Most people in the industry would likely tell you that AI plays a significant role in digital advertising. But they’d likely have a harder time telling you what its exact impacts have been.
That’s in many ways due to haziness around what does and doesn’t count as AI. This haziness in turn has allowed a plethora of tech companies to slap an ‘AI’ label on their products, regardless of whether there’s any real intelligence in their technology or not. As such, it’s hard to say with confidence whether an ad has been touched by AI or not.
A new report from GroupM aims to quantify AI’s influence, predicting how it will change advertising over the next ten years. And the media agency predicts that though AI has undoubtedly made its mark on the ad industry, we’re “only just beginning to glimpse” its potential.
Using a fairly wide definition of AI (including technologies and algorithms such as machine learning, neural networks, computer vision, natural language processing (NLP), and intelligent process automation), GroupM judges that currently 45 percent of all advertising is “AI-enabled” – meaning that AI plays a role on the media side (this figure does not include revenue from creative ideation and production).
By 2032, GroupM expects this to rise to 90 percent, covering around $1.3 trillion worth of advertising. And for areas of advertising already touched by AI, it will in most cases play an even bigger role in ten years’ time.
Helping DAI reach its full potential
One of the areas which is relatively untouched by AI at the moment is TV.
The growth of connected TV has brought AI-enabled targeting and delivery into the TV world, but the volume of transactions which are AI enabled are still relatively small (since not all CTV ads, or even programmatically bought CTV ads, involve AI).
Kate Scott-Dawkins, global director of business intelligence at GroupM and author of the report, says that the growth of streaming TV over the next ten years will see AI play a bigger role. More TV ad inventory will be addressable, meaning more inventory will be AI-enabled.
Scott-Dawkins also sees opportunities for dynamic ad insertion to start living up to its potential. “That would be the ability to use things like computer vision to gauge the context of what’s on screen, or derive context from the audio,” she said. “Then you could match that with a product and either show an ad for it, or even digitally insert the product into the content.”
There are plenty of tech companies in the market who already talk about these sorts of capabilities, but its still very rare to see this sort of placement actually happening. In ten years’ time, it could be much more common to see these sorts of contextually relevant ads.
Interestingly the limiting factor for swapping products in and out of TV shows and films may be legal issues, rather than technical ones. “There are so many regulations and mores around AI which we’ve yet to figure out,” said Scott-Dawkins. “Currently in some markets you need to get the director’s permission for a product to be shown in a film or TV show. And it can’t be programmatic if you need to go and ask for permission! So one of the areas we’ll see evolution is in how those sorts of questions get worked out in the next few years.”
Putting the AI in creative
AI’s role could also change significantly on the creative side of the industry.
AI already pops up in the creative process, for example in creative optimisation, and analysing audience data to sport trends which inform creative strategy. But you don’t see AI pitching wholly original ideas for a big new TV campaign.
Kate Scott-Dawkins says it’s not impossible that in ten years time, AI will be more involved earlier in the creative process.
“AI is already really good at synthesising massive amounts of information and then finding complex relationships and correlations, which can guide creative strategy,” said Scott-Dawkins. ” With generative AI, the other opportunity is actually throwing out an idea that no one had ever thought of before.”
It’s unlikely that creative directors will have lost their jobs to AI by 2023. Scott-Dawkins says AI will likely play a supportive role, as human input will still be necessary to separate the good ideas from the bad.
“Maybe the AI spits out 100 ideas, and then the humans do what humans are really good at, which is filtering and refining,” she said. “So maybe the creative team picks the best of those hundred, and asks for ten more ideas along those same lines. Then out of those ten, they pick their favourite, and ask for ten more along those same lines, and so on.”
Better, faster… fairer?
Of course, AI’s overall impact will stretch far beyond these specific examples, partly through existing technologies and techniques being increasingly adopted, and partly through entirely new use cases emerging.
This will help make advertising quicker and more cost efficient, as intelligently automated processes reduce time-intensive manual labour. But Scott-Dawkins says there are still gains to be made in effectiveness too, particularly as AI gets better at showing consumers ads they’re actually interested in.
“We can already see improvements in effectiveness playing out in areas of AI that are more mature,” said Scott-Dawkins. “So Google’s search algorithm has gotten much better at figuring out what users are searching for. And the better they’re able to understand the intent behind a search query, the better they are at placing the right ad in front of the user. That’s better for the user too, because the ad matches what they’re searching for.”
“I do think we’ll see that play out in other areas,” she added. “For example if I’m going to see an ad for a car, maybe the car is shown in a colour that I would actually buy, and on a road that I recognise.”
This however raises another question – will the benefits of AI mostly accrue with businesses like Google, which are best equipped to ingest huge volumes of data?
“You can certainly see how currently companies like Google and Meta have benefitted, using AI to benefit their platforms and DSPs,” said Scott-Dawkins. “But it has benefited small businesses as well to an extent. And I’m hopeful that there is an opportunity moving forward for large advertisers and agencies, who have a more holistic view across the media landscape, to utilise AI more, and not leave it solely in the domain of the Googles and the Facebooks.”
“I think some of that may come from advances in ‘low code’ and ‘no code’ tools,” she added. “For example with GPTs [Generative Pre-trained Transformers, a type of AI capable of producing text], you don’t have to know how to code to use them. Those sorts of solutions could really help large advertisers.”