Catching the Generative AI Wave
Not sure if you’ve been paying attention, but it’s happened again. Yet another, seemingly magical, Generative AI (GenAI) tool has just been released - this one’s from OpenAI and it’s called Sora.
Sora can render full HD video assets using only basic text prompts. It’s remarkable stuff; you can check out some examples on their site. But as GenAI continues to explode in a myriad of creative directions, it’s worth stepping back to consider how (or even if) all this innovation can be meaningfully and practically applied to our brands.
But before we dive in, a quick point of order. GenAI we’re talking about here specifically refers to the creative AI tools (both the large language models and visual programs) that you’re seeing popping up all over the place. It’s not hyperbole to say AI in general has already changed the world; it powers much of our digital infrastructure, systems and workflows; you probably found this article thanks to one. But the creative AI tools are, in many ways, still nascent technology from a brand adoption point of view. At least in Australia. And that bears some consideration.
The first thing to know is that many of the major global brands have already started to experiment with, or wholesale incorporate, GenAI to drive what’s been dubbed ‘synthetic asset production’. This is happening because, once trained, the GenAI tools provide unparalleled speed and efficiency in the creation of new assets. No reshoots. No locations. No need to re-book talent (see the section on AI ethics later in this piece). You can now create virtually endless sets of assets (stills and videos) based on your trained models. Just imagine - you’d never need stock photos again!
But how good are the assets? Tests conducted with Google Lens show that synthetic assets are already indistinguishable from actual assets up to 99% of the time. So not even Google’s own AI can pick up when it’s being shown a GenAI asset. But before we all go skipping off to our nearest GenAI tool to start making stuff, there’s an important distinction to note. The GenAI tools that brands are using to do this are not the off-the-shelf versions you and I can access online.
GenAI Maturity
There’s a big difference between experimenting on your desktop and plugging in GenAI to power commercial creative asset production. To do the latter, you really need two things - a brand proprietary version of the creative tool you’re using, and someone who knows how to drive it. The first is something you can only develop through experimentation and trial. The second is a person who’s skillset most clients probably don’t have inhouse, yet. In the same way we have retouchers who are the masters of PhotoShop, GenAI will need its own personnel (we just don’t have a name for them yet).
The adoption and eventual application of GenAI is a certainty for most clients. The question is how and when. Oversimplifying, the process looks like this:
Brands can only realise the significant benefits of GenAI once they reach the second stage. But we’re not all moving along this path together. When a brand reaches GenAI maturity solely depends on when it starts its journey. Those that start first, get there first. So start! Look for ways to experiment, even on small scales. The more you learn, the faster you’ll move.
What GenAI is Not
Yes friends, you heard it here first - GenAI can’t do everything. Sorry. And there are still some issues relating to its application that brands should definitely consider before they start rolling out assets. So, in no particular order, here are a few important things to remember about GenAI:
It’s Not Strategic
GenAI cannot think for you, and it cannot provide strategic or creative insights to help you outmanoeuvre your competition. You can try this for yourself. Log on to your preferred large language model (eg ChatGPT) and ask it to give you a marketing plan for your brand and your biggest competitor. Ask for specific strategies to deposition them. What you’ll get back is very generalised marketing information. Same goes with creative. It can only make what you tell it to; TL;DR - you still need to come up with the ideas. Your instincts, creative skill, and brand/market experience will remain as critical as they have always been.
It Won’t Challenge Your Brief
This is really important. GenAI tools are incredibly powerful, but they can only create what you ask of them. No-one likes it when their brief (or creative work) is challenged, but it’s a necessary part of the process. It means concerned brains are interrogating what’s happening; in general, you want that because it leads to better work. GenAI will never tell you what you’re making is a bad idea, which means somewhere in the process humans need the ability to do this on the AI’s behalf. Don’t skip it.
The Ethics are in Flux
Also something you should really think about. There have already been multiple legal cases regarding misappropriation of creative work. And there was the most recent Hollywood Writers’ strike regarding the use of ChatGPT. But in addition, commercial GenAI works by first training the AI models on subjects - both products and people. If you’re using people, what are the royalty arrangements moving forward if you continue to use their likeness? To what degree are you comfortable manipulating their images to create ‘other people’ in new assets? These are conversations that brands need to have well before assets are rolled out.
It’s Not Always Correct
This one might be obvious, but it’s worth repeating. Tools like ChatGPT are quickly becoming the go-to option for desktop research because they can draw on such large data sets. But you should always verify those data. I recently asked one to write my bio for me, and while it pulled all the relevant info, it mixed up what happened at which agency. If you’re using GenAI for this purpose, no problem, but treat the source validity in much the same way you might a Wiki search.
The Bottom Line
Like Photoshop did with film, GenAI will undoubtedly revolutionise creative asset production in advertising and marketing. I can say this because it already is overseas. If you follow any major brands on socials, it’s entirely likely that you’ve already seen synthetic assets and not even realised. The question for Australian marketers is when? Sooner or later, someone in your competitor set will make the jump. When it happens, those that haven’t will only then start their journey toward GenAI maturity (most likely, in a scramble) while your competitor is already there, enjoying the benefits.
Unless of course, you get there first!
