One of Australia’s hottest AI startups, Springboards, has issued a warning to marketers: relying on AI tools throughout the creative process is a race to ‘average’ and ineffective advertising.
Although generative tools are accelerating and advancing at pace, the creative standard is not. The industry does not need faster shortcuts, but rather stronger ideas supported by tools built specifically to elevate the work.
To illustrate the point, Springboards created a parody ad based on a well known OpenAI spot that shows how quickly large language models can generate work that appears finished, but frequently crosses copyright lines and collapses into familiar patterns.
In the ad, a brother and sister are driving on a scenic country road in the US after prompting an LLM to create a holiday itinerary. Springboards, teaming up with an AI production team, played around with ways to show the risks when everyone is sent to the same place.
The work that came back made it crystal clear just how easily AI can blur creative lines. Springboards dug deeper into themes like imitation, IP and how fast things can feel off, even with the best intentions.
“We thought it would be funny to then take a very well known spot and show what actually happens when everyone gets the same answer and everyone ends up in the same spot,” Springboards co-founder Pip Bingemann told B&T (see Springboard’s version above, and the original ad below).
“If everyone gets the same answers, you’re going to end up in the same spot. So rather than ending up on this road trip where it’s a nice little trip, a little unique in the spot, and they’ve ended up in a massive traffic jam.”
Founded by Pip Bingemann, Amy Tucker and Kieran Browne, Springboards is used by more than 200 agencies and companies worldwide and was built to help creative teams explore a broader range of ideas without sacrificing the craft, judgement and originality essential to great work.
Bingemann describes his business as a ‘self loathing AI company’.
“Most AI companies in the AI creative space try to build tools to make execution faster and cheaper, and we just don’t believe that. We believe that good creativity needs to involve people,” he explained.
“We think that is a very dangerous thing when everyone defaults and thinks they’re getting one on one personal advice from these machines, but they’re not. They’re just getting the most average probable answer that it’s given to everyone. And so for creativity, that’s a bad thing.”
To prove the point, there is a simple experiment Bingemann recommends. Ask any LLM to choose a random number between 1 and 10 and you will receive the number ‘7’.
B&T tried this on ChatGPT, Claude, Google Gemini and Meta’s AI tool; each responded with ‘7’.
The false promise of AI
Bingemann said that relying on AI tools to create ads becomes a problem because these tools often produce average answers.
“The most disturbing thing is when you remove human judgment from that entire process, and often that’s where we see things go wrong. It suddenly feels like judgment at every stage of the process has been outsourced to a machine, from the start of the idea to the production of the idea, to what the thing actually looks like,” he said.
“The scary thing that we quickly discovered when we produced that spot, is that it was frame by frame perfect in trying to replicate the original ad as a parody, which obviously jumps into issues of IP infringement and the rest of it.”
Bingemann understands why marketers, under increasing budgetary pressure, would be tempted to use AI technology to cut costs on production, but warns it’s a fool’s errand.
“Everyone is being sold this false promise of AI productivity, and so they’ve been told to cut budgets,” he said. “High production values are easy to replicate at speed — everything looks better, more polished, more finished and ‘good enough’, but quietly we are losing originality and the subtle nuance only humans can really protect and detect. Judgement gets blindsided. When speed becomes the goal, the question stops being ‘is this a good idea?’ and becomes ‘is this done?’. And those are very different things.
“What people forget is that, like the cost of advertising isn’t just the cost to make the spot, it is also the effectiveness of the advertising. So it doesn’t matter if you save $100 or $1,000 or $1 million dollars producing your spot, if it is not effective, the cost doesn’t matter.
“I would rather spend a million dollars on a spot that brings in $5 million than spend $10,000 that brings in a spot that brings in $50,000.”
AI ‘won’t take your job’
Springboards recommends using AI tools for inspiration and to carry out repetitive low value tasks in the media planning and buying process.
“The way that we build AI tools is to focus on variation and use it as an inspiration tool, but then let good people do what they’re best at, which is to come up with original, creative ideas where they can feel stuff,” Bingemann said.
“Creative teams need the same thing as what they did 10, 20 or 30 years ago. A tight brief with an interesting insight, hook, observation or piece of data. The problem with most AI is that it gives you average, and average doesn’t inspire. That’s kinda why we built Springboards. Not to try and jump straight to execution, but to spark ideas in people’s heads by providing a different POV or jump-off point.”
He said that creatives and others in the industry should not fear that AI will take their jobs; AI tools are just another “productivity hack”.
“The printing press came along and people kept working Monday to Friday. The Internet came along, they kept working Monday to Friday,” he said. This is just another productivity tool. We just do things differently and more efficiently, and we do more stuff and more interesting stuff.”
He said that AI tools like Springboards should spark ideas, not execute them, and that creative teams need to be given the variation and space to unlock new creative directions while keeping the taste, judgement and originality human.
“Going through this process was eye opening; these dangers are real and people need to start talking about them,” he said.
“You can see how easily these things drift into the average and if you don’t have human judgment pulling it back into line and staying true to what the creative idea is about, that’s when you get in trouble.”
Bingemann’s good examples of AI in advertising
Heinz, ‘A.I. Ketchup’
This Heinz ad uses AI to reinforce its dominant position in market.
Amazon, ‘Five Star Theatre: Penguin PJs’
AI is to sift through thousands of customer reviews on the platform and surface the reviews and help find the most heartfelt and hilarious ones for this Amazon spot.
Bingemann’s bad examples of AI in advertising
Google ‘Dear Sydney’
Google has used AI as a replacement for the human part.
Coca-Cola, ‘Holidays are coming’
I hate this ad and the 2024 version as well. The irony of the tagline ‘real magic’ – not sure if that is on purpose but a brand as big as Coca Cola don’t need AI to produce great spots. They’ve done wonderful spots throughout history and just seems lazy – as one YouTube commenter put it ‘Nothing says Christmas quite like “dad got fired because AI replaced him to do animation’.
Read next: ‘Are You Pranking Us?’ Jeanswest Ridiculed Online Over ‘AI Monstrosity’
