Creative & Content

Less, but better: where AI creates real value in the content engine

Firehose of slop

‘For the head of a content at scale organisation to call for less may seem paradoxical but what he’s calling for is an end to the firehose of slop, and a more intelligent approach to meeting escalating content demands better’.

AI makes it easy to do a lot, fast. That is unarguably true. But what brands should be seeking is AI that can do what is needed, better.

From the moment of conception to delivery and beyond, AI has the potential to refine every step of the content generation and delivery process – provided it is put to work in the right way.

That’s the argument put forward by Paul Ward, CEO UK & Global President of Prose on Pixels (POP), Havas’ AI-powered global content at scale network.

For the head of a content at scale organisation to call for less may seem paradoxical but what he’s calling for is an end to the firehose of slop, and a more intelligent approach to meeting escalating content demands better.

But before we even get into the route to better, he wants to make one thing patently clear: We must get over the hype around AI and origination.

“I think if one of our creative directors caught one of the creatives coming up with ideas for ads simply using unfiltered AI, they’d fire them because there’s a reason why you’re hired into a creative department. It’s because you are an innovative, lateral thinker.”

Don’t think Ward is an AI refusenik, “we’re using it everywhere, across the whole business, all the time”. Indeed, POP does have its own origination tool, Vermeer.ai, allowing controllable, high craft AI creation that guarantees consistency, brand fidelity and quality, responsibly.

This sits under the wider Havas Converged.ai strategy that brings together data, creativity and AI capabilities.

“Havas talks a lot about being the business that’s AI led, but human curated”, Ward suggests. “We haven’t got an army of people just spitting things into a machine and seeing what ideas come back.

We still have experienced, innovative creative teams coming up with ideas with AI helping them filter out the noise.”

Part of the challenge with the current use of AI is too many people asking ‘AI is the answer, so what is the question?’. It’s back to front.

Instead, Ward insists, start looking at the content operation from beginning to end. Where are the roadblocks? Where are the inefficiencies, the lack of efficacy? And why is it all costing so damn much?

The sky isn’t the only limit for AI

From running ‘synthetic audiences’ where POP can use AI to identify a customer persona, develop a pencil portrait and then bring them to life and actually ‘talk’ to them (“it’s synthetic and not 100% accurate, but way more accurate than not doing it at all”), to hashing out long copy variations the team can then refine, to cutting a three-day, fully crewed international photoshoot to a single day – there are incremental gains all over the place.

It’s about using AI to chip away at challenges and reshape process into more efficient workflows that support market-defining creativity.

It’s all about understanding where AI should stop and the human should start in the creative process and that comes down to developing the creative too.

With brands on a global stage, there is a need for localisation. For it to resonate, there needs to be relevance in every aspect. Not just the copy but the locations, the sets, the clothing – the people. The more you localise in the traditional sense, the more time consuming and expensive it becomes.

With AI in the mix, there is the potential for infinite localisation – but again, there are limits. Not necessarily in terms of cost this time, but in credibility.

AI avatars are commonplace but they still tip into uncanny valley territory. Despite the obvious logistical and financial savings, creatives realise this is where they lose their audience.

“More and more, we’re seeing – quite rightly – that the industry doesn’t want to use synthetic humans on film. We want authenticity. We want real people,” Ward insists.

One example in how they might strike the balance, according to Ward, is using human actors, and then using AI to localise the environment around them. “An AI environment led by an AI artist is still the machine doing the work.

But if you’re if you’re shooting a human performance and pulling them into a generated environment as if they were shot on a green screen, we’re still filmmaking.”

This approach takes targeting to a new level. Not only can you now talk to a 30 year-old father in LA, Montreal or Atlanta in a more targeted way, but you can now talk to the same 30 year-old in Montreal who also happens to be part of a Mandarin-speaking community.

“If you wanted to talk to Mandarin speakers in Montreal, you never would have hyper targeted before, because it would just be too expensive. That shift alone makes the impact of AI on audience-first thinking undeniable – it changes what is viable, not just what is possible.”

Fix it in post?

The problem with expanding the realm of the possible is that people rapidly expect it to do the impossible. This is where using AI gets us into the realm of unintended consequences.

Shooting that ad for the father in Montreal becomes a template for other versions. Procurement gets excited – we can make umpteen variants for a fraction of the cost, we’ll just tweak it in post-production.

Then you realise there are still hurdles; the size of product pack you used can’t accommodate the headline in a different language. The lip sync you’re trying to overlay doesn’t match the actor’s mouth shapes.

Suddenly, you’re fighting a bunch of fires you didn’t even realise you had lit. “It’s like painting the Forth Bridge, when you’ve finished fixing one thing, you find something else,” Ward sighs.

So, once more, before AI can really deliver on its promise in the content pipeline, there needs to be a great deal of strategic upfront planning, human creativity and problem-solving.

A male actor filmed in real life speaking German, for example, will move their Adam’s apple much more than in other languages. The creative solution? Not spend hours blurring it in post.

“We actually put the actor in a polo neck to start with…that’s true production problem solving at work – AI won’t think about that for you”.

This is a critical point, not just in avoiding wasted work using a tech that was supposed to make everything more efficient. It becomes a critical procurement and finance issue.

There is no getting away from the fact that clients want agencies to do more with less. Or at least the same volume and better for less. The argument is that AI is taking up the human slack so the agency cost must be less, too.

The above is a prime example of why that is simply not the case.

“We’re in the midst of a revolution, which is bringing efficiencies for sure. Clients are rightly demanding to benefit from them and so they should. There are loads of things our business and our agency partners did previously manually, which we no longer do.

However, it hasn’t simply made us 30% cheaper, I think it’s probably making us cheaper but way, way more impactful. What that costs, I think we’re still trying to work out”, Ward admits.

“But then there are whole new problems created by AI,” he adds.

Whether those are the series of unintended consequences that mean lengthy ‘fixing it in post’ sessions, or because it’s positively impacted the campaign by revealing new opportunities such as hyper targeting, correlating agency remuneration to use of AI is far from simple.

“You can go down some mad rabbit holes. Developing new campaign ideas, new capabilities that before AI, you never even thought you could do, hyper targeting – that’s both an opportunity and a challenge.

You don’t need the same levels of targeting for a private jet customer as a toothpaste one so there’s also the conversations to have around ‘just because we can, should we?’.”

AI may be machine-based but it behaves like a living, breathing thing. It evolves and it makes the people around it evolve too. “Our AI is everywhere in the business”, Ward concludes.

“We weren’t around when the wheel was invented, but for me it feels like the closest analogy.

Just like the innovator who thought ‘put that wheel on’, I think there are niche areas popping up saying, ‘I’ve been thinking about using the AI platform that was built for that other purpose and now I’m going to take it into my area of the business and use it for this too’.”

To keep up with this evolution, agencies and clients alike need to flip their approach from ‘AI can’ to ‘I wonder if WE can’ and then explore the options, all the time applying some very human thought to the process.

“My old boss always said, ‘do interesting things and interesting things will happen to you’. Well, bloody hell, this AI revolution is interesting for sure.”

Paul Ward is CEO UK & Global President of Prose on Pixels (POP), Havas