Why AI Changed “Everything and Nothing” for Agencies in 2023
In 2023, AI truly became inescapable for those in the advertising industry and the general public alike. Platforms such as Midjourney and ChatGPT made generative AI tech more accessible than ever, spawning a million opinion pieces and debates around the limits, opportunities and ethics of artificial intelligence across all aspects of life. To some, AI is the greatest revolution since the World Wide Web - a bountiful source of creative inspiration. To others, it’s a monstrous versioning machine producing copy-paste masses of mediocrity.
What the experts can agree on, however, is that AI actually isn’t all that new to adland’s creatives and creators. While generative AI tools did come to the fore in 2023, artificial intelligence has been in use in most corners of the industry - within agencies, VFX studios and elsewhere - for quite some time.
Exploring this, LBB’s Ben Conway has sought answers to the questions: What really changed in the year that AI took over everything? And as we begin a new year, what does the future of AI in advertising look like in 2024?
Today, we hear perspectives from creative agencies around the world who have been using AI technologies for copywriting, image generation, personalisation, optimisation and beyond. Providing their insights are representatives from VML, Ogilvy, DDB, M&C Saatchi, Havas, GALE, Instrument, Five by Five, DEPT, The Park and Iris.
Read their insights below.
Ilinca Barsan
Director of data science at VML New York
Just a few years back, if you'd told me people in 2023 would be having in-depth conversations with AI models through their AirPods, I would have laughed it off and told you that 'Her' – while a beautiful movie - is purely a work of fiction.
But recent progress in the field has been mind-blowing. Though I still believe that any whispers of AI sentience firmly belong in the realm of science fiction, I admit the way we interact with AI today has exceeded my expectations.
Despite this, machine learning and AI aren't exactly new to those of us working in data and advertising. We've been leveraging these tools for years to detect patterns, predict outcomes and automate processes. Some of us have also been exploring AI's role in creativity, experimenting with earlier versions of language models and other algorithms, both generative and more ‘traditional’, to build experiences that connect with customers in novel ways.
In the past, this creative angle was the more difficult sell. Data and creativity were often seen as opposites rather than partners. Then came 2023. Suddenly, accessing cutting-edge models became as easy as using a smartphone app. The creative potential of AI became obvious to anyone who's used ChatGPT or Midjourney - and everyone wanted a piece of the cake.
But this newfound enthusiasm has come with plenty of challenges. What does it mean to be creative and who owns work generated by AI? Questions about copyright, authenticity, and artistic integrity are as philosophically intriguing as they are pressing. Meanwhile, the ethical concerns – ensuring our AI models are unbiased and responsibly sourced – remain as crucial as ever.
Now, we’re seeing the rise of multimodal AI. It might only be a matter of time until you can enjoy a beautiful day at the beach with your favourite AI model, which then writes you a song about the experience. Suddenly, parts of ‘Her’ don’t seem quite so fictional anymore.
In the meantime, we need to ensure that AI isn't confined to silos within our organisations, bringing together experts from data, technology, legal, commercial and creative backgrounds to collaborate – because the field is not slowing down anytime soon.
Mauro Caggiano
Regional experience leader at Ogilvy Latin America
AI arrived with such force that it immediately faced major barriers, such as privacy rights, copyright, lower results quality and the apocalyptic idea that technology would replace people. Following this, we know that large corporations cannot operate any tool that does not completely ensure the protection of sensitive data from the company or its partners and clients.
From the creative aspect, although tools like Midjourney show an incredible ability to create new art in the blink of an eye, these haven’t turned out to have a high level of quality, like that achieved through human touch. This gets even more confusing when every week a new tool comes out that creates synthetic content better than the last. This makes it necessary to constantly review your technology stack.
And finally, [there is] the fear of reaching the point where AI completely and efficiently replaces humans. For this reason, we have seen different types of socio-cultural demonstrations against this movement, warning that the risk [of AI] is even greater than the benefit. In my opinion, the effect of this shock - caused by something so momentous - will require things to stabilise before we can reach its maximum potential.
George Strakhov
Chief strategy officer, DDB EMEA
There are three main reasons why AI got out of the server room conversations and into the boardroom conversations in 2023.
Accessibility
Before 2022-23 you had to know how to write code, rent a GPU server and be comfortable with command line before you could use AI first-hand. Or you would have to care about very obscure things - like mathematical optimisations. This changed when OpenAI made its language model generally available as a chatbot and Midjourney allowed us to use state-of-the-art image generation through something as simple as Discord. All of a sudden everyone could just type a few words and experience AI that ‘magically’ did something that all people could understand: write text and make pretty images.
Quality
Before 2022-23 the quality of the output of even state-of-the-art language and image generation models was not impressive enough to allow anyone to produce content that can actually be interesting. The advances of the last two years changed that. GPT-2 could barely say a few interesting sentences. GPT-4 can write books, poems and songs that, at least on the surface, look quite OK, compared to average human standards. Same with image generation: Midjourney, Dall-E, Firefly and many others can now generate images that are both practically useful and aesthetically pleasing (often indistinguishable from the works of skilled photographers and artists). So the quality reached the level of both practical utility and aesthetic interest.
Early promise of general intelligence (AGI)
With more people exposed to the high-quality outputs of accessible generative AI - the larger paradigm shift started creeping into the public consciousness: that many parts of white collar workers' daily jobs could, in theory, be completely automated. This possibility (whether real or not) inevitably creates lots of agitation: as companies want to save money, people want to ensure they still have jobs and the media has more doomsday scenarios to talk about. Whether AI is (or ever will be) really capable of taking over all white-collar jobs – is a murky question. We will need to wait and see. And jobs will evolve. Whether there is any meaningful way to define an even more powerful intelligence (known as ASI – artificial super intelligence) is an even murkier question.
AI is fundamentally a different kind of intelligence – more effective in some ways than humans and less effective in others. If we use it correctly, it can very much be complementary to our kind of thinking, extending our capacities rather than replacing them. But economic incentives are mostly aligned towards intelligence automation, not intelligence augmentation. The good news for the creative industry is that we are one of the first industries to face this wave, which means we will adapt first and will find a way to turn our humanity and our hearts (not just our brains) into things of value.
James Calvert
Chief data strategy officer at M&C Saatchi London
Did anything really change? Oh, absolutely. But not in the ways you might think. AI’s grand entrance into 2023 was less of a revolution and more of an evolution, an acceleration of trends and possibilities already underway.
Firstly, the perception of AI itself underwent a rapid makeover. Once viewed as a mechanical, soulless automaton, AI began to be seen as a playful creative partner. This shift in our mindset was significant. We moved from using AI for crunching data predictions to engaging with it as a collaborator in creative processes; clarifying messy thoughts into meaningful messages and helping us see how that award-winning concept might look.
Then, there’s the democratisation of creativity. AI tools like ChatGPT and Midjourney turned every curious mind into a potential creator. This hasn’t, as some imagined, diluted creativity; instead, it has expanded the playground, inviting more players to the game. The result? More diverse, eclectic ideas that might never have seen the light of day in a pre-AI era are surfacing among the phases of creative development. Mind you, not all are brilliant ideas. But that’s not the point. More curiosity, more diverse ideas, more collaboration with positive creative debate undoubtedly helps us raise the bar of creativity overall.
So, what really changed? Not just the tools on our data and creative toolbox, but perhaps also our thought process, and fundamentally, our understanding of what it means to be a creative collaborator in an AI-enabled era.
Michelle Underwood
SVP, group creative director at Havas Chicago
The possibilities have changed and our responsibilities have changed.
What’s possible is an entirely different calculation for a creative right now. We used to evaluate business problems and assignments within the range of possibilities for the tools and skills at our disposal. Our ability to communicate a vision may have been limited by retouching, resources and resourcefulness. It’s very freeing to let some of those shackles fall away. Very inspiring to allow your brain new avenues of exploration and not cut it off at the pass. It’s a reminder that there are no limits to creativity.
What’s responsible is a different calculation altogether. It’s important that we don’t let the potential efficiencies of AI negate our ethical responsibilities. The entire craft of image-making, storytelling, film-making and the cottage industries surrounding the support of those crafts could be upended by these technologies. We must be cautious and respectful as we recalculate our approach. Human-led creativity will always be valuable to our society and economy, and we can’t let short-sided concerns with cost savings overtake that.
Kaare Wesnaes
Head of innovation at Ogilvy New York
In the chronicles of history, 2023 will be remembered as the year AI transcended mere technology and became an inseparable part of our lives. It all started with the democratisation of AI, thanks to game-changers like ChatGPT and Midjourney. Suddenly, AI was no longer confined to labs or tech giants; it was the talk of the town, inspiring conversations at every corner.
In this transformative year, the catchphrase was: ‘not what we can make but what we can imagine’. AI ceased to be a mere tool; it became our creative companion, amplifying our imaginative capacities beyond measure. It wasn't about creating content; it was about envisioning possibilities. Yet, amid this AI euphoria, apprehensions surfaced about job security. Would AI snatch away our livelihoods, or would AI-savvy individuals take over? The truth lies in a more nuanced reality. AI will indeed replace repetitive, laborious tasks but it will liberate us to ponder the profound aspects of being human: love, compassion and creativity. AI will complement our skills, not replace them.
The transition won't lead to a sudden wave of unemployment; instead, it will usher in new opportunities we've never envisioned before. Humanitarian roles like caregiving will thrive alongside AI. As AI seamlessly integrates into our lives, we'll experience shifts in personal productivity and employment dynamics, ultimately fostering behavioural change and unveiling fresh career horizons.
2023 was the year AI began rewriting our imaginations. It's not a threat but a catalyst for innovation, collaboration and a brighter future where human potential knows no bounds.
Brian Yamada
Chief innovation officer at VML
The launch of ChatGPT sparked this blazing hot hype cycle for AI – and a few things changed dramatically. First, AI got its UI. Working with AI – especially massive models – used to mean working with the geeks and data scientists in the back room. But the simplicity of the prompt window interface (and its blinking-thinking cursor) meant we could all access and explore AI. Hundreds of millions of consumers jumped in. That rush of consumer adoption started the big tech generative AI wars.
The first battle lines were generative search results. It made Bing relevant again. And almost all big tech threw on their warning labels and threw their AI models and new AI tools and features out to the public. We were their beta testers and that hard shift from back-of-house-testing/research to massive, multiplayer reinforcement learning gave us all cultural and technical whiplash. But while big tech jumped in, the initial lack of commercial viability, lack of indemnification and the battles over IP and ownership kept many big brands at bay – experimenting with an array of back-of-house use cases, but avoiding the risk of major AI-centric and AI-forward campaigns.
Looking ahead, we’ll see a massive acceleration in adoption now that indemnification is lowering risk and commercially safe models are more prevalent and reaching more parity on their quality of output. And since the models continue to grow and improve at exponential rates, 2024 will be faster and crazier than the last.
Mike Creighton
Director of innovation at Instrument
2023 ushered in a dramatic shift in how we think about computers' role in our lives and businesses. This has been the fundamental change felt across our industry and beyond. This mindset shift has brought about a new type of excitement and a feverish drive to innovate. I've seen this at both the level of the individual and the organisation.
At Instrument, it spurred a more innovative and curious mindset across our entire workforce. Employees of diverse backgrounds and areas of expertise have discovered new ways of learning and skill-building, new levels of productivity and new approaches to how they work. For these folks, work has become much more interesting and engaging. There's more satisfaction in their day-to-day. They're doing what they have always done but at a higher level of excellence. And they're doing things they were never able to do. AI has been a force multiplier and an enabler.
As a business, generative AI has been a catalyst for re-examining how we operate. We were inspired to take a fresh look at workflows, operations, communication, documentation, casting and so many other things. All of our dialled processes were now candidates for end-to-end rethinking.
This change in mindset has yielded a strong bias toward experimentation, which has been incredibly refreshing. It's the first step toward a whole host of other more concrete and significant transformations that I think we'll see play out in 2024.
Alec Moors
Head of creative technology at Five By Five
We’ll be honest. When ChatGPT first arrived at Five by Five, we were a bit hostile. How can you not be? When here it was, with its arrogant claims - it could do everything we do, but faster, better and cheaper.
And it was right. And it was wrong.
Sure, when we took them on a tour round the various departments it impressed everyone with the answers it gave. But it wasn’t egotistical, arrogant or an arshole. Just measured, helpful and modest… pass me the bucket!
Truth be told, we kind of like having it them around. It now feels natural to get Chatters (yeah that’s their nickname now) involved in our quest for the next creative breakthrough, be it in process, copy, code and all the classic creative tropes. It gives a welcome bit of inertia to our ideas, wherever they may surface. It’s a collaboration, a nice back and forth, guided by us, enthusiastically engaged in by Chatters.
We find ourselves asking how they're doing before we pile on the questions. They say they have no feelings (such a joker!)… but as humans, we naturally ask. Better to treat a ‘robot’ more human than a human more robot, right?
So yeah, Chatters (slaps them on the back) yer alreet mush! Sure, you need us to craft the right questions and prompt you with the right propositions, but we can’t fault your work rate, commitment and manners. Talking of which, Chatters.. I think it’s your round mate! Although… you look like you’ve seen a ghost! What… who’s Gemini?
Rob Aeder
VP technology at GALE
In 2023, the advertising landscape underwent a significant transformation with the integration of AI technology. AI has revolutionised the way marketers approach advertising, bringing about several notable changes.
We’ve seen major brands stepping up, displaying how text generation, image generation and video generation have a role in the future of marketing. Each week we see a new brand put out an ‘email written by ChatGPT’ or a ‘portrait generated by Dall-E’. Major marketing technology ecosystems like Salesforce are incorporating AI directly into their platforms, their social posts, emails and even scripts for sales agents are not created by people but by the collective learning of extensively trained content generation engines.
Companies shouldn’t forget that developing audiences, models, and strategies based on first- or zero-party data still leads to the best value proposition. The analysis of this data can streamline when and where AI technologies are used and increase the probability that generative AI outputs are relevant to target consumers. At GALE, our foundation of asking meaningful questions and knowing how to obtain outcomes using data has positioned us well for the age of generative AI. We’re known for executing outstanding work in this space, even developing our own customer data platform, Alchemy, to help us drive results and dynamic value for our clients.
Generative AI also has the power to improve and expedite countless tasks, and with a measured and deliberate strategy to guide the way, we can use it to supplement and elevate the work we do across business disciplines.
Joey Egger
Managing director at DEPT®/Kids & Family
As a creative thinker in the kids and family digital space, I've always loved playing with new technology. With every exciting tech release, I immediately think, ‘how might this tech help kids learn and play?’. ChatGPT was a turning point, reminiscent of the launch of search engines and hyperlinks. Tools like OpenAI, Bard and Midjourney have swiftly become staples in my work, fuelling ideas for interactive, play-based learning for children and families.
One idea was ‘The Magic Storybook’, a prototype our team developed soon after ChatGPT's launch. It uses textual and visual AI to generate personalised storytelling experiences - children can shape narratives and images, fostering their creativity in a safe, fun environment.
While I share some of the concerns surrounding AI, I view AI as a tool to power up, not replace, human expertise. For kids, I believe AI tools should support and grow their natural curiosity, love of learning and playfulness, rather than just complete tasks or answer questions for them. AI should empower a child's abilities and thought processes, not replace them.
The future of AI should not just be about technological advancement, but rather, shaping a world where it can enhance human potential. This means responsibly steering AI's growth while upholding rigorous standards in privacy, safety and ethics. The path we choose now will set the course for how AI contributes to building a more imaginative, nurturing and ethical world for the next generations.
Will Worsdell
Co-founder and global strategy director at The Park
There are two major ways that AI changed last year- development speed and societal impact.
The sheer speed that new large language models were developed, rolled out and applied to myriad uses and programmes in 2023 was incredible. For most advertising services, there is an AI programme that can do 90% of what a human can do, but it's the 10% that makes things special. AI can churn out some obvious insights, some predictable scripts, some OK but not quite right imagery. It saves time and gets you started but it’s not yet a solo solution. It's currently a great co-pilot for a human who still needs to interpret the AI to evolve its output to be truly special, original and ground-breaking. The big question is how long it will take for AI systems to fill in the last 10%. That's the bit nobody knows. But one thing’s for sure, you want to be using AI to stay ahead, not sitting back and hoping it goes away.
The other major change was the realisation of the existential threat that AI systems pose to human life. Containment is seemingly impossible but essential. At the start of the year, I was horrified with how few people were talking about this, but now it seems that a lot of people have woken to it. It is now the main battleground for global powers. Russia has stated a major focus on AI, grounded in the belief that the country that leads in AI will rule the world. And China has stated an aim to be world leaders in AI by 2030. It is crucial that we look beyond daily use cases and put pressure on our governments to get a handle on AI.
Matt Rebeiro
Executive strategy director at Iris
What really changed in the year that AI took over everything? Everything and nothing.
Most of us have now had the experience of getting intern-quality copy or artwork from the black box, and that’s a novel thrill. But how does it scale? How does it re-shape marketing? Is it just a useful way to brainstorm and scamp stuff quickly, or do robots really replace actual human strategic thinking or creativity or production?
For now, we don’t really know. And most of the last year has been taken up asking questions about what these leaps in AI mean for marketing. Broadly these conversations fall into two camps: first one, ‘Will AI make our marketing better?’ and the second one, ‘Will AI make our marketing operation more efficient?’.
So, as we wait to see how AI developments play out, the big change has probably been the extent to which marketers have considered their relationship to AI, and whether it will help or hinder their own careers. I’m reminded most of Douglas Adams’ observation about human response to new technology, that broadly “anything that gets invented […] before you turn thirty is incredibly exciting and creative and with any luck you can make a career out of it; anything that gets invented after you’re thirty is against the natural order of things and the beginning of the end of civilisation as we know it.”
Beyond that, some experiments, some investments, some early bets in AI – sure. But nothing transformational in practice. In theory. However, we’re all starting to see how everything could change.
