← The Digital Marketing Podcast
The Digital Marketing Podcast
Generative AI Update - What the Latest Model Wars Mean for Marketers
00:16:31
Loading summary
Transcript1 lines
- [00:00]
A
Welcome back to the Digital Marketing Podcast brought to you by targetinternet.com My name is Daniel Rolls and in this episode I'm going to take you through the latest generative AI news and updates. Just before we get going, just a quick message if you fancy spending half a day with me every month. What a joy. A Target Internet membership is currently half price at just 30, 30 pounds a month and you get our half day master classes and we have got a masterclass coming up on the latest trends in digital marketing. We are talking about all the latest AI updates. We're looking at AI agents hand on, we look at vibe coding and a load of other stuff as well but at the moment you get those with the 30 day free trial as well and you can cancel anytime. So really worth signing up. It'll be lovely to meet more of you. You've already got a great community so it'd be great to meet you and hopefully see you on the next masterclass. So let's get into it and the end of 2025 we saw a flurry of AI updates and changes and a bit of heating up of the AI model war, particularly between OpenAI and ChatGPT and Google Gemini. So it's kind of worth starting there. OpenAI released ChatGPT 5 and actually that was adopted into Copilot. So something Copilot leaped forward as well. And it seemed like it was kind of really racing ahead of Google Gemini and ChatGPT5 was smarter and it was a lot more autonomous and it just kind of seemed smart all around. And then suddenly Gemini 3 was released and in terms of writing and logical responses, it kind of leapt ahead of ChatGPT. So it was really good at work out what you're asking for. And it seemed a lot more reasoned than it had done before. But then suddenly a big story appeared in the press about Sam Altman from OpenAI reacting to that and releasing a code red and saying, right, we need to stop all non essential projects and we need to kind of leap ahead again of Google Gemini. Now I should say at the same time he did come onto Twitter and congratulate Google on what a great model Gemini 3 is. Then within, within days in the following weeks we saw ChatGPT 5.1 and then 5.2 and the big leap for 5.2 is ability to do real world tasks kind of more effectively. So if you've ever got ChatGPT to kind of create a project plan in PowerPoint or create a spreadsheet, it was kind of reasonably limited. Suddenly 5.2 is massively better at doing those kind of things. So in reality, now, what it means is that depending on the task, you might actually select a different large language model for that particular task. Now, I haven't mentioned Claude here as well, but Claude has also taken a leap forward in terms of coding as well. But for my use case, for example, generally I'm still using ChatGPT for most things, and that's, that's mainly because I've created lots of custom GPTs. I can connect it to lots of different tools, which we'll come back to in a moment, because that's taking a leap forward as well. But for coding, Gemini feels like it's just miles ahead as well. So we'll, we'll come back to that. That coding piece to me, and actually the interface for making that coding accessible to marketers, to entrepreneurs and so on is a bit of a different thing. Over the same time period, we saw the release of Nano Banana Pro. So that was Google's image generation tool, and they had nanobana already, which was already pretty impressive. But Nano Banana Pro has just kind of really leapt forward. And then we had the release of new ChatGPT images. And in both cases, what we really saw is a leap forward in the ability to edit existing images and particularly photographs. But the photorealism of what was being created as well. Now, I have to say Gemini is. I feel generally better at the moment of photorealism, but they're both incredible. The big change really is now that ChatGPT can edit and recreate photos. So previously, if I taken a picture of me and I'd said, put this on me on a beach. Same person, but just put me on a beach, it was kind of prevented from doing that due to a safety restriction. And the safety restriction, you couldn't take people and put them in new situations. It would recreate them, but it wouldn't quite be them. The faces would change. Now, the, the big shift is you can now do that. Now, if you look at why Google did it in the first place, it's. If you go into Google Ads, there is a tool there so you can take a product and you can put that product in any scenario. So you could take a Target Internet mug and I'll upload the example of this to the show notes targetinternet.com forward/podcast, and I can then take that item. It will remove the background for me and it'll put it in loads of different situations. And you can do that at scale. Now, incidentally, if you Want to do that? Go to Google Ads. You click on Tools on the menu, you can access the ability to add images and you can download those images. You can also create videos and things in there as well. The big change for Google with nanobanana Pro is extreme photorealism, like lighting and subtlety is incredibly better, but it's also really caught up in be able to create text effectively. So it was a little bit, it would create kind of hieroglyphicy things sometimes as well. So it's much better at doing that. But it also now does something that ChatGPT has done for a while, which is using source images to create on brand images. So if you're not familiar, you can do this in both platforms. Now you upload a whole sequence of images, you say create me something like an infographic explaining something. You define what it is, but use these images as source styling and it will copy the colors, the fonts, the spacing, the layout, those kind of things as well. Another big change for ChatGPT is that connectors are gone and they're now called apps. And the key thing here is, first of all there's a lot more of them and they can do a lot more. So you've got a massive list of tools now you can connect ChatGPT up to. And that is one of the limitations of Google Gemini is that Gemini is kind of locked down a lot more to the Google platforms. Now in a lot of large corporations that's a, that's a good thing because they don't want it to be connected up to lots of different things. Within ChatGPT, you can set this at an administrator level, so you don't need to give people access to it if you're doing it as part of a large team. But for example, there's a Canva plugin. But instead now me just asking canva to list, when I access the projects, what are my latest projects and things like that. I can actually say, okay, create me a design using my brand guidelines and it'll go on and it'll give me, for example, four designs and I can select from those and I can refine them, I can update them. And it's not just canva, it's Adobe Photoshop, it's Adobe Express and various others as well. So suddenly I'm interacting with apps like canva and Figma and lots of others as well from within ChatGPT itself and doing a lot of the work there as well. Now I mentioned this earlier on, but really for me, the big leap for Gemini is its coding ability. Not just that it can code well, but the logic in working out what I want it to build from my prompt is really incredible. And the code is noticeably more reliable than ChatGPT from my experience currently. The big thing really though, from a kind of marketer entrepreneur's point of view is the, is the coding interface. So both tools have something called Canvas Mode, which I will come back to in a moment. But Gemini has a little selector tool where it will get it to write some code. It will then preview what that code, like the interface that whatever you've built as well. And then you could just highlight a little bit of that page and you could say, oh, change this text to this or redesign this button or lay this out differently or add this functionality here. If you want to see a couple of interactive pages we've done recently, go to the show notes targetinternet.com podcast and you will see those in there. And we've managed to create these interactive pages, tutorials that are really completely on brand. And that what it created the first time was okay, but by tweaking it again and again with my brand manager, we got it to the stage we're really happy with and then we can take that code and use it again in the future. So the idea of Vibe coding going backwards and forwards that we covered in a recent episode, really, the interface in Gemini really helps you with that. Now bear in mind at the moment what I can do is say create me a one page app in HTML, maybe build in the cascading style sheets, the CSS and then maybe use something like React. And this is all stuff that can run in a browser with no backend. So what does that mean? It means you could copy and paste that code into your content management system if your content management systems allows you to do that. And you could deploy that relatively easily. And we're doing that across target Internet a fair bit. Now that essentially means that you creating interactive content as a marketer is just a whole lot more accessible if you can liaise with your tech team to actually deploy that stuff. A stage further on from this is that Gemini now allows you to add AI features to whatever you're creating. Okay, so give you an example. Fairly easy with any of these tools at moment to create some interactive content with charts or animations and things like that. But actually now I could create the interface for a tool that creates a Persona and I could fill in some basic details and then I can add an AI button. I hit the button and it fills in the rest of the Persona for me. Or I could go through and say, right, I'm going to give you a topic, a target audience, a number of channels. I want you to build me a marketing strategy. Now, deploying these kind of tools is more complex because you've got to include calls to the API that's in Gemini within a website. We're working on doing that at the moment. But there's nothing stopping you creating apps within Gemini that you can just use for yourself. So you could build yourself a strategy builder with beautiful interface. So the idea being that at the moment you can in ChatGPT you can create custom GPTs. Create a GPT that's a specific thing for you. In Google Gemini they're called gems. Within Copilot, they're calling them agents. It's all pretty much the same thing. The difference here is that what I'm doing is creating an interface for a tool. It's not just text. I can actually kind of fill in some fields and things like that and get it to something for me. Now, if you're interested, we have a live one hour update session that's free to newsletter subscribers on Vibe Codings do exactly this and we're going to build something out and then for members we've got half day Masterclass and remember, it's only 30 quid a month at the moment and there is a free trial for 30 days as well. So if you're not signed up the newsletter, targetinternet.com newsletter you get that one hour session and for 30 pounds a month come and get signed up to Target Internet and spent half a day with me building some of this stuff live as well. Now I want to come back to that point about Canvas and it's a little missed kind of feature. So Canvas, if you go to both ChatGPT and Gemini, there's a little plus sign on ChatGPT you have to go to Tools. In Gemini you just kind of go through the list and essentially what people use it for is kind of interactively writing something. So when you're writing an article, you could just highlight a section of the article and get it to rewrite that bit, which is a nice interface for editing content. But actually what I've seen people miss a lot and I've done a lot of training sessions where I'm doing something and people are going, well, how, how have you done that? Have you got a preview? And what I mean is if you get any of these tools to write your code, it'll write the code out and you're like, well that's great, but I've now got a cut and paste that out. I've probably got to create an HTML file, I've got to drop it into that and I've got to preview it. Okay, if you have Canvas mode switched on, you will get a little preview button and that will immediately show you what that code when it's running is going to look like as well. So it's a real game changer because it makes it way more accessible to go through and create code and then preview it. So the idea I was speaking about a moment ago, you go to Google Gemini, you get it to create a tool that would be useful to you. You build AI into that and there is a button just to add Gemini, Gemini AI to it. And then if you preview you can use that tool. So it's a really nice technique for building things for yourself. And then potentially you could deploy that stuff if you wanted to as well, but obviously that's a bit more complicated. Okay, last small piece to talk about in this quick update session. ChatGPT Atlas, that was ChatGPT's browser, was released for Macs. Now it's only Mac at the moment and it's only Apple silicon powered Macs as some of the latest Macs as well. It's been around since mid October though and we've been playing with it loads. Just to say there is a PC version coming, we think late January or February. There's no update. But don't forget you can achieve very similar things. Except the agent mode stuff I'm about to talk about with Microsoft Edge, which has got Copilot built into it as well. And Copilot is now running the very latest version of ChatGPT which is 5.2 at the moment anyway, so you can get the same kind of results. But what's really exciting about this is that I still use Google Chrome as my standard browser, but where Atlas is really excelling is comparing and contrasting pages I have open. So look at these five web pages that are all selling maybe a similar product and create me a new page that's better than all of them. Or run analysis of all these open reports and create me a summary report. So imagine you go through, you open up your web analytics, your email analytics, your social analytics, your ads. You open each of the tabs and you create a prompt that you run once a month to go through and compile a standard report from all of those things in one place. So suddenly it becomes way easier to understand data, to pull insights out of that data, but doing a quick, efficient way. Analyze this page from an SEO point of view. Analyze this page from a conversion rate optimization point of view as well. But what's really interesting is anything you can open if you put it into agent mode, you can then edit it. So if you open a spreadsheet in Excel in Office, you can then put it to agent mode and it can edit that spreadsheet. I had great fun making it argue with Gemini. So I open up Gemini Inside Atlas, open us ChatGPT in the little side tab and then got it to interact and have a conversation about which of them was better and it went on for some time. So basically if you can open a window, you can then get it to interact with that as well. So lots of opportunity for workflow improvement. But I think that reporting piece is probably one of the most important. And we're going to have an upcoming podcast all about creating automated reporting as well. So some tips there, some updates on what's happening with the major large language models and what it means to us as marketers and entrepreneurs as well. Keep listening to the Digital Marketing Podcast. We'll have more of these regular short update sessions coming up and thank you so much for listening. For more Episodes resources to leave a review or to get in contact, go to targetinternet.com forward/podcast.