AI is here to stay, the genie is out of the bottle and there’s no going back, however disconcerting that may be.
The decision now for advertisers and agencies is how to harness the power of AI to boost – not hinder – creativity and performance, while somehow retaining that vital but fragile bond of trust with consumers.
“AI has the potential to become one of the most important advances in technology in human history,” asserted Lloyd Davies, the UK managing director of Making Science, a digital marketing agency and technology company. Headquartered in Madrid with an office in the UK, 400 of its 1,300 plus employees are data scientists and engineers. “We see AI as the third big shift following the internet and mobile phones,” he added.
McKinsey estimates that generative AI could increase the productivity of the marketing function between 5-15% of total marketing spending.
1. Creativity: Here to help, not replace
The capacity of generative AI is already mind-boggling whether that be basic text prompts creating hi-res images or simple sketches being turned into fully rendered 3D models. Does that mean AI is coming for creative jobs?
Not according to Peter Gasston, creative innovation lead at VCCP, who recently set up its own AI creative agency, faith. “We’ve taken an actively positive view that the responsible use of AI is an accelerator of human creativity, not the opposite,” he said.
“We believe that it empowers creatives rather than replaces them. AI is not creative in any sense – it’s a tool that creatives use, such as Photoshop. Anyone can make an image with AI – not everyone can make a good image.”
Costanza Ghelfi, operations director at ad-machina, a Making Science company, agreed: “Google is democratising the access to creativity so established brands will have to supercharge their creative team. It will be a much more strategic process than before.”
“Surrealism is a big trend right now,” according to Nora Zukauskaite, global marketing director for three independent beauty brands, citing Jacquemus’ recent stunt of sending giant bags through the streets of Paris. “AI serves as a springboard – not a replacement – for ideas. We will probably all need to adapt, learn new skills and pivot in a certain way. That’s not necessarily scary or bad.”
2. Performance: Personalise at scale
Ghelfi is also the co-founder of ad-machina, one of Making Science’s three key AI platforms that uses natural language generation and a brand’s product feed to generate thousands of search ads tailored to individual search queries, generating more sales at a lower cost.
RIU, the Spanish hotel chain, used ad-machina to automate their personalised search ads and saw a 41% increase in conversion rate while almost doubling their number of bookings.
“If you want to really apply generative AI to performance advertising, you have to do it at scale,” said Ghelfi. “The real opportunities on the performance side are in personalisation. If we integrate generative AI in a safe way into our system, we can generate specific, relevant messages. We can finally combine automation and relevance.”
3. Human insight: Still vital
When creative agency Uncommon pitched a campaign using AI-generated art to its client the World Wildlife Fund (WWF), it was wary. “AI was uncharted territory,” said Holly McKinlay, director of strategic communications and brand for WWF UK. “We knew there was a risk and we’d had a backlash over NFTs.”
The campaign (‘Bringing our world back to life’) was designed to highlight how the UK is one of the most nature-depleted countries in the world. WWF created a digital and physical exhibition of AI-generated art based on the Romantic period of British painting by the likes of Constable and Turner.
Because of the historical nature of the source material, there were inherent biases perpetuated by the generative AI so WWF used illustrators to augment the images and make them more diverse. They also used a professor from the Royal College of Art to assess the artwork and help to ensure copyrights were not being breached. “AI’s great but it doesn’t replace experts and scientists,” said McKinlay, referring to the need to ensure that the art produced – showing how the UK’s landscape might look decades from now – had scientific credibility.
Zukauskaite made the point that “AI can’t yet replicate brand and tone of voice but it’s fantastic for tactical and technological skills, such as making sure that captions are search engine optimised.”
4. Metaverse: Endless opportunities
Zukauskaite believes that the “the combination of AI and metaverse is only a matter of time”, meaning an “interesting future of virtual and 3D worlds” where “the opportunities are endless”.
5. Regulation: Responsible, respectful marketing
Given how thorny the issue of cookies and third-party data has been for the advertising industry, it seems naive to think there will be a global standard to regulate the use of AI any time soon.
“We, as advertisers, have a really big responsibility,” said Ghelfi and offered hope to those who fear that large language models (LLMs) have the potential for rampant data breaches. “The combination of predictive AI and generative AI allows that personalisation without needing to access the personal data of the user. It is respectful marketing – a kind of personalisation that generates value to the users but is not invasive.”
6. Education: Shared experiences
“It’s our responsibility to educate as well,” said Ghelfi. “All our companies and brands need to invest in education not just for our teams, but externally too – and we need to really share our experiences. Each company is much more about the industry and building together.”
7. Consumer pushback: Communication and transparency
Fake news is old news. Now the issues are deep fakes and doctored imagery. AI has great power for good but it can’t be used at the expense of consumer trust. “Opinion will continue to be mixed,” said McKinlay. “That’s why we employed real artists [as well as using AI]. And, as the art historian we used said, ‘AI is still art’.”
Zukauskaite urged brands to be transparent about content that is produced by AI, saying: “It depends on how you communicate – it will be very difficult to define what is real and what is fake.”
8. Lawyer up: Be as watertight as possible
It’s important to be aware of possible issues of IP ownership and infringement: do you own the rights to the content you’ve created and have any rights been breached in the process of its creation?
JJ Shaw, of law firm Lewis Silkin, suggests creating a company policy for generative AI, “a little bit like you might have a health and safety policy or an information security policy”; and also to utilise sensible prompts when using AI platforms. For example, prompts such as “create a song like Adele” or “create art like Andy Warhol” is an immediate copyright red flag.
Shaw’s four tips for legally sound use of AI are:
Check your subscription level and rights on the platform you’re using
Ensure you’ve read the terms and conditions
Agencies and clients need to have clear conversations about risk
Perhaps they need to enter into an indemnity letter to share the risk. If that’s not practical then it’s better not to use AI.