Mike Sharpe, Executive Creative Director at design and motion studio ManvsMachine is not new to the creative industry. “I graduated in 2000, starting out in film and then moving into animation as technology evolved. And in that 25 years, I’ve seen the birth of a whole new industry, one that’s more and more able to do things in the computer that you would normally have done on set.”

At first, he was taken aback by AI. “I was an AI sceptic to begin with,” he admits, “but when I started to understand how we were using it, I realised it wasn’t just some machine spitting out random images. It’s just the next step of that marriage of creativity and technology that’s been going on for decades. And like everything that’s come before, it’s humans that stay in control.”

For ManvsMachine, the power of AI lies in the little things, accelerating processes behind the scenes, rather than producing end products. “Something like character design can be a time-consuming process. Just turning a sketch into a 3D would take ages. But with AI and WPP Open's Creative Studio, we’re able to show clients a multitude of ideas rather than just a handful, and do it much more quickly,” says Mike.

A similar advantage is seen by Priti Mhatre, Chief Product & AI Officer at Hogarth. "At Hogarth, we view generative AI as an amazing opportunity to deliver industrialised scale to our clients. We combine the power of generative AI with our craft expertise and cultural insights to supercharge content supply chains for our clients and drive their growth."

The power of AI in creating the next generation of movies, TV shows, and commercials isn’t just confined to backend either – it’s also letting teams do things in front of the camera that were truly impossible with older methods. Kat Sullivan, Senior Director of Interaction Design at WPP cites a client who wanted to film a car advert in the Namibian desert. Due to local filming restrictions, a traditional shoot wasn’t an option – but thanks to AI, the team didn’t have to do a traditional shoot.

“Instead, seven of us went to Namibia, where we scanned the roads in this beautiful desert environment,” says Kat. “Then, back in London, we trained an AI on the scans in a way that let us generate not just landscapes, but also reflections of the desert on the car’s body, or dust being kicked up by wheels. We produced car commercials on roads that were literally impossible for cars to access.”

But creativity isn’t just about the creator. It’s also about the audience, about who gets to experience it. And AI has powerful implications here for the way content is accessed. Josh Loebner is Global Head of Inclusive Design at VML, and gives the example of how AI tools are being used to create subtitles that truly transform live events for people with hearing disabilities “Sporting moments, for example, are very powerful, but if you’re deaf or hard of hearing, you’re often cut off from the emotion,” says Josh. “So, as well as basic descriptions of where the ball is, or who is doing what, AI is allowing us to auto-generate captions that convey information about the crowd, about the excitement in the stadium.”

Josh highlights how this bridges the critical but often-overlooked gap between information and experience. “If accessibility only allows people with disabilities to be ‘informed’, then they don't have the whole story,” says Josh. “They might know what’s going on, but they don’t have that opportunity to cheer and share in that camaraderie with others. AI has the power to change that.”

The ability of AI to broaden access to content – for viewers and for creators – is one of its most exciting applications. But it also points to a critical issue that concerns all use of AI: it’s just a tool, and like any tool, it won’t result in the desired outcome if we don’t use it with care. And that’s particularly true when it comes to inclusivity.

Josh – who is himself legally blind – gives the very simple example of generating images of people with disabilities. “If I ask an AI to generate a picture of blind person using a cane, what I'll often get back is an image of someone with a blindfold, holding a candy cane,” says Josh. “The algorithm is getting it wrong, basically, because there aren’t people in the room giving clarity on what that should look like.”

And the issues don’t end with disability. Without human oversight, any marginalized group can find its experience shortchanged by AI. Kat gives the example of a campaign in which her team was asked to show a client’s product alongside meals, which needed to be localised to match the cuisine in different markets. “The problem was that food eaten outside of the western world is less likely to be photorealistic in a generative AI model, because realism is influenced by how much imagery that model has access to” says Kat. “And there are lot more photos of western food in AI training data sets.”

If AI wants to avoid these issues, then underrepresented groups need to be built into the design process. Kat co-founded WPP’s Creative Tech Apprenticeship in 2022 to train the next generation of AI talent, and ethics and inclusion are central to its curriculum and recruitment processes – part of its goal is to ensure that future AI technology truly reflects the diversity of human society.

Because there are better ways of doing things. Josh highlights a project with automakers Honda which involved building an app that would give audio descriptions of the scenery outside a car window for visually impaired passengers as an example of this done well. The project did not dictate what it thought users with disabilities needed. Instead, it engaged individuals from local blind and deaf institutes to evaluate the solution’s effectiveness and source feedback that could be used to refine the user experience.

Having the right people in the room means you can also fix any issues that do emerge. Kat mentions a campaign in which AI was used to generate images of users as children. It was a cute campaign, with just one snag: if the source photo had facial hair, this would appear on the child version of the face as well. But with human oversight “we were able to teach the AI to assess whether the starting image had a beard or a moustache and then remove it,” Kat says.

It’s critical to remember that without humans, AI can't create anything. In the words of Ben Affleck: “AI can write excellent, imitative verse that sounds Elizabethan. It cannot write you Shakespeare.” If you’ve tried to coax a generative AI to produce exactly the image you had in mind, you’ll know it’s not always the most precise instrument. In fact, it’s only by enhancing the granularity with which we can control AI can we make sure it truly delivers on its promise, whether you’re designing a 30-second ad, or shooting a two-and-a-half-hour blockbuster.

For Mike, having humans in the driving seat is something that’s embedded in his approach to creativity. “We've developed our own system, to be able to art-direct the AI rather than the AI art-directing us. We're still in control, there's still a human behind those decisions, controlling things visually, not just with prompts.”

This isn't just beneficial; it's critical. "I think the biggest change in perception is yes, AI can do a great creative… but will it do the exact, precise creative you need?" says Priti. "That’s where the difference lies. As production we need to do exactly what is required and is legally compliant and has the right copyright. There are a lot of details that go into production beyond the creative ideation front. The art of using AI is about guiding artificial intelligence through sophisticated prompt-engineering that codifies deep production knowledge, while simultaneously navigating the complex territory of compliance, ethical use and brand guardianship.”

Carlos Cortés, Lead 3D Designer at ManvsMachine, is similarly unambiguous about the importance of human control. “What makes a good image or good footage are things like composition, colour, and lighting: variables that an AI can’t decide. Humans always have to make the important decisions. Good AI enhances our work: as designers, artists, directors. It doesn’t just deliver premade content.”

That’s what has guided the ManvsMachine approach to AI. “What the team have developed allows users to not just describe the shot, but actually design the shot,” says Carlos. “We can say ‘let's bring the camera back’, ‘let’s change the floor material’ or ‘let’s put a city in the background’.”

And rather than just regurgitating other people’s ideas (with the attendant IP and copyright issues that this entails), effective use of AI actually enhances the ability of human creators to develop projects based on original design work. Mike gives the example of a recent project where the team produced a series of adverts in claymation style.

“We got an illustrator to sketch the assets we needed,” says Mike. “We then turned those sketches into 3D shapes. Thanks to AI, we could then iterate those ideas. They are still our creative ideas; there’s still a human driving the work. All AI does is give us the muscle to move through that creative process at speed.”

And Priti, too, sees the role of AI as fundamentally one that augments current practices, rather than replacing them: "Rather than viewing AI as simply driving faster content production, we see it as supercharging human craft with technological amplification to drive better outcomes."

AI is a tool, and tools are about serving people. Yet for too long, AI has been seen not as a tool but as something else: either a wonderful cure-all that can and should solve every problem – or an existential threat to people’s livelihoods. Both interpretations overestimate what AI can do on its own. And paradoxically, in exaggerating its power, those extreme positions also undervalue AI’s genuine potential.

“There is a misconception that AI can solve every problem with the click of a button. It's often a lot more complicated. We shouldn't fear generative AI, but we also shouldn't treat it as this magical thing that will immediately create the perfect asset.” says Kat. “There must always be a fusion of talent and technology. If we get that right, then the applications are truly incredible.”