Beyond ChatGPT: What's next for AI and the marketing industry?

This article was first published in Campaign Asia on May 24th, 2023

Generative AI technologies like ChatGPT, Midjourney, and Dall-E have already impacted the marketing industry. But beyond these well-known tools, we explore other ways agencies and marketers use AI and how it might change the industry in the long term.

AI interns Aiko and Aiden are employed by tech marketing agency Codeword, within WE Communications

Whilst ChatGPT has taken the world by storm over the last six months and is the poster-child of the current wave of AI tools, the reality is that AI has been in use across the marketing industry for several years.

"We've been working with earlier versions of these types of tools over many years – both within strategy, data, analytics and creative departments," says Oliver Kunze, chief data officer at TBWA Asia. "In fact, as far back as 2018 we were already experimenting with early versions of GPT."

Michael Lee, partnership chief strategy officer at VCCP, says that AI underpins much of the media buying, measurement and optimisation process of any typical media agency.

"AI tools have been in use within our agency for several years especially within our VCCP Data and VCCP Media parts of the business. Many of our data tools like ContentScoringEQ and FluencyIndexEQ rely on AI-powered natural language processing (NLP) and image recognition. Within our production capability, Girl & Bear, we've created a piece of software called Pathway which incorporates a number of AI tools to deliver content personalisation, adaptation and optimisation."

And for creative agencies, while many are already using Midjourney and Dall-E when ideating and establishing a creative vision, some are also starting to use Google Bard or Claude from Anthropic as well as a host of others.

"We also have teams going much deeper, using Google Colab or Jupyter Notebooks to run Stable Diffusion algorithms or alternative large language models (LLMs) to build prototypes and proof of concepts," says Laurent Thevenet, head of creative technology, Publicis Groupe APAC & MEA. "And we are seeing the rise of RunwayML as a favourite tool among creatives willing to explore the emerging video-to-video and text-to-video territories."

Anthony Baker, VP, head of technology strategy Asia at RGA, says that beyond ChatGPT, they’ve been actively exploring a range of other LLM tools that are finding their way into their regular toolkit. “Among these tools, Google Bard, Bing AI, Canva Magic Write, and GitHub Co-Pilot have caught our attention,” says Baker. “These tools have high potential in assisting us with various aspects of our work, such as generating creative written content, refining search engine optimisation strategies, enhancing design workflows, and streamlining code development.”

AI companies & tools. Clockwise from top left: OpenAI, Stability Diffusion, Claude by Anthropic, Runway ML, GitHub Co-pilot, Midjourney

But while Baker and RGA are exploring the capabilities and limitations of these tools, they are also developing their own set of best practices and safeguards at the same time.

"We are keeping a close eye on the development of copyright policies, and making sure that anything we use for client work has very clear data training and governance policies to ensure responsible data and content sourcing," says Baker.

AI is already in so many everyday work experiences

While the story the past few months has been dominated by generative AI, we’ve already been leaning on other AI tools in our workplaces for quite some time.

"AI is in so many everyday work experiences," says Kyle Monson, founding partner at Codeword. "Video conferencing software uses machine learning to make the picture and sound better. Smart Reply helps me stay on top of email.

Search and calendars and docs and so many other tabs on my work laptop have AI baked into them." The same is true for commonly used tools like Grammarly, the typing assistant, or popular transcription service Otter.ai, as well as tools like Audiense, Talkwalker, and Signal AI that help measure sentiment analysis, surface conversation themes, and forecasting.

The use of AI is already vast. But while it's one thing having all these tools at your disposal, knowing how to use them is another. In recent months, there's been much talk of agencies hiring for specific AI expertise like prompt engineers and such.

But one agency, Codeword — a tech-marketing agency within WE Communications – has gone a step further and hired the first employees created by AI in the agency world.

AI interns enter the marketing world

Back in January, Codeword welcomed two new interns by the name of Aiko and Aiden. But unlike regular interns, they were not of the human variety. Instead, they were created using various technologies like ChatGPT, Midjourney, Dall-E 2, and others.

As new members of Codeword's 106 staff, you can find the AI interns on the agency's employee directory — an effort by the agency to demonstrate that they are “full members of the Codeword team”. Each of them has a designated reporting manager and receives monthly feedback based on their performance, just like any human intern.

"The intern program was an attempt to figure out how our creative teams can use generative AI tools, and it's going OK," says Kyle Monson, Codeword’s founding partner. "Right now, they’re useful as a research and writing assistant, but they’re very far from being able to do meaningful work with any kind of quality or reliability. That’s why the intern metaphor is good for this experiment. We’re treating the tools like we’d treat an intern’s work."

Monson adds that one of the big takeaways from the AI intern program was the realisation that we shouldn’t treat AI expertise as a specialty.

"There’s been talk recently about 'AI prompt engineer' becoming a specialty skill, and I think that’s the wrong approach," says Monson. "Every creative should be developing AI skills, and learning how to use the tools available to them. It shouldn’t be the purview of that one AI expert over there who you turn to for help."

Where are these AI tools leading us?

Will the tools remain just tools or will they evolve to be more? In recent weeks the news has been awash with stories about 'the great replacement', of how AI could soon take away people's jobs as companies look to increase efficiency and reduce costs. But defenders of AI have argued that the technology will never fully replace humans at work.

However, days after it was granted Microsoft's Azure OpenAI service license, top Chinese marketing agency BlueFocus hit the news after announcing it would replace human copywriters and creatives with AI, and “fully and indefinitely” end the outsourcing of creative design, copywriting, planning and programming, and interim employment.

But experts argue that the BlueFocus announcement was more of a PR stunt than a realistic plan, unless quality is not a concern for them.

"What these AI tools produce today still needs strong human guidance," says Thevenet. "I have been playing in this new field since 2019 and the working model remains the same for now. It's about humans and machines working together."

Others agree that whilst generative AI has the potential to revolutionise marketing, it is not a replacement for human expertise and creativity.

"AI should be used to augment and enhance our teams to increase productivity, efficiency, and return on investment," says Cheuk Chiang, CEO of Dentsu Creative. "A great analogy would be generative AI as a sous chef or kitchen assistant. A sous chef helps with the preparation in order to realise the master chef’s vision. AI helps prepare our thinking and craft in order to realise our creative vision."

Jon Williams, founder and CEO of The Liberty Guild, has already hired a conversational AI specialist at board level to guide his team and their clients. He says it’s far more than prompt engineering or direction, it’s about ‘fine tuning’ AI models.

Williams was also recently on the ‘Turing Jury’ for the BRXND conference in New York trying to tell the difference between human generated ads and AI generated ads. "It was harder than you’d think," says Williams. "My view is that AI is currently at the level of a junior art director. It still needs a lot of creative direction. But that won’t last forever, and the acceleration curve is terrifying. Pretty soon it will BE the creative director. At that point we get our coats as we leave. Change is coming. The CEO of Microsoft put AI on a level of the discovery of fire. Think about it."

But as AI continues its rapid acceleration and becomes increasingly powerful, it's widely agreed that AI needs tighter regulation. Even the creator of ChatGPT, Sam Altman, the CEO of OpenAI, has called on US lawmakers to regulate artificial intelligence.

Samuel Altman, CEO of OpenAI, testifies before the Senate Judiciary Subcommittee on Privacy, Technology, and the Law May 16, 2023 in Washington, DC. The oversight hearing focused on rules for artificial intelligence. (Photo by Win McNamee/Getty Images)

"We urgently need policies to ensure change is managed in an ethical and equitable way," says Chris Dodds, co-founder of Icon Agency. "For example, if a large language model or image generator is trained on a writer or artist’s work, there should be mechanisms in place to compensate that person or IP holder. Where is user data stored and how is it being used to train future models? How will AI impact our sense of self-worth as experts participating in the workforce? What is our purpose when knowledge and memory become commodified? In the coming years, we will face existential challenges that require curiosity and an understanding of the opportunities and threats presented by AI."

How will AI impact the marketing industry in the long term?

The likes of Facebook and Google have already announced that they will offer ad clients AI generated marketing campaigns similar to ones created by humans at agencies, but the news doesn't have agency leaders reaching for their coat just yet.

"I'm deeply skeptical that AI will ever be able to create human, differentiated experiences," says Monson, founding partner at Codeword. "Coming up with a novel idea that's never been done before isn't something AI is good at. It's also not something most agencies and brands are very good at, which might eventually lead to problems for them. But that's way down the line."

Kunze at TBWA Asia says that the biggest risk he sees with the mass adoption of AI, at least within the strategic and creative disciplines, is the race towards parity of thinking and output. "When all the prompts have been optimised and we all have instant access to everything, it’s going to be the agencies who can go draw on their own, proprietary methodologies and skilled people that will create the most disruptive and effective work."

Assuming the IP, regulatory and ethical aspects of generative AI are sufficiently resolved, Lee at VCCP believes that the creative output of generative AI can be used on a regular basis. "The most obvious uses of generative AI will impact more immediately at the mid-to-bottom funnel level of creative production," says Lee.

One thing is certain, the genie is out of the bottle and AI is not going anywhere soon.

"It will help us work better, open new creative doors we never thought about and enhance personalisation at scale," says Thevenet. "All teams will be using AI for their own respective needs. It will also most likely be the enabler of future technologies."

But, Thevenet adds, we will also see a rise in AI ethics with risk teams being set up to ensure we are using AI responsibly.

"This is already happening internally at Publicis Groupe and we are also seeing governments such as the US or Singapore providing guidance in this space for companies to be responsible."