With the likes of ChatGPT and Claude, you can build a website, fill it with text and generate images in a pinch. You can plan a marketing strategy, design a logo, create social media posts, and make posters and adverts too. You could even automate customer emails, handle their queries and write staff emails, and you could also write a press release. Basically, you could use AI to do all of your marketing for you.
But just because you can, doesn’t mean you should.
This summer, Linkedin introduced a very well received ‘seems like AI slop’ button for users to report bad AI generated content. This was introduced after it was reported that 41% of all long-form posts were machine written. Within its first few weeks, more than a million users used the option to flag bad posts that had plagued the platform.
At around the same time, the European Union introduced new laws that require businesses that are either in the EU or which advertise to people in the EU to label most content that is AI generated content.
There is certainly an increasing level of policing of AI content (as opposed to there being none at all a few months ago) but there is also a change in the wind regarding how ordinary people view AI. More and more, people mistrust and even resent generated content that is bleeding into all other aspects of marketing.
There’s also the problem that some of what AI generates will be rubbish, and you might not realise it.
So how much should AI be used in marketing, and how much is too much? Read on to find out…
Why the AI marketing shortcut is so tempting
If you’re the owner of a small business and you’re keen to do some marketing and don’t have much room in your budget, you can achieve an awful lot with a few prompts and some trial and error, even on a basic subscription for most AI software.
From a starting point of having no website and no real marketing to speak of, doing something is generally better than doing nothing, even if you are using AI to help you out. AI can help you quickly build up a brand and establish a presence that you might have been lacking before. You could even make a half decent website in under an hour that might be all you really need.
A more complex, less generic and more authentic-feeling website will take a lot more work and a graphic designer or a web developer will still see through your AI efforts straight away, but most people might only get the feeling that things are a little ‘off’. If that’s not a dealbreaker, then AI can be a great way to get your marketing off the ground.
Using AI to create social media posts can also be very quick and very effective. AI has learned from so much social media content, so it knows exactly how to write engaging content. 13.8% of all social media content is now AI generated, and while it performs comparably to human content in terms of comments and engagement, AI-led accounts acquire more followers per post than human-led accounts.
Of course, that’s not the end of what AI can do for your marketing, but the problem is that a lot of it won’t necessarily pass the sniff test and it could do your brand more harm than good.
Can people tell when marketing content is AI generated?
AI generated imagery is hard for most people to detect. A recent survey found that even though most people are confident they can spot AI generated photos, 57% of consumers can’t identify them in practice. Writing, however, is much easier to spot. Try as they might, AI generators have some very clear patterns and habits in the way they write and that means 50% of consumers can correctly identify AI generated copy.
The telltale signs can be pretty obvious. AI loves a predictable structure that’s overly formatted, with too many emojis and bullet points, it also tends to use phrases like ‘game-changer’, ‘look no further’, and posts seem to always end with an ‘inspirational’ conclusion. Something like: ‘it’s not just annoying – it’s the next level of irritating’.
The more time you spend using AI generators, the better you’ll be at spotting when they’ve been used as well. You’ll start seeing those simple cartoon faces with dead eyes, the unnecessary use of sepia tone, the very annoying ‘tech-bro’ way of writing, and you’ll notice when an article is saying absolutely nothing of substance whatsoever.
When it comes to graphic design and logo design, it’s even less convincing. Image generators famously struggle with text and they make designs that are often overly intricate, overly rendered and old fashioned. It’s important to know that you’ll also have no legal ownership of your AI generated content either, because even if you did spend ages tinkering with the prompt, you didn’t make it, so it’s fair game if someone copies it.
Oh, and then there’s also the massive use of almost identical, nonsensical AI generated posters and flyers that are plaguing just about every sector at the moment.
Consumers don’t trust AI generated content
But is it a problem if people can tell what’s AI generated and what’s not?
While it might sometimes be hard for people to detect it, there are many studies that show that once they realise something is AI generated, they’ll turn against it.
53% of UK shoppers say they explicitly mistrust AI generated social content, with 51% saying its use risks eroding their trust in an entire brand. There’s also evidence that this is making people more skeptical and cautious in general because 72% of Britons say they’re now unsure whether any online content can be trusted because it might be AI-generated – and that’s sensible because 74% of new pages are now made with the help of AI.
Meanwhile, American surveys have found that 61% of shoppers say they’re unlikely to engage with content if they think it’s AI, 78% think that AI makes adverts less authentic and the same proportion think that brands that over-use AI are “cringey”. 63% say they would be less likely to purchase from a brand that uses AI ads.
This also plays out in the performance of advertising, with Linkedin ads that are human-made outperforming AI generated ads.
That sounds pretty damning, but I’d still say that if AI is the only way you can get your marketing off the ground, then it might be worth the gamble.
One thing you really shouldn’t do is use AI to write press releases for you. Trust is a very important commodity in the media, and a lot of publications absolutely will not publish AI generated articles for that reason, so it could risk your long term relationship with a publication if you trick them into publishing your GPT masterpiece. The same goes for AI generated imagery unless it’s clearly labelled as such and it’s the only possible image you have to offer. Every mistake or AI-ism will also affect their interest in your content, because most good publications choose their content based purely on merit.
And, of course, there’s the law to contend with. The EU has introduced its new requirement for lifelike AI generated imagery to be properly labelled, but the UK also has the Advertising Standards Authority and Trading Standards to contend with. These organisations don’t care if you wrote it or your AI generator did, if your marketing material or website says things that aren’t true about your product or if you’re passing off AI generated as real photos, you could get in serious trouble. That’s a real concern because AI does have a tendency to hallucinate, exaggerate or be a little careless with the facts.
So, while using AI generated content might be quick and easy, its practical use in marketing is somewhat limited.
AI and SEO
There’s another problem with using AI to create and populate a website with content; it might affect how easily people can find your brand on search engines.
You might have heard that Claude recently began embedding imperceptible, machine-readable watermarks into all of its generated text, along with metadata that describes how it was made. In future, that’s likely to become a more common and important factor, but did you know that Google is already detecting and penalising AI generated content in search results?
To be clear, that doesn’t mean that all non-human content outright is being hidden from search, it’s just the most egregious examples that they’re trying to filter out. Most of the ways they are detecting AI content won’t be a problem unless you’re really going overboard but Google’s E-E-A-T quality scoring system (Experience, Expertise, Authoritativeness, and Trustworthiness) is worth being aware of. This is important because it’s a measure of how useful and usable a website actually is, so if your website is a bit sloppy, it might be missing out on search traffic.
Where AI excels in marketing
Ok, so we’ve covered the downsides of using AI but it absolutely does have its place. So let me explain how our PR team uses AI here at Carrington.
We don’t use AI to do our writing for us, but we do use it to critique our work and suggest ways to improve it. We might do this during the writing process and we also use it at the end as a check after an article has been proofed by two people already. We might also use it to search for data and evidence when writing articles that need to be grounded in fact, or we might ask it to teach us new subjects when we’re writing content for other professions, although our writing is always checked by the actual experts before it’s published.
After we’ve brainstormed and set out our own ideas for a campaign, we might ask AI what it suggests in case we’ve missed anything obvious so we don’t miss any low-hanging fruit.
Similarly, we might also use it to analyse data which could make potential news stories. Because we don’t fully trust it, we do this after we’ve been through the data ourselves and we make sure to double check its suggestions to make sure the data does stack up, but sometimes it can help spot things that we miss.
We don’t like using AI generated images if we can avoid it, but we sometimes also use them if we can’t find suitable stock photos or if a client doesn’t have the budget to buy stock images. If we use them this way, we make sure they’re labelled and we don’t simply use them as they are, but we mostly use them as background images with additional graphic elements over the top. We absolutely do not use AI generated images to accompany press releases unless there’s a good reason, such as if we’re sharing an impression of how something could look and the budget isn’t available to hire an artist.
We’ve also used AI to create systems that complement our existing PR coverage checking processes. This includes automated email updates for newly published articles that mention our clients so we can share these with our clients and stay on top of how they’re being talked about.
The way we use AI focuses on doing the work that doesn’t warrant our skills, talents and creativity. We get it to do the dry, mundane background work that can make us more effective and more efficient. That might not be quite as exciting as letting it rip on an entire strategy, but we get much better results this way and we can take much more pride in our work.
Is it better to stick to human marketing?
With AI becoming increasingly widespread and the public becoming increasingly wary of it, good old fashioned human beings might be seeing something of a comeback: In particular, there’s a real push in marketing for more authenticity in content because people are seeking it out more and more.
AI generated content can certainly help you out in a pinch – say, if you need to get a basic website up and running in a couple of days – but it shouldn’t be seen as a long term solution and it should be used sparingly.
For this reason, and those outlined above, we’re not jumping fully on the AI bandwagon – it’s more like we’re skitching along when it suits us. AI undoubtedly has its place in a behind the scenes capacity, and, for the most part, we will continue to use it in this way only.
Afterall, recent polling shows that 43% of shoppers say the rise of AI has made authentic human-made content more important to them, with that figure rising to 62% of Gen Z shoppers. Meanwhile, 81% of people say they value human-made content more than AI content, preferring its authenticity, originality and appreciating the connection between them and the creator’s own experience. In addition, 69% of people say they trust human content more than AI content.
All of that shows that if you want marketing done right, it’s always best to use real people…. like us!
