When ChatGPT launched in November 2022 it was an overnight game-changer.
Within five days it had a million users. Within two months there were 100 million, and it now has over 25 million daily active users. From complex to mundane tasks, the potential of the technology seems limitless: ChatGPT can now help you write media plans, map out resource needs and explore game mechanics.
Companies are using AI to build tools enabling individuals to turn creative ideas into reality – like Roblox’s generative AI for 3D modelling, Adobe’s Firefly for design and OpenAI’s demo of turning a scamp into a website in a single click. “Artificial intelligence has escaped from the minds of developers into the hands of everyone. If you think it, you can make it,” says Isabel Perry, VP of emerging technology at DEPT.
What sets the likes of ChatGPT apart is the ‘T’ in its name, which stands for generative pretrained transformer. Using ‘transformer’ architecture can create AI models that don’t have to be trained on a specific use case; instead they are all-purpose models, immediately useful for almost whatever you apply them to.
“At DEPT we’re helping companies combine their proprietary data with pre-trained models to build bespoke AI solutions for any number of use cases,” adds Perry. “Transformers allow businesses to leapfrog 20 years of technological progress and allow you to be as good as your best competitor.”
For marketers in particular it represents a step change in the accessibility, affordability and efficacy of AI. So, what are some of the key ways that this type of AI could disrupt marketing in the coming years?
Creating assets at scale
One area undergoing rapid development is cost-effective asset creation. In short, using AI to cut out the laborious (and expensive) work of creating, replicating and tailoring assets for different markets, channels and campaigns.
Already automation is making a significant difference here. At Just Eat Takeaway, for example, the team has used AI to generate creative assets for various different campaigns, while deploying a creative automation tool that provided the ability to tailor that message across 17 different markets too, for example showcasing different restaurant partners in different cities.
“This has been transformational for our business, allowing us to build templated creative that answers our business needs while still delivering strong creative that speaks to our customers,” says Richard Burgess, the company’s global principal for brand digital and social media. “We’ve seen improvements in media metrics versus objectives when running through creative automation as well, as a reduction in resource and effort needed, with a huge ability to drive scale.”
Now the company is exploring how AI can help increase productivity and effectiveness further.
Benefits have been recognised at eBay too. “Within the last 12 months, we’ve been using generative AI in innovative creative campaigns,” says brand director Alex Allcott. In October 2022, for example, it invited attendees at Comic Con to immortalise themselves as avatars within bespoke trading cards, which were designed with the help of generative AI. “We’re now excited to take generative AI out of our ‘innovation’ process and embed it within our ‘assets at scale’ production process, to achieve more with less.”
The power of AI to support marketers here is only set to increase, says Perry. DEPT is currently building proofs of concept for tools that facilitate better layouts and designs, as well as those that can check for linguistic errors, analyse text padding and assess whether or not a word is too long in a given layout. In so doing, it could free up to 50% of the time currently spent on these kinds of activities within some roles in marketing teams.
Reducing media costs through segmentation
A second area where AI looks set to be increasingly disruptive is in optimising paid media, says Jonathan Whiteside, global SVP for technology and engineering at DEPT – by supporting areas including value-based bidding, improving search and discoverability, and SEO keyword clustering to name but a few. In particular, it allows brands to be far more targeted and cost-effective, cutting the cost of content by at least 50%.
Using AI to refine customer segmentation, for example, has “been a real breakthrough for one beauty client”, he explains, with AI being handed the data and asked to suggest customer personas based on what it revealed about buying behaviours. “It gives us a different insight rather than coming in with preconceived ideas, which makes our work far more effective.” Overall, the approach has driven a reduction of 85% in media costs at the brand, he says.
Third is the capacity to finetune these pre-trained models to create purpose-built applications for a brand. The potential here is limitless, she says. “It could be the TikTok of retail, the ChatGPT of white-collar industries, or the new Getty of stock imagery.”
And this is just the start. “As technology continues to evolve, we’ll also see opportunities for AI to improve the assessment of customer sentiment, and flag any early risks or attrition indicators,” says Dominic Allon, CEO at Pipedrive. “Overall, leveraging AI to improve sales and marketing will be crucial for businesses to succeed in the next decade, letting skilled people bring more of their empathy and insight into the tasks that matter the most.”
“We’ll get to a point where, if you can imagine it, you can create it – no skills required,” echoes Whiteside. In the near future, it’ll be possible to create mock-up designs, anticipate best-performing content, and craft automated media plans and channel distributions based entirely on AI.
“There’s going to be a trillion-dollar application spawned from these technologies in the next couple of years,” he adds. “If you’re not using AI your competitors will be – to be faster, more accurate and more cost-effective, but also to enable entirely new applications and experiences.”
For marketers then, the message is clear: embrace the potential in this generation of AI tools or get left behind by your competition.