The advent of Generative AI is ushering in a revolution transforming the creative arts in ways few expected. Who would have thought that AI could write better than we do, compose music that moves us – and now, create ads as good as those for power brands on national television?
That day is now upon us. And advertisers are starting to embrace it.
The Interactive Advertising Bureau (IAB) estimates that 39% of digital video ads will be built using generative AI within a year. In all likelihood, many – if not most of the ads you encounter going into the future will be AI-generated.
But for a transition such as this to really take hold, AI ads will have to demonstrate their capacity to produce ads on par with those created by humans. Is this really likely?
That challenge has two parts to it: First, that AI systems be able to produce ads at similar production quality to premium human-produced ads. Second, that it be able to conceive of concepts without relying on human intervention.
AI Production Capabilities
At MediaPET, we’ve already solved for the first challenge. Our platform now delivers at 98% of the perceived production quality of ads for power brands like Apple, Doritos and Snickers.
To validate this, we recently conducted a study ‘re-creating’ premium ads. We created prompts that described, shot by shot, what happened in an ad and had our platform create the ad. The results, independently analyzed by researchers at the Ehrenberg-Bass Institute, one of the leading academic research centers in marketing science, demonstrated no significant difference across any of the classic ad impact metrics between the MediaPET ad and the original premium ad.
And recently we conducted a similar validation study recreating the recent Coca-Cola ‘Holidays Are Coming’ ad. While the Coke ad was produced using AI, it still took a team of over 100 people thirty days to produce (a considerable advantage over human production which previously required a full year). And despite being AI-produced, the Coke AI ad was effective, in our testing, significantly outperforming holiday ads for Dr. Pepper, Pepsi, Sprite and Fanta.
To test the efficacy of our platform, similar to our previous study, we recreated the Coke ad using MediaPET. But our recreation was produced in just two hours by a single editor, at what would have cost a subscriber less than $100 with all its edits (and longer duration). And the results? Once again, we delivered impact on par with the original ad that outperformed Coke’s competitors.
What these studies demonstrate, above all, is that the power to produce cinematic-quality ads, of the kind that traditionally was the domain of power brands, is now available to local businesses, small brands and single person digital creators. This is a powerful equalizer and potentially, this could revolutionize advertising – dramatically enhancing the quality of the ads we see. Because as we move forward into the future, there is no excuse for low production quality advertising. Everyone now has the ability to produce a Super Bowl quality ad.
AI’s Ability to Conceive Ad Creative
But here is where the second challenge kicks in.
In the above examples, the ads were recreations and not original concepts. Humans were still critical to conceiving the ads in the first place. Without that magical human touch, the impact of the ad would be lost entirely.
So while AI can now deliver premium production quality, it’s still constrained by the creative concept itself. A weak concept, no matter how well executed, will still deliver weak results.
Is it possible for AI to create ‘original’ concepts – that is, scripts which an AI system creates on its own accord, with human guidance limited to a prompt alone?
This is the focus of our R&D initiatives at MediaPET. Our platform already does this – you can simply instruct it to create an ad for a flower shop, for example, and it will do all the work, from conceiving a script to fully executing it, even with a custom musical score (it’s a fully ‘agentic’ system).
And the result, even at present, is actually remarkably good. I worked as a copywriter at one stage of my career and won a few awards in my day. So, I did a head-to-head contest where our AI competed with me in creating an ad concept. My team’s consensus? The AI won!
But currently, our AI is not consistent in producing good concepts – so our challenge here is to build an AI agent that can consistently outperform humans. And in this task – we have a massive advantage. Because MediaPET was created by our sister company, MediaScience, which started off as Disney’s media and advertising innovation lab and has gone on to do similar innovation research for almost every major US TV network over the past 17 years.
MediaScience has some of the deepest data in our industry on how ads work, with neurometric measures of ad impact including galvanic skin response, heart rate variability, eye tracking, facial expression analysis, EEG, response latency as well as traditional survey-based measures. Most of MediaScience’s clients have agreed to allow it to train AI models against this incredible data trove. The result? AI algorithms reliably predicting ad performance.
So in the near future, MediaPET will be able to harness this data in the formative process of creating its ads. This will ensure that ads produced in MediaPET perform better than the average national premium ad consistently. And this is a vision that is only months (or even weeks) away.
This will truly deliver on our vision of leveling the playing field – the marketing equivalent of David vs. Goliath.
And for big brands still wed to traditional ad production, this affords an exciting opportunity as well. Now, concepts can be reliably tested prior to actual production, so that only ‘great’ ads get produced in the first place.
Again, collectively, this means that the quality of the ads we view should rise considerably in the age of AI.
Implications
On one level, the implications here are devastating. If you work as an ad creative, this is clearly a terrifying prospect. Many such creatives will be in denial, hoping it will all prove to be little more than a passing fad. But with the passing of every day, AI will continue to persuade advertisers – and more and more ad creatives will have to figure out how to adapt and capitalize in the new landscape (or risk being relegated to being obsolete).
And there’s no sugar-coating that reality. It’s bad news for ad creators by any measure – or at least challenging news.
But on the other hand, think about what this means for local business, for the 33,000 small digital agencies that have only a few employees, for small brands and for content creators. This is a revolution in the making – the democratization of human expression – where you no longer need massive budgets and exceptional talent to create a great ad.
Increasingly, the power to tell compelling stories that move people will be a medium available to all.
Naturally, there will be some social resistance to this new frontier at first. When the new AI-created Coke ad was unveiled, many criticized it at levels of nuance: the size of the trucks were inconsistent, the number of tires on these trucks varied from shot to shot, the whole ad was animated rather than photorealistic (an easier task for AI), etc. But remember, in our testing, the ad still significantly out-performed other holiday ads for Dr. Pepper, Pepsi, Sprite and Fanta – and those were all ads made entirely by humans.
As people become accustomed to seeing AI-generated ads, they’ll increasingly come to accept it. And as AI goes from making good ads to creating great ads, they may even soon come to expect and appreciate them.
But no matter how you slice it, the genie is definitely out of the bottle. There’s no looking back now. AI-generated ads are here to stay. The question everyone should be asking is how they can get ahead of the curve and best capitalize on the new opportunities at hand.
