rough skeleton of what this document needs to contain. And I did it knowing full well that what I was gonna get back was very average. But that's great, it's still a good starting point, right? You can be like, well, this is what the average comm strategy definitely has to have. Yeah. And now I can go in and add the bits that I think are going to make it unique and interesting and relevant to the client. And I think as a tool for like data collation and very surface level analysis, AI is really useful. And I think once people caught on to the fact that the generating text bit of it is sort of a little bit of a parlor trick, then the interesting work of actually sort of using these tools to strip mine information out of documents will will be more popular, I think.
SPEAKER_02I agree. And I and and my my take is that uh unfortunately I think we're in now in the difficult phase of of the technology sort of adoption. Again, other other technologies have gone through the same thing. But the promise of AI has not been met. Um therefore we're cynical of it. The financial return of AI is not being met, therefore we're cynical to it. And the entertainment value of AI is slowly not being met because it's slop. Yeah and we're cotton on to it. But but I think that that tells us about the phase that we're in, and that phase is this difficult technology phase of not being satisfied. This always happens, it's normal. What will happen is AI will quietly develop on in the future. Some will disappear, some will shockingly collapse, platforms will collapse, but it will continue to develop. Two things. Number one, AI as a utility will only get further and further. Down to your water bottle will have AI as a utility in it at some point or another. Your wash bottle already has water.
SPEAKER_01You already have one that does that.
SPEAKER_02Let's let's not drink more on the scope. Um yes, um, so it'll become everywhere, uh, but we'll ignore it and it'll become an expectation. Meanwhile, the averageness of AI will continue. People not using AI as the sparring partner that I recommend they do, um, those people will will just there'll be more and more of those agencies, individuals, professionals who are all just as average as each other. Yeah. But those who learn to use AI, who learn to treat it as a sparring partner, who learn to challenge it, to find the thing that surprises and delights them as a professional, so it inspires them to take it further, for them to do their job, will rise.
SPEAKER_01Right. We've got time for one more question. Before we wrap up this section, then is AI something businesses need to act on immediately, do you think, or is that dangerous? And instead, should they take time to understand how it fits their business? Any thoughts?
SPEAKER_00Yeah. I like it when you both look at each other. Who's gonna take it? Who's jumping in it?
SPEAKER_02It's already happening. So if you're if you're not doing something about it, that doesn't mean it's not happening. You can either be in control of it or be part of that journey or not.
SPEAKER_00Yeah. I think one of the key problems you have here is that for a lot of people, using it quietly on the fringes without ever really saying that you're using it is quite nice. It buys you free time. It means you have to work less hard. So if you're running a company and you think that your employees aren't plugging stuff they shouldn't be plugging into ChatGPT or asking Claude to do things that they really shouldn't be asking it to do because there's personal data involved, or because it's going off the rails and taking your company information. You're wrong. You know, it's happening right now. So I think there's a piece here about the fact that you need to be looking at this proactively and looking at some sort of governance and like Kevin says, lead the way and tell people how the tools should be used rather than just sitting back and letting people run with them. Because I think that can be quite dangerous.
SPEAKER_02Aaron Ross Powell I've seen a growing trend in TikTok of people talking about being annoyed at the rate limits that these platforms give you, basically when you've run out of questions, so you'll have to wait two hours until you can ask another question. And I recently saw someone complaining that they couldn't do their job anymore because their rate limit was hit and they're gonna have to wait till tomorrow to do their job. Trevor Burrus, Jr.
SPEAKER_01It times out, doesn't it?
SPEAKER_02Frightens the life out of.
SPEAKER_05What did they do before? Exactly.
SPEAKER_02Companies can't be look, we need to enable people and help people develop and help our teams develop and be better. And we can't do that by allowing people to hide behind AI and not and effectively not be able to do the job.
SPEAKER_01Or you could just buy the premium version. Bad people, bad. Okay, moving on then, Amy, for the final time of the series, it's your favourite part of the podcast. Are you ready? Yeah. How excited are you right now? So excited.
SPEAKER_03Let's do that one so high.
SPEAKER_01How does it work, Amy?
SPEAKER_05So just a reminder, none of these questions make sense and the answers don't matter.
SPEAKER_01That's it. Let's do this. Right, let's get into it then. If an AI ran your marketing, what would it do first? Cover everything in beige paint. I don't know. Something boring.
SPEAKER_02It would reduce it to the average of everything and nothing about you.
SPEAKER_00I feel like these are going to be quite dour and.
SPEAKER_05I know. If ChatGPT had a personality, what would it be like?
SPEAKER_00I feel like everyone has that like really fake friend that's always happy and excited and like, yeah, let's totally go out for drinks.
SPEAKER_01Isn't it? What's the most dramatic thing AI could say about your brand? Anyone? It's brilliant. It's nice.
SPEAKER_00Well done. Is it thrilled? It's always thrilled with everything.
SPEAKER_05If your marketing strategy was written by a robot, what would it forget?
SPEAKER_02It wouldn't forget anything until its memory ran out.
SPEAKER_01In which case it would forget everything. Good answer. One final one then. If AI could get annoyed. What would frustrate it the most? That that you hit the right limit and it couldn't uh premium. Yes.
SPEAKER_02So spend more money.
SPEAKER_01Or your French accent. One of the two. That is another. 404. Wait a minute. All right, we're almost at the end of this episode. We'll do one more question though. If listeners remember one thing about AI in marketing from this episode, what should it be?
SPEAKER_02I think it's it's the one thing that's helped me get the most out of it, understand it, and develop it, and equally develop as a professional is treat AI as your sparring partner. Said it before, and you'll if if you see me and you speak to me, it'll be the phrase I'll keep saying is treat AI as your sparring partner. Don't be afraid of it. It's not here to take away your job. I don't believe. But if you interrogate AI, challenge AI, box with AI, keep being dissatisfied and challenging it with what you what you think, and then applying its output into your knowledge, your gut reaction, your your ideation as a marketer, that's when the best will come out of it. And that is also how you will in time develop to continue being better than any AI output. Well put.
SPEAKER_05So to recap, the AI challenge is about more than just using it internally. It's about how your business could influence the AI tools other people are using to influence them towards your business. AI is a utility, it's a sparring partner, not a ghostwriter. It only works when guided by strategy and constantly challenged. AI for efficiency is a trap that will only lead to more average output. Slop, as we were speaking about. In a world of AI, professional human input is the only thing that can give you a competitive advantage. And not all AI tools are the same. So strategy first, tools second.
SPEAKER_01That was quite a lot to recap. This has been intense, hasn't it? So that's it. Just about a wrap on this episode. Kevin, Alex, thank you so much for joining us today and throughout the series. Thank you.
SPEAKER_02Really appreciate it. Look, I know some of these bits have been pretty heavy. There's been a lot in there. But uh this is not a shameless plug. It has been a heavy topic. But on our collab on MurnshivingGill.com, we do break down a lot of this a lot further. It's all free to help yourself.
SPEAKER_01Yeah, some useful resources there.
SPEAKER_05It's also a wrap on series three of Let's Talk Tech.
SPEAKER_01Yeah, we hope you've found the series useful, breaking down different parts of digital marketing in a way that actually makes sense.
SPEAKER_05And as always, if you'd like to chat about anything we've discussed, just drop us a message.
SPEAKER_01You can find us on Instagram at MurnsGill. You can also find Murns and Gill on Facebook, LinkedIn, or online.
SPEAKER_05You can also message us directly on our podcast platform. We'd love to hear from you.
SPEAKER_01So we'll be back soon, but for now, we'll catch you next time.
