There is a glimmer of good news for all those bright, talented creatives in the advertising industry who think artificial intelligence is going to steal their jobs.
A study by Australia’s Monash University on behalf of the advertising agency TBWA has found that human creative concepts always outperform generative AI creations.
Does that mean that Liz from Kerikeri should not have bothered handing over her biometric data to become the cloned frontwoman of those Skinny mobile tv commercials that proudly proclaim to be made with the help of AI? No, because the irritatingly obvious creations were the brainchild of real people and the use of AI was trumpeted by Skinny and its parent company Spark as a clever marketing ploy.
The Australian study was aimed at testing whether artificial intelligence could, in fact, replace creatives in dreaming up advertising ideas and slogans. The finding: AI consistently fails at creativity.
The Monash researchers took 1000 creative advertising campaigns and fed them into large language models (LLMs) that were first asked to strip the advertising messages down to single sentences. The cunning academics then fed the single sentence into the AI machines and asked them to create advertising campaigns.
TBWA Sydney chief creative officer Matt Keon told The Australian that the LLMs had quickly removed all of the creative elements while reducing the campaigns to single sentences and then consistently failed to produce such elements in their recreations of the campaigns.
“What we found is as it started to rebuild, it used more flowery language a more descriptive language, but it left out the things we would all consider creative: Things like humour, metaphor, lateral thinking. It was very consistent across many generations of this experiment, and the top line is that they fail at creativity. They regress to the mean, pretty much, conclusively and consistently.”
The findings did not come as a surprise to me, but it was useful to see the challenge put to a scientifically based test. Why wasn’t I surprised? I had recalled a piece written early this year by a former chairman of leading international agency Ogilvy & Mather. Kenneth Roman writes a column called Madison Avenue Insights and in it he asked two AI engines to tell him what they could and could not do.
The first told him: AI can assist, enhance, and perhaps even inspire, it will never replace the soul of creativity. At the heart of every campaign there is not a circuit but a heartbeat.
The second responded: AI is a tool – albeit a powerful one – but creativity is a uniquely human trait, born from lived experiences, emotions, and a nuanced understanding of culture. A machine can analyse data, recognize patterns, and even mimic style, but it lacks the very essence that makes creative work resonate: authenticity.
Even AI knows AI fails at creativity.
The people who haven’t woken up to that fact are the business owners and brand managers who think they pay too much to advertise their goods and services, and are trying to find ways to reduce their marketing costs. To them, AI sounds like mana from heaven. It offers an opportunity to remove a whole (expensive) layer from the advertising campaign – the work of the creative team. Just go to a generative AI site, describe your product and ask for a slogan, then pass it to the advertising agency for action.
I put it to the test and asked ChatGPT to provide an advertising slogan for the journalism commentary website Knightly Views. It responded with “Knightly Views: Cutting Through the Noise, One Insight at a Time.” Wow, I thought. Wow! Positively brimming with creativity and deeply symbolic metaphors evoking chivalry, battle, armour, champions and the whole valiant caboodle. Just joking: It scored a miserable fail grade.
I like an expression Kenneth Roman used in his column (in which he recounted some of the lateral thinking applied to campaigns by advertising geniuses like David Ogilvy). The word was fingerspitzen (technically, finerspitzenefühl) meaning a feeling or sensitivity in the fingertips. Its metaphorical application to advertising is what sets human creatives apart from plodding AI would-be usurpers.
Fellow Ogilvy alumni and its North American Head of Innovation Kaare Wesnaes has put AI in proper perspective. He says it isn’t here to replace creativity but to rewire it. It not only plays a role in creative execution but, from the outset, it can help to challenge assumptions and overcome some of the less helpful human traits such as confirmation bias.
In other words, artificial intelligence is a tool not a solution. Or, more accurately, a kit of tools. However, even those tools need to be handled with care.
In August, the Interactive Advertising Bureau (a global network of 45 organisations including New Zealand) warned that, while adoption of AI was accelerating, safeguards were not. It carried out a US survey that found 70 per cent of marketers have encountered AI-related incidents in their advertising efforts – including hallucinations and bias – yet less than 35 per cent planned to increase investment in AI governance or brand integrity in the next year. The survey findings are set out here
The IAB study concluded: : “The data is clear: AI is undeniably transforming advertising but incidents are already happening, current safeguards aren’t keeping pace, and marketers need better solutions. This isn’t a future problem to solve but instead a present reality demanding immediate action to unlock AI’s full potential. With a few practical steps, responsible AI is not only possible – it can be the norm.”
Responsible use of AI is not only in the best long-term interests of the advertising industry, it also makes good business sense for their clients.
The research company Kantar has found that high quality creative marketing can drive 4.7 times higher profit.
However, customers are not fools. They react negatively to falsehoods, and AI fakery creates a minefield for both creatives and marketers.
A study published in the journal Internet Research looked at consumer awareness in AI-generated advertisements related to corporate social responsibility. It found AI-generation was linked to perceptions of insincerity when the imagery was seen to be false. Consumers perceive AI generated advertising as less sincere. Assigning the promotions of initiatives to machines, such as AI, instead of humans could potentially yield negative consequences.
Logic, the public interest, and plain old common sense tell us that it is vital to maintain the creative human element in an industry that markets goods and services to the public. However, logic, the public interest, and plain old common sense count for nothing in the face of Mark Zuckerberg and his Meta megalith.
The owner of Facebook and Instagram has announced new services that will enable advertisers to fully create their own advertisements and target campaigns using artificial intelligence tools. The services will be in place by the end of next year. They have the potential to remove not only human creatives but entire advertising agencies from the equation.
Cheap? Yes. Nasty? Almost certainly. Effective? That will, I have no doubt, depend on who is doing the measuring.
The cloning of Liz in the Skinny commercials is deliberately amateurish. Spark’s advertising agency could have made the process undetectable but it wants to get your attention by leaving you in no doubt it is playing with the new teck-er-nolurgy. When Meta puts AI in the hands of true amateurs the resulting advertisements will range from laughable to regulation busting.
If it is allowed to happen, the only winner will be Meta. The victims will be the creatives whose uniquely human talents have made advertising such a powerful medium. But the ultimate losers will be the poor suckers on the receiving end of the latter-day equivalent of snake-oil ads.
The smart money will be on paying humans to create advertisements. These will be the ones that rise above a mess of messages with all the appeal of a fake cat video.
Advertising…mmm. Boring. What’s happening in journalism? Are reporters being replaced more as we speak (I’m sure you’ll have looked at that in the past – but the AI world learns and moves fast. It’s still a bit like Wikipedia in its early days (Terry Brown at RNZ forbad his reporters from using it.
You might find it boring, Mr Tucker, but it paid a large part of your wages for a very, very long time. And fear not: The use of AI in journalism is firmly on my radar.
I use it often. But the accuracy is problematic. Last week I asked it half a dozen times to draw me a simple graphic/map of the New Plymouth coastline showing the pollution gradings for the half dozen rivers emptying into the sea. It got them horrendously wrong every time, on one occasion even swapping the sea for the land. In the end I did a screen grab of the Taranaki Regional Council one. I’ll be writing a column about it myself someday soon.