AI tools like ChatGPT have received a lot of attention lately. From social media to blogs, AI tools and the capabilities they have is being discussed everywhere you look. But will AI tools truly change the way we approach marketing? Could they even take your job?
We address this in our on-demand webinar ‘Will AI Take Your Marketing Job?’, as explore the advantages of AI, but also the limitations marketers can face when using these tools. We cover:
- AI technology for marketing
- How to take advantage of AI
- How good at marketing is AI today?
- Overview of the best AI tools
- What the future might hold
Register to view our webinar on demand by clicking here, and why not get in touch to let us know if our insights helped you.
Napier Webinar: ‘Will AI Take Your Marketing Job?’ Transcript
Speakers:Â Mike Maynard
Hi, and welcome to our latest webinar from Napier, where we’re going to ask whether AI is going to take your marketing job. So it’s great to see a lot of interest for this webinar. And what we’re going to do is we’re going to do a very quick whistlestop tour through AI. And look at how it’s affecting marketing, and really trying to understand the implications for people who are involved in marketing.
So, in terms of our agenda, we’re firstly going to ask whether we should worry about AI. I mean, maybe it just isn’t that big an issue. We’ll then do a little bit of a dive into what AI actually is, and what we mean, when we talk about AI. And in particular, I think a lot of the hype at the moment has been around generative AI. So we’ll talk about that. We’ll discuss whether AI works, I think that’s very important. Depending on who you talk to, there’ll be very different views about the quality of the content that AI generates. We’ll spend a little bit of time talking about chat GPT. And then other writing tools. Obviously generated text is one of the areas where AI has been incredibly successful. And this is very broadly applicable across a wide range of different marketing disciplines.
So we’ll take a look at some chat GPT. And writing will give you a little word of caution about some of the issues surrounding AI. We’ll look at what I can do that’s beyond, you know, generative AI. So this is going to look at some of the things that you can do with AI, that perhaps is going to help speed up other aspects of your job other than creating content. And lastly, a summary. And I would very strongly ask anyone who’s got questions, please put them in the chat as we go along. And we’ll deal with all the questions at the end, if that’s okay. So firstly, I mean, there’s obviously been a lot of hype about AI. We’ve all seen, you know, the news about how massively important AI is. But the question is, you know, should we worry. And I guess, you know, one of the things we can look at is the impact on jobs. And AI, according to Goldman Sachs could replace equipment of 300 million jobs. And if we look about what they were saying, it was saying that the report that they wrote was saying about a quarter of the work tasks in the US and Europe could be replaced. And interestingly, the US and Europe are in line to get hit worse than most other areas in the in the world. And that’s because there’s far fewer manual jobs. And obviously, AI is not going to take over manual jobs at the moment, although robotics is doing a good job of, you know, coming in and doing similar disruptive things to manual jobs.
But I think what was interesting was as part of that report, when the BBC reported it, and they talked to some experts in AI, and they pointed out that actually, it might not be a case of fewer jobs, it might be the face of the case that there’s more competition, so you’ve got competition from other people, and competition from AI. And here we see a professor suggesting the journalists will face more competition, and that will drive down wages. And the analogy they gave was the introduction of GPS, and platforms like Uber, and that actually resulted in lower wages for taxicab drivers. Not fewer drivers. In fact, if anything, it resulted in more drivers over a period of time. So it’s going to be very interesting. But clearly AI is going to have a very big impact on what we do. And that might be in terms of reducing the number of jobs it might be providing competition to what you do. And that competition clearly can drive down wages. Now, I’m actually okay. Because I don’t know if people listening to this webinar know but one of my hobbies is short track speedskating. And I asked chat GPT about short track speed skaters, he gave me an excellent response, listing some of the best short track speed skaters ever. But of course, I had a bit of an ego. So I decided I was going to tell chap GPT that I should be one of the best short track speed skaters in its list. Now, this doesn’t happen every time if you want to go and try it sometimes check GPT claims and it’s never heard of Mike Maynard and however much I try and persuade Chet GPT that I am a good speedskater It completely blanks me. But other times it comes up with his completely made up story about this Canadian short track, speedskating called Mike Maynard, who, as far as I can tell, never existed only exists in chat, GPS, memory, and yet, you can see, chat GPT has very confidently given details about my career as a Canadian short track speed skater.
And I think this is something that’s very important before we, you know, really rush into AI, and all get panicked, AI makes mistakes, and actually AI probably makes worse mistakes than people do. Because AI is always incredibly confident. And so whilst Unfortunately, my career as a Canadian is short track speed skaters entirely false. The reality is, is that AI is not going to be able to completely replace humans, because there’s a risk that AI gets it wrong. And in fact, there’s even a term for this, it’s called hallucinations. So AI is hallucinating something that really didn’t exist. So now, hopefully, we’re a little bit less worried. Although we’ve been told it can impact jobs, we know that AI is not perfect.
So let’s have a look at what AI is. And basically, AI or it’s also called machine learning is really simulating human intelligence. And typically, what it does is it uses a neural network. So you feed this network data, the network is modified based upon the data. And then it can process new data in the same way. And the reason we do this is it’s modelled on a very simplistic understanding of how the brain works, where the brain house neurons, and those neurons are connected together. And those connections are made stronger or weaker, depending on our experiences. And that’s what learning is, very crudely speaking. So neural networks attempt to replicate this mathematically. And basically, a neural network is a number of nodes or neurons, as they will be in the brains. And there’s input layers, output layers, and then there’s a hidden layer in the middle. And there can be multiple hidden layers. So if we look at chat GPT, one of the big things that chat GPT did was when moving from GPT, to GPT. Three was they massively increased the number of hidden layers, and that gave a much more powerful neural network. And all you’re doing is you’re feeding numbers, which represent anything from text to images in one side. And the other side is outputting numbers, which can then be turned back into text and images.
So it’s a very abstract thing. And this is one of the challenges of neural networks and machine learning is that actually, when you build a neural network, you don’t have a deterministic understanding of how it’s working. And so it can be very hard to know exactly what’s going on. And this is one of the reasons why you have these hallucinations. And they’re very difficult to deal with if you’re building a neural network. So that’s all very technical. So what I’m going to do is I’m going to get my friend Shrek to explain it. And Shrek had a conversation with Donkey and talked about ogres and tried to explain that Ogres are like onions. And it’s not that ogres make you cry, or ogre stink, or indeed that they go brown in the sun. Really, the thing is, is layers, onions have layers. Ogres have layers.
So they both have layers, and layers in Shrek mind was complexity. And that is exactly the same in neural networks. So basically, a neural network, the number of layers is very important that impacts the complexity. Although not everybody likes onions as Donkey pointed out, and probably not everybody likes layers as well. And one of the things we’re talking about is why those layers make it difficult to build these complex neural networks. So let’s get into what a neural network can actually do. I mean, we’ve had a very brief theory around how it’s built. But let’s see how we can use AI. And so AI that creates things is called generative AI. And what you’ll see is you’ll see throughout some of these slides, you’ll see images and most of the images in the slides from now have actually been created using generative AI. So I’ve used darly, which is an image generator a lot of people are familiar with. Obviously, there’s other ones like stable diffusion, and you can see it generates some interesting images as we go through But although we’ve generated images, it’s interesting, this looks at funding for European AI startups. And if you look at it, the vast majority of startups have been funded around text generation.
And the reason for that is, it’s now relatively simple to generate text that is more or less equivalent to human quality text. And that text can be used in a very wide range of different applications. So you can, for example, create blog posts or articles from Ai generated text, you can also generate sales emails. And you know, something that people involved in PR may have seen is there’s already PR systems offering to do AI pitches to journalists, I’m sure journalists have seen it as well and feeling terrified about the onslaught of spam they’re going to get from machines, which of course is going to be one of the challenges is that creating volume with any of these AI tools, whether it’s volume of text or volume of images, is trivial. It’s the quality that really matters. So anyway, most people are focused on text. And a lot of startups are focused on tax because they’re using the same model as chap. GPT. So, for those of you that don’t know, there was an organisation formed called Open AI, very crudely formed to make sure that AI didn’t do bad things to humans. And, and they created these GPT models started with GPT.
One, we’re now at GPT. Four. And these models are basically what’s called large language models. So they’re very big neural networks that can actually create text. So a large language model is a neural network that’s typically been trained with large amounts of data has many parameters. And it basically does self supervised learning. So you feed it data, the neural network kind of learns from that data. Or, to put it more simply, what the neural network does, is it takes a lot of data, so a lot of text. And it tries to use that text to predict what the next most likely word is if a human was replying. So if you put in a query, the neural network is trying to predict the next most likely word, and then the next one on the next one. Interestingly, though, if you do that the responses sound very automated. So actually, what happens with neural networks, and this is exactly what happens with chat GPT is it will produce the most likely word around about four out of five types. And then one out of five times approximately, it’ll produce something that’s, you know, likely, but not the most likely, and that produces a much more natural response. But crudely speaking, chat GPT, and any of these GPT based models, they’re returning what they think or what they aim to be the average response.
So one of the first things to say is, if you’re concerned about your job, and whether AI is gonna take it, if you’re above average, you’ve actually got an advantage over AI, so you should feel confident. So hopefully, at least half of the people listening to this webinar are now feeling a little more relaxed, a little more confident about the situation. So large language models are very important and a chat GPT. And the GPT models that underpin it are probably the most talked about models, in terms of AI today. So one of the issues we’ve had is that open AI has gone from being open as it named suggested. And actually, it was designed to be a nonprofit to being a company for profits, basically driven by the success of chat GPT. So we know how GPT three was trained. And in fact, there’s a table here on the slide that explains exactly where that data came from. So if you look at it around about 85% of the data that was used to change, chat GPT was from the web. So that’s something called common crawl, which is a database of text from the website, from websites online. It’s something called Web text two, which is a smaller database of website information, and Wikipedia. And those three things formed about 85%. And then about 15% of the training data came from books, there were two book datasets, book one and book two. As you can see, whilst AI people create very, very clever neural networks, they’re not great at branding and naming. And this was the GPT three training set. And it allegedly cost around about $12 million in terms of computer time to train GPG, three GPT four, as I say, open AI has now become for profit, they’ve not revealed how they’ve trained GPT four. But Sam Altman, who’s the CEO, has actually said it was probably over $100 million to do that training. And this is very interesting, because anyone who’s used chat GPT and asked for up to date information will have found that chap GPT says, oh, no, I was trained about a year ago. And I don’t know anything that’s more recent. And obviously, any of these neural networks, it’s difficult to keep retraining them, because the cost of doing that is incredibly high. So you’re talking $100 million. That’s compute time.
And obviously, a lot of compute time actually is electricity to power those computers. So it’s not necessarily particularly environmentally friendly. And it’s not cheap to keep retraining. So it’s likely moving forward that what we’ll see is we’ll see a lot of these large language models being trained on a periodic basis, rather than being continually trained. If anyone’s use Bing, and use the chat tool on bing, bing does something slightly different in that it actually runs a web search and feeds the data into the chat into the GPT engine, which then produces the Bing chat output. So we’re seeing basically, what would be if you like the first half of a page of big results actually being presented as pros, rather than being presented as a list of websites. So it’s really using that GPT engine as much as possible to present results, rather than necessarily to go find that information. That’s different from asking Chet GPT a question where it relies on its training. So hopefully, we’ve not gone into too much detail on AI and what’s happening with AI. Let’s look at how it can be used in marketing?
Well, the first thing is, is there’s been a lot of issues with using AI. And whether that’s Amazon realising that they took all their prejudices, when they were hiring, and programme them into the AI they used as a recruiting tool, all the way through to CNET, using AI to generate articles on personal finance, and getting a lot of plagiarism identified in that article, or in those articles. So there are problems with using AI. And a lot of people have tried to use AI and failed. But what I’d like to try and do is I’d like to try and investigate whether you can spot AI. And so what we do is we’ve got a n, two images of the dog here, and I’ve popped up a poll. So you should see that in your right hand side. So please go to the poll on the right hand side and tell us which one is real. One of the dogs is real. And one of the dogs is generated by AI. And we’ll run this for about another 10 seconds or so if you can put your your votes in.
So thank you very much everyone. This this is very interesting, because the one thing we can say is that people are certainly struggling to work out which dog is real and which dog isn’t. So 45% think the dog on the left is real. And 55% think that dog on the right is real? Well, the good news is, is that the hive mind of everybody on the call has actually identified the right dog. So just 55 to 45% we actually got the answer, right. So that’s good news. But clearly, it’s not obvious which one is real and which one isn’t. We can do a similar thing if we look at text. So what I’m gonna do is I will get a pull up again for the text. If you can just have a quick skim here. There’s text on the left text on the right, so two columns. And one of the columns is real text and one of the columns is AI from chat GPT and I will give everyone about 45 seconds to read the text and try and predict which one is real and which one is AI.
That’s great everyone’s voting, we’ve got about another 15 seconds for people to finish reading and work out, which is which.
This is very interesting, we’re a little bit more confident here. So 58% of people actually has gone to 6040. Now it’s gonna last voting 60% of people think the text on the left is AI. And 40% of people think the one on the right is AI. And I can tell you, I put this search term into Google. And so one of these is actually the top result in Google. And the other one is AI generated. And the good news is, again, we’ve got this right, so 60% of you thought that the left one was AI. And you’ve got that absolute right, the left one was AI. This reflects exactly the results. I’ve done this same test three times now. And generally speaking, it’s about 55 to 60% of the audience, know which one is AI generated, whether it’s the image or the text, but you can see that the quality of AI is actually very good, it can fool a lot of people. To be fair, the quality of the human text in this example, perhaps isn’t the best quality, but it was at the top result. So now we know it can fool people it can work. Are people using AI in marketing? And the answer is absolutely. I mean, for sure. I think anyone who’s been on LinkedIn recently has seen some LinkedIn posts that are maybe less well written or they’d expect, and they’re clearly AI generated. But if you look at it, some people are doing some really good work. So on the left hand side, there’s an example of an agency.
There’s using AI to generate backdrop images. And also, what we’re seeing is particularly around programmatic ads, a lot of focus on AI for generating headlines, and body copy, to make it quicker to generate ads. So we’re definitely seeing people use this both in terms of releasing products, and also using more general products to drive AI. So this is not coming, this is something that’s actually happening now. So let’s have a look at some of the tools you can use. So, chat, GPT, I’m sure most people on the call will have at least seen it. It’s really good. Because you can do an awful lot of things with chat GPT, it’s not great if you want long form or detailed copy. So if you want it to write an essay, it’s not fantastic. It can often be a bit simplistic. You know, if you ask it for an agenda, it’s probably going to have, you know, opening, it’s probably gonna have closing, you know, and a prep two or three bullets that are related to whatever the agenda is for. But it’s not necessarily hugely insightful, but it is great, fun and great for generating ideas. And so we asked him to write 10 Google Ads headlines, promoting a webinar that discusses whether AI will take marketing jobs.
And you can see it’s actually produced, you know, a really good selection of, of headlines there. So I think it’s, you know, clearly able to write good effective headlines, you might not want to use all of them. So future of marketing with AI question mark, probably not the best headline. But certainly in terms of removing writer’s block and speeding up the process, we can see chat GPT is really, really effective. And it’s definitely a tool that can be used for many different content creation projects. I mentioned it can also be fun as well. And so here we have a press release in cockney rhyming slang, promoting a webinar that discusses if AI will take marketing jobs, so Iliev and generate a Cockney style, sort of East London, press release, great fun, not quite sure how you’d ever use it. But again, it can be kind of amusing. It’s also excellent writing songs as well, if you want to get some fun, some song lyrics written. So chat GPT is great. It’s probably not great in terms of writing in depth articles. And if you think about it, what chat GPT is doing is it’s trying to predict the next word. It’s not really thinking in terms of structure of an article. And so what’s happened is there’s a number of different products that are now available that are designed to help people write long form content. And so two of the most well known are writer and Jasper and what they do is they introduced this concept of structure. So, you know, very simply, you can go to the tool you can say, I want to write an article about out whether I will take marketing jobs, initially, the tool will produce a list of bullet points. So the key points you want to make, you can then edit those bullet points. And then what it will do is it will write paragraphs or sections around each bullet. So you basically build a structure by working with the AI tool. And then you produce the content. And this produces much more engaging and interesting content, then you get if you just asked chat GPT to produce an article.
So the question is, do people actually read it, though, so we decided to try it. And what we did was we actually posted a range of content on our blog. So we posted content that was written by experienced writers. So basically, some of our PR pros, we posted content that was written by people who don’t normally write. So typically our design team, and we posted AI articles. And we measured the time on page for each of these articles. And you can see this kind of a grouping. And the orange ones are to the, to the left, or the least time on page. And the red arrows to the right tend to have more. So clearly, the red is typically better quality than the orange. Now, interestingly, if we actually look at what those arrows mean, the inexperienced writers performed worse than the AI generated content. Now, to be fair, we didn’t take raw AI generated content, we use Jasper, it was an interactive process. So we were guiding them the AI. And we also sub edited the content afterwards to make it read better. So in fact, because of the editing and the interaction, it took us almost as long to generate the AI content as it would just write from scratch.
But doing that work makes the AI content more effective, more engaging, than someone like a designer writing and our designers are not, you know, terrible writers. I mean, there’s some good content there. And experienced writers starting from scratch still, on average did better than AI. But you can see that whether an experienced writer starts from scratch, or starts with an AI, it’s actually getting very close. And I think that’s very important. One of the things I would say is these numbers would probably be very different if we just use the raw AI output, because in long form content, there’s always sections that really don’t read well. And the AI models, the large language models are still progressing. So today, it still does need sub editing. And as I say, to do a good job, it takes about as long as writing an article from scratch. So it’s not quite there in terms of accelerating, although if you have writer’s block or don’t know how to start, it’s absolutely fantastic. One thing that is worth mentioning is we left our blogs up for a couple of months before telling anyone they were AI generated, and no one noticed, which I think, you know, certainly, you know, highlights the fact that the quality of the article was pretty decent, it wasn’t a poor quality article. Of course, one of the challenges we have is whether these AIs are actually able to do what they’re doing. And what they’re doing is they’re taking the information they’re given.
And they’re using it to learn and then they’re regurgitating it as outputs. And here you can see, I mean, this is an AI generated image that still has a Getty Images watermark. So this article talks about Getty Images, claiming that stability, AI unlawfully scraped millions of images from their site. And obviously, pulling in copyright images is a breach of copyright. And anyone who’s dealt with Getty will know that Getty is incredibly proactive and protecting its copyright. So there is a huge risk of using this content. And the same thing happened with CNET, where plagiarism was found in the articles that they wrote using AI. So in terms of protecting a brand, I’d say that using AI is potentially a very high risk at the moment, because of the potential to actually include copyright material. It’s stable diffusion, for example, their CEO, was asked whether they used any copyrighted images in their training. And his answer was, well, we’re pretty sure it’s in the hundreds of millions of copyrighted images, but we can’t be certain how many so clearly, you know these these tools are very definitely trained on copyright content. And there is a real potential risk and that this is something I think that companies looking to rush into using AI needs to be very careful about and particularly with images because quite often with things like watermarks, you know it can be easy for able to track that. And I think we’ll see more and more lawsuits around ownership.
Another thing worth mentioning is whether you actually own the content. And this is fascinating because there has been one court ruling in the States. And that first court ruling, which was a lower court, so it’s not necessarily going to stand. It says that if you use AI to generate content, you don’t own the copyright because you didn’t create it. So if you use AI, you can’t claim copyright in the US at this stage. With anything you create, there hasn’t been anything to set precedents in the UK. And I’m sure this is something that’s going to play out in the courts over many, many years. But you know, looking at AI, there are certainly going to be issues around law, and around IP, that are gonna take a long time sought out. But the good thing is, even if we’re not gonna use AI as generative AI, it could do more things than generate images and text. So AI can have an opinion. And this is something we tried at Napier, we wanted to see whether it could predict more effective or less effective headlines. So he asked it, whether get your free sample or free samples will be more effective. And actually, the answer that chap GPT gave was really very good. I mean, it was very clear. It used sensible logic. And it did pick get your free sample, which is more likely to perform better on Google ads, than free samples.
Obviously, it does depend on a lot on context. But, you know, it’s great, you can ask chat GPT and get an opinion on the quality of headlines. You can also take the pain out of data analysis and, you know, people I know, hate these, this sort of stuff, you know, is a click through rate of 1% better than 2%. If the sample sizes are 1000 2000, you want 95% confidence. So this is a statistically significant difference. So are we saying that there’s a 19 out of 20 chance that the click through rates are genuinely different, and 2% is better. And it’s not driven by randomness. So we can ask chat GPT lists, and chat GPT loves this because it can produce lots of lots of content. And it will come out and it will walk you through the process that you need to do for analysis. And you can see it’s got the formula, it’s got the calculations. But most importantly, the last sentence says, with 95% confidence, we can conclude that the click through rate 1% is better than 2%.
So this means that if you’re a B testing, you can actually have you know, a very easy way of testing to see whether your two ads one is better than another. And that’s a really important part of a b testing, because a lot of ads get picked as being better, although in fact is randomness. And you could be going down the wrong routes. So it could do analysis. It can also do things like answer complicated questions. So one of the things you can do is you can get a plugin for chat GPT that sits in Google Docs. And so here we have an extract of a spreadsheet. And this spreadsheet had a bunch of customer names. And we just asked chat GPT to identify which of the following industries communications, industrial automation, security, agricultural or other is served by this company. We then give it the next cell we drag it all down and chat GPT very quickly will categorise the industries. As you can see on line seven, it’s not perfect. It’s categorise something as other because it didn’t know. But basically check GPT has done a very good job of categorising the majority of companies. So if you’re looking for example, to run a campaign, you want to cluster by industry, chat. GPT is a great solution. If you’ve got this Excel, if you’ve got the sorry, Google Docs plugin, Google Sheets, plugin, sorry. And you also are probably using AI anyway in marketing. So whether it’s programming, programmatic advertising, whether it’s text to speech or speech to text, whether it’s editing, spoken content, or video content, whether it’s recommending content on your website, or even chatbots. Typically, AI is now becoming embedded in a lot of tools. And we actually think this is going to be one of the things that’s going to be a big trend.
The interesting thing is the general AI models, the GPT models, they weren’t really well, they weren’t actually probably better than specialist advertising language models. But they can be embedded easily into tools. So they almost disappear into the background. And you’ve kind of got this wizard or helper that’s going to help you use either, you know, Google ads or whatever. and optimise it and make it better. So we see this as being a big trend. And it’s certainly something to watch with any martec tool you’re using, because I’m pretty sure they’re looking to invest in integrating AI. Some other uses of AI one is summarising, you know, in marketing, we all have a lot of content. And Microsoft, if you’re a Microsoft company, they’ve already said that they’re going to produce tools to summarise meetings, and produce different summaries of meetings. So it’s definitely something that that’s happening, a lot of people are using it. And you can actually also have tools that will write summaries of web content. So if you’re writing newsletters, for example, there’s tools like glass that will pull in multiple articles from the internet, summarise them and create your newsletter automatically. Generally speaking, these generative API’s are very good at summarising.
So they do tend to produce pretty reasonable summaries, although again, they can’t be relied on to be perfect. It always needs a proofread. And certainly, you know, if you’ve seen some of the experiments where people have asked for summaries of books on which the AI’s had been trained, then often book summaries are not so good. So definitely summaries are something we think is going to happen. editing video, is another thing that looks like it’s gonna happen. If you look here, this is a tool called Content fries. And what it does is it takes long form videos, and it pulls out highlight clips. So that’s really exciting, really interesting, and could potentially help with repurposing quite significantly. And you can even have artificial people. So this avatar is called Emma. And we’re using her for some outbound marketing. Now, although she is a real person in real life, she never says any of the things. She says in our videos, we literally type a script, and we get a, you know, very realistic human moving and talking with the words that we type in. So AI powered avatars, I think are going to become a very big thing. And I’m pretty sure that you’ll start seeing a lot of videos coming into your inbox, that are AI avatars that look like that they’ve recreated a personalised video.
But in reality, the AI is generated, and it’s all been automatic. And then lastly, I mean, as an agency, we’re always on the lookout for threats. And there are tools that claim to effectively replace agencies. So digital first is an example which claims you can log in and create an execute marketing plans in seconds. We have tested this, I would say it’s a long way from creating and executing marketing plans. With AI. It’s not very creative. It’s very limited in what it can do. But certainly, I’m pretty sure that they’re working hard to find ways to actually help drive campaigns. And I certainly think if we look forward in AI, things like Google search campaigns, increasingly, what we’ll see is a lot of AI acceleration around that. That’s a fairly formulaic approach. And I think what we’ll do is we’ll we’ll see for a lot of campaigns, you know, some of the keyword research being done by AI, and also some of the ad content being generated by AI, at least the first pass. So, in summary, we think current AI technology has its limitations. But it is going to be massively important. If anybody is working in marketing, they should be looking at AI and looking at how they can use it.
If you’re talking about creativity, I mean, just be above average, and you’ll be okay. So, you know, hopefully, we should be able to all manage to maintain our jobs in marketing, by introducing creativity and innovation. But I think, you know, imagining that because AI doesn’t completely replace you, it can be ignored, it’s completely wrong. And using AI to accelerate work is definitely something that needs to be done. Obviously, be careful. And particularly Be careful if you’re taking AI generated content, that you’re not plagiarising either images or text. And today, you know, I’d be really clear on this. Napier is not using AI to generate content. And there’s a couple of reasons for that. Firstly, you know, if you look at our business, we’re not at the content farm kind of level, we’re at the higher quality level. And it’s probably the case that AI is not quite at that level yet.
In terms of is offering the ability to write deeply technical specialist articles for our clients. And I say yet rather than it won’t be, but also, we wouldn’t want to expose our clients to risks of being seen to have taken content from other copyrighted sources. So, you know, please use AI, please make it part of your day. But the good thing is, is I asked Dali, to create an image of a robot holding a sign saying you’re fired. And it didn’t quite manage to make you’re fired. It was honestly the first attempt it came back with. And so I feel that although AI is definitely something that’s going to change our jobs, as of today, it’s still got some limitations. And I think the future is still bright, particularly if you’re creative and innovative in marketing. So thank you very much for listening. I very much appreciate the time you spent. I’d be really interested to know if anyone has any questions. So if anyone’s got any questions, please type them into the chat.
Okay, I’ve just got one question. So the question has been asked whether the Add In for Google Sheets is free, or if it’s a paid for it. And so obviously, chat GPT has a couple of options. Most people know about the fact you can access check GPT for free. And you can pay for an upgrade that gives you access to the new model, the GPT four model, which produces better content. There’s also something called an API the interface for computers to talk to. And this is what the Add In for sheets actually uses. And that API is also paid for. But the cost of using it is pretty small. So unless you’re dealing with huge datasets, it’s a pretty trivial cost.
So it is paid for, but it’s not expensive. And I think that’s something we’re gonna see a lot of with AI is that using AI more and more, I think will become paid. But the reality is, is those AI queries don’t cost a lot to run. And so the actual cost won’t be too big. Well, thank you very much for listening. Everyone. Obviously, if anyone does have a question, and they haven’t. Oh, actually, I’ve got got one other question here. Are writer and Jasper free? Or is there a charge? Both writer and Jasper are companies and they are paid for products? They do have free trials. And I can’t remember exactly what the deal is. I think Reiter has a free tier. And Jasper has a free trial if I’m right. But but don’t quote me on that. But they’re, they’re definitely paid for tools, if you want to use it. And I’ve also got one last question here about whether I’m aware of any ad or campaign done by AI that’s been super successful? And the answer is I’m not, it’s not something we’ve particularly looked for. I’m sure people have done campaigns out there.
I know recently, there was a I think it was a Sony, or it could have been a Canon photography competition that was actually won by an AI generated image. And it was won by a photographer who generated the image, specifically to highlight the potential impact of AI on photography. So I really appreciate everyone listening. I think I’ve probably still got a couple of questions that might be coming in. But what I’ll do is, if anyone’s got any questions that I haven’t covered, please feel free to email me. And what I’ll do is I’ll send an email back and if I do have any coming through on chat afterwards, I’ll send those through as well. Thank you very much for listening. I hope this was useful. And as I say, please send me an email, Mike at Napier b2b And I’d be happy to answer any questions. Thank you.
