Augmented Potential: What Does Generative AI Mean for Marketers?

Marketeers have been working with AI for longer than most - but can AI-generated copy and images really help the human effort?

Will Freeman

Anybody with a few years in marketing behind them has likely already worked in partnership with artificial intelligence.

Employing AI as a tool for analysis and reporting is a well established approach in marketing, and has proved a tremendously useful application of the technology, taking the legwork out of trawling through vast data pools. But AI is increasingly able to do much more for marketers, generating consumer-facing copy and highly tailored images that precisely target given audiences. That opportunity begs an important question: do we really want to hand the creative part of the marketer’s craft over to machines?

Certainly, it’s an idea that not everybody will feel immediately comfortable with. Us humans can host something of an instinctual uncertainty around letting AI take over ever more of the work that we do; work that so often defines our personal sense of self-worth.

It’s likely somewhat to do with decades of science fiction films that posit that humans will be dashed aside by artificial intelligence, as machines created to serve us become our masters. And, of course, for anybody who has lost even a single Word document to gremlins in a laptop, it can be a little tricky to entirely trust a computer.

However, AI that generates convincing, engaging and readable copy is very much becoming a reality, as evidenced by a quite remarkable article written for The Guardian by OpenAI’s GPT-3 language generator. The AI author was asked to convince readers that robots indeed come in peace; and it did so with surprising aplomb.


The idea of AI that creates copy is that it can do so through machine learning. While ‘artificial intelligence’ is an umbrella term that covers the concept of computers thinking independently in general, ‘machine learning’ (ML) describes an ability to improve and adjust abilities through experience. Google’s striking Deepmind project, meanwhile, saw computers with ML abilities observe the world’s best esports pros playing StarCraft II, before eventually going on to beat them at their own game; more of a challenge to a machine than beating a human at chess, remarkably.

ML is essentially a specific type of AI. ‘Generative adversarial networks’ or ‘GANs’, meanwhile, are a class of ML that lets computers create new data based on that which they have observed. For example fashion retailer Zalando experiments with GAN's to render outfits individually created by their customer.

So after viewing millions of marketing images for different games, a GAN could then begin to create images for a specific game, using its gained experience and some provided assets. ‘Generative AI’ is increasingly used as a catch-all term to cover forms of AI that can create copy, images, assets and other entities itself.


GANs have improved considerably since they emerged from academia in 2014, to the point that they can now write readable copy for a national newspaper. But why would a marketer need a computer that has learned to write copy, or build custom marketing assets?

Ultimately, it comes down to efficiency and scale. While you may be able to write nuanced, impactful marketing copy for emails, editorials, ads and social media, generative AI might create hundreds of different versions of the same copy, each highly tailored and targeted for a particular platform, publication, or even individual consumer. That copy could be extremely optimised for SEO, and informed by datasets that could take a real human a lifetime to digest.

Equally, generative AI can constantly analyse, edit and even remove older existing content. In such cases machines can monitor and maintain legacy libraries of published and new copy to make sure contradictory or dated information is eradicated, while marketing messaging is kept contemporary and harmonious, taking into account the likes of updated marketing plans and the context of global events.

And you can even apply generative AI to copy written by your own hand, letting a machine learning analysis refine your work at a granular level to generate more click throughs and other defined actions, or even refining internal copy and the impact it has on your marketing planning.

Additionally, using GANs to build highly realistic, highly targeted images for you campaigns brings powerful efficiency, letting you save on the time and budget needed to create images from scratch, or source ideal and appropriate existing materials.

Through all these examples, it’s worth noting an important point. We are nowhere near the day where computers can do all this entirely unaided. Generative AI - and AI broadly - offers a means to augment human abilities; not replace the need for a human.

In the case of The Guardian article, an editor had to set out a highly precise brief and examples of how the robot might write. On producing eight articles, highlights were selected, combined and tweaked.. In fact, in analysing the process, the newspaper team concluded that working with GPT-3 took at least as much human effort as seen when receiving copy from a homosapien contributor.

Through all the examples above where generative AI may help the marketer. Humans will always be needed. The idea is not that machine learning means you can down your writing tools forever. Rather, computers can handle a significant majority of the process; analysing data, tracking shifts in audiences, optimising precisely for click throughs, and creating first drafts. All being well, that should leave you with more time for the actual craft of marketing


As we’ve seen, AI can be powerful. So powerful, in fact, that there are challenges around ethics. If machines can generate convincing copy and highly realistic images, the potential for erroneous, false and deliberately misleading information is considerable. Deep fakes are a common concern, and rightly so, but in the realm of marketing trust specifically may become a key issue around the use of AI. Marketing broadly is about installing consumer trust in brands. The layperson chooses a brand they trust over one they don’t, and marketing endeavours to build and maintain that trust. As such, marketers need to be thoughtful and considered with regard to how the misuse of AI can undermine trust. Because if brand trust is damaged, an otherwise valid marketing effort could all be for nothing.

Certainly, using generative AI comes with great responsibility. But it may be close to inevitable that human marketers will see even the most creative parts of their work augmented by machine learning.

Here at actioncy., our agile marketing planning and delivery platform absolutely serves to reduce process and give marketeers more time and headspace to focus on the human craft of their work - all while ramping up efficiency and impact. As such, we see the rise of generative AI for marketers as a powerful opportunity that strikes a chord with our goals. So do get in touch if you’re keen to discuss the opportunities in combining generative AI and marketing craft - or if you want to learn more about actioncy.’s research in the area.

Marketing has always been a role that necessitates working with the cutting edge of technology, and most of you will have been doing that for years. Equally, marketing is often at its best when highly collaborative.

It may just be, however, that in the near future you will be collaborating more closely with machines, empowered by a better toolbox that lets you deliver better results.