Coming from Tim O’Neill’s background in running one of Australia’s leading digital agencies – Reactive – selling and then meeting Jason Ross at Accenture, they’ve jumped to now officially launch a next-gen generative AI agency called Time Under Tension.
While there is much AI hype around creation and ideation, there is also much caution and concern.
Anton talks with Tim and Ross about where and how to test generative AI, how to educate and engage conversations across an organisation in terms of usage rights, copyright protection, contracts, and cost management, brand proof of concept use cases, and the gains to be had in harnessing the most suitable AI tools.
You can listen to the podcast here:
Follow Managing Marketing on Soundcloud, Podbean, Google Podcasts, TuneIn, Stitcher, Spotify, Apple Podcast and Amazon Podcasts.
There are eight good tools that you could use. They each have their pros and cons.
We’ll help you evaluate which one is best for your needs, and then we can provide outside of the design workshop training on how to use that tool.
Transcription:
Anton:
Hi, I’m Anton Buchner, senior consultant at TrinityP3. Welcome to Managing Marketing, a weekly podcast where we discuss the issues and opportunities facing marketing, media, and advertising with industry thought leaders and practitioners.
Today, we’re talking again about AI, however, honing in specifically on generative AI and where AI meets design and user experience.
And whilst we all know there’s a lot of AI hype around creation and ideation, also a lot of caution and concern and watch outs.
My guests today are the co-founders of Australia’s first generative AI consultancy, Time Under Tension, officially launched just over a year ago in early 2023.
So, please welcome to the Managing Marketing Podcast, Tim O’Neill and Jason Ross. Welcome, guys.
Tim:
Hi, Anton.
Jason:
Thanks, Anton.
Anton:
Good to talk with you. And I’m really interested to hear your views today. AI is such a beast or such a wild west, and I’m sure I don’t have to tell you guys that, but I’m interested to hear your perspectives.
Maybe before we jump in, Time Under Tension, what’s behind the name? How did you guys meet each other? I know you’re ex Accenture, so maybe quickly backwards to where you’ve come from and where you’re going.
Jason:
Yeah, sure. So, I first met the Tims around 15 years ago. I’m an agency guy by background and I was running digital agencies out in the UK.
I was looking to make a move to Australia, and that’s when I was put in touch with the two Tims. Tim O’Neill here and his business partner back then, Tim Fouhy, who were running one of Australia’s most successful independent agencies called Reactive.
So, I was really excited when they offered me a role and to come and run their Sydney office, and we’ve pretty much been working together ever since then.
Tim has a interesting story with regards to Reactive, which he ended up selling to Accenture. And at that point, we parted ways. I think that was around 2015 or around 2016.
And we stayed in touch, obviously. I stayed in Accenture. I then ran frog at Capgemini, a design agency.
And around 2023, when generative AI sort of sprung onto the scene, we started to ramp up our conversations about how excited we saw what’s happening out there and how we believe that there’s going to be a whole raft of new experiences being built on this tech.
So, we got very excited and we formed Time Under Tension and launched a business back then. And with regards to the name and a bit of background to himself, I’ll let Tim cover that.
Anton:
It’s a great name, Tim. I’d love to know Time under Tension. Tell us more about that.
Tim:
Yeah, thank you. Well, as Jason said, so there are three co-founders, Tim, and Jason, and I. When we decided we wanted to set up an agency focused on this area and generative AI, it was I’m part of an industry group called SoDA, the Society of Digital Agencies.
And there was this totally unrelated thread. And someone had dropped in the phrase and not talking about generative AI or AI or anything like that, but they’d used the phrase, time under tension, which I’d never heard before.
And I literally just replied to the guy that said it, this guy in the US, an agency owner, and said, “Hey, that’s a great name for an agency.”
And then a couple of weeks later, Jason, and Tim, and I are having lunch and going, “Okay, we need a name for this agency.” It’s like, “You know what? I’ve got this name for an agency just in my back pocket.”
And we were like, “Yeah, actually, it’s got a nice ring to it, but how does it relate?
And ChatGPT had just been launched and we put into ChatGPT, “We are going to launch a generative AI agency. Here’s the name, can you relate it to this sort of topic area in some way?”
And it came back with what we thought was an amazingly good justification for the name against generative AI and around a metaphor of so time under tension is a physical training technique or basically, you do strength training, you put your muscles under tension for a certain amount of time, and you build strength and agility.
And there’s a metaphor there about training AI models. So, you put time and energy and resource into training an AI model. And you do that once, and then from that AI model, you get power and agility and speed like an elite athlete.
And we’re like, “Okay, cool. That works for us. So, there’s the name.”
Anton:
Got it. Stress testing. Excellent. And you’re right, I mean-
Tim:
Bit of reverse engineering on the name.
Anton:
Oh, that’s good. I thought it was Tims Under Tension, but it’s Time Under Tension, so that’s good.
You’re right, the ChatGPT exploded over a year ago now. Obviously, AI had been around for a little while before but certainly as society and marketers jumped on the bandwagon and we’ve had a myriad of solutions being launched at seems every day since.
I’m really keen to hear your view of sort of where you’ve honed your agency and where you’re working in when you say generative AI. Whether it’s copy, whether it’s imagery, whether it’s video. What sort of creation are you focused on?
Jason:
Yeah, that’s interesting, Anton, because I’d say over a year in now, things are still changing for us as a business.
I can tell you that when we set the business up, we were sort of so excited and convinced that there’s going to be a whole raft of new experiences that are generative AI enabled. So, conversational interfaces, conversational commerce.
And we really had a view that will set the business up as a build agency, which can learn and understand how to build using some of the early stage APIs that were emerging back then from OpenAI and how do you actually build stuff.
So, we did a whole bunch of R&D back there to A, educate ourselves and then B, get a point of view on how easy is it to build on them.
The reality, fast forward a year has been that actually, we’ve been lucky enough to be building, but we also, had to do quite a lot of education around those services as well, which is something we underestimated at the beginning.
A lot of our clients, a lot of brands, a lot of companies are still trying to understand just how broad is the impact of generative AI on their business. So, we’ve ended up alongside building stuff. We’re also, helping educate and inform clients on what generative AI means.
Some cases we’ll walk into boardrooms where board members don’t even know what the potential use cases for generative AI are. So, we sort of have to go from zero all the way up to a hundred in the space of an hour sometimes to try and do those education sessions.
Anton:
Yeah, I think that’s interesting because we’re seeing across the board whether it’s businesses we’re working directly with or obviously in the media, brands testing.
So, whether it’s literally millions and billions of variations, data driven, trying to be personalized at scale. But whether that’s copy only or it’s imagery and copy, that sort of lower level creative testing seemed to be the first stage and first place that brands played in.
And then of course you’d know that we started to move towards imagery and video and that brought with it a variety of dangers and challenges. Is it on brand, is it off brand? Is it licensed, is it not licensed?
And we’ve seen the train wrecks on royalties and licensing issues TikTok facing at the moment with Universal.
And restrictions now, starting to creep in I think can make legal teams a little bit nervous around how far can marketing or brands take this in terms of generative AI.
What are your views on where that’s currently going? Is it still in that hyper testing hype stage and brands are getting away with testing as much as they can?
Tim:
We see huge amount of variety of the level of understanding and maturity in businesses with generative AI, which is not sort of really that surprising. It’s all very new and there’s new things that are coming along all the time.
But I totally agree, like content copywriting is the most mature area and probably also, one of the safest areas for people like a brand to experiment in. There are definitely things that a brand or an agency need to be aware of. But it’s a good place to start.
In terms of where those use cases are. There’s copy writing, copy generation, there’s testing against government rules and secure safety and so … sorry, by what … sorry, I’ll scratch that. Stop.
Anton:
All good.
Tim:
Okay. How far should I go back?
Anton:
As far as you want. Want to start that again, or?
Tim:
Okay. So, what we see is there’s big changes in level of maturity within agencies and also brands of their understanding of generative AI and how far they’ve progressed so far. And totally agree, content creation is a really good place to get started.
And so, most agencies and most brands have at least sort of experimented and piloted, even if it’s just using ChatGPT to assist with copywriting is probably the very first thing that people can do.
But in terms of image generation, video generation, there are lots of pitfalls. And I think agencies tend to be a bit further ahead than brands and experimentation.
They’re probably a little bit more open to the risks of it, less likely to have an IT department locking stuff down than within a big corporate. Which is sort of a good segue into some of the things that we were going to talk about and from the procurement side as well.
Anton:
Yeah, that’s a good segue because I think our biggest learning or biggest insight from helping clients assess AI or look at AI solutions has been what is the agency actually offering? And I think that’s been unclear.
Agencies are learning along the way. The amount of experts that have popped up on LinkedIn now, that are suddenly AI experts, I think it’s clear that it’s a new space, a new frontier, everyone’s learning.
But to procure those resources, typically, we’ve worked with procurement departments that are very tight. Very tight scope of work, very tight allocation of resources, very tight licensing agreements around technology.
This space doesn’t seem to have that yet but it’s starting. We’re starting to see restrictions in contracts where marketers are trying to prevent agencies going too far around use of data.
What are you guys seeing from your side of the fence around this?
Jason:
Well, we’re seeing a whole load of different behaviors from brands. We see, for instance, some of our global clients, they have global policies in place, which have just done blanket bans on any form.
I was just at a client yesterday and they’re the Australian subsidiary of the global brand, and they’re not even allowed to use ChatGPT or anything at all, basically. So, they’ve blocked everything down. They have some plans to release some tools, but they haven’t sort of socialized that at all.
On the flip side of that, some of our clients are engaging us to jump straight into to do proof of concepts and experiment with the tech and try and figure out as they go, sort of build that plane and fly the plane at the same time.
What does responsible safe usage mean as we’re building those proof of concepts out?
So, we’re seeing a range of different behaviors out there. I think what sits across all of them is that the education that’s required to shift an organizational sort of attitude towards generative AI, is much bigger than probably any of them have estimated.
This is something that at the end of that, it impacts every member of staff, whether marketing or otherwise.
So, there’s a huge burden of responsibility on an organization to increase awareness of usage of the tools that are available to staff.
Because even with those blanket bands, you can almost guarantee from the anecdotal surveys that we run in all the sessions we do, we always ask how many people are using generative AI in their personal lives. And that’s 90% plus have tried and played.
And then you sort of ask that from a business lens, how many of you use it in business for your work? And it’s maybe 1 in 10, it’s the sort of converse of that, but-
Tim:
1 in 10 admit to it.
Jason:
Admit, exactly. So, the burden of education is I’d say one commonality that sits across a lot of the clients that we work with.
Anton:
Right. And I mean, of course, you’re starting to see the moral and ethical issues as well around AI. So, we are seeing that the different departments are now, getting engaged. We’re seeing legal heavily come and look at it from a contractual point of view.
We’re looking at the ethical moral side of businesses from the c-suite down, starting to say, “Well, actually, do we want to be positioned as a brand or a company in this space and utilize these technologies?” That’s another big question.
And then even if we are, what are the ethics or morals of whoever coded the original system, whether it’s through an agency or an off the shelf solution?
Tim:
Yeah, that’s really interesting. We just, prior to this podcast, we had a call with the client and they have a whole lot of different use cases that we were discussing around generative AI and how they could put this into practice.
And one of the things that the client highlighted was that in this particular area, there’s a lot of potential gains to be had, and it’s a really exciting, really good topic, a really good sort of place for them to play.
But they’re very cautious because of their role in the kind of creative community in Australia and the backlash. So, they can, and they could do this, but there’s potential backlash if they do.
And yeah, we’ve seen this with some pretty famous examples out of the US like the Under Armour ad from, I think it was only a few weeks ago, which was produced with the help of AI, and there was a huge backlash without people really understanding it wasn’t actually made with AI. AI was part of the production process.
And I read a pretty good full article about it, and it’s like that seems like totally reasonable that they would do this. They had all the usage rights, they had permissions from all the agencies.
But still, it’s a hot topic and some brands quite rightly don’t want to kind of wade into that territory too soon.
Anton:
Yeah. And it’s also, there’s good and bad uses. The deep fake area, which sort of exploded and brands have tested. Which morally could be wrong or morally could be right. You can argue on both sides of the fence.
We’ve seen David Beckham speaking millions of languages, and other brands doing hyper localization. Obviously, not true and as a customer or consumer, I can see it and go, “Well, that doesn’t seem right.”
But they’ve created at scale this volume of generative content which could never be done before. So, I get the good and the bad.
Tim, you touched on this, the education and you, Jason, as well. What are the biggest barriers in that training or that education type work you’re doing at the moment? What’s stopping people testing and going forward?
Jason:
So, well, quite a few. I’d say the first one, and I think Tim mentioned it, is that agencies are a little bit ahead from the experimentation standpoint than potentially brands are.
But there’s limited understanding from the top down about what’s the approach you take with this? Do we unleash the tools to the organization and see where it lands? Do we need to lock things down? So, I think there’s a general understanding facts that-
Anton:
Sorry, Jason, I lost you on what’s your approach. You cut out.
Jason:
I might-
Anton:
Lots of brands, lots of approach.
Jason:
Yeah. Sorry, I’ll start that bit again. So, the question was, okay, what are the challenges that we are seeing on the education side of things?
Anton:
The education front, yeah. What’s stopping people going forward?
Jason:
Yeah. So, I think what’s stopping people, what organizations specifically in pushing ahead on the education front is an understanding from the top down about generative AI.
I think we are seeing, when we walk into boardrooms or c-suite meetings, that there’s a very mixed understanding.
There are some real advocates for generative AI in the room. And then likewise, there’ll be people that just think generative AI might be an avatar or ChatGPT. They have a very particular view of what it is then. So, I think there’s an alignment and understanding question.
And then there’s still it took us, I don’t know how long, I think you are both probably old enough to help me remember, but how long did it take for the W3C guidelines to come out back in the days of the internet?
I think there are still emerging standards, so there’s no one right approach. So, if you talk about, hey, how do we implement AI ethically in our organization, there’s no one approach.
There are multiple and a number of which has resulted in a bit of yeah, I guess over confusion about how to take that next step. I think that’s also, plays a role in there. Where do we start, basically?
And that’s where we have a methodology. It’s not the only methodology out there, but we have a point of view and an opinion on how to go about it.
And it starts off with a very wide and broad what we call informed process where we upskill as many people in the organization as possible about what is generative AI.
And that’s sort of the starting point for a lot of our engagements because we find that that’s necessary to get everyone onto at least a similar baseline of understanding.
Anton:
Fair enough. So, it’s a challenge to get them aligned. If we go through each of those teams and departments, so you’re noticing legal, for example, are they fairly opinionated and voicing concern around copyright infringement or copyright protection? Is that becoming a hotter topic?
Jason:
Tim ran an interesting session specifically on the topic of IP. I would say, just to answer it very quickly, probably relating to the education piece that I mentioned, we don’t see a lot of challenges coming from the legal team.
Admittedly, there should be, and they’re not even asking the right questions, I would say at this point.
Tim:
Yeah, I totally agree with that. I mean, there are different kind of considerations with if you’re a brand or an agency, you are less … let’s talk about it from a brand point of view.
If you’re a brand and you are using generative AI yourself personally or via your agency, there are so many considerations. Like the benefits are pretty well known. Like people can see, “Okay, cool, this will be unlock more productivity or allow us to do things that we could never do prior.”
So, for example, the Under Armour ad. The talent was not available to do any new shooting. So, they’re like, “Well, what can we do create new with existing footage and with AI?” And that’s what they did.
So, being able to create new experiences or new content that you couldn’t before, as well as the productivity benefits, that’s all kind of pretty easy to understand.
But what brands not really as aware of is all the other considerations and the potential risks. So, there’s some that are reasonably well known, so data security and privacy.
So, there’s some really famous case studies with people using ChatGPT and then that content going into the training data. So, most people at least vaguely familiar with that. Definitely IT teams are familiar with that.
From the legal side, there are some considerations in particular around copyright potential infringement of someone’s copyright or your ability to protect the copyright or the IP of things that you create with generative AI.
And I, as Jason said, don’t think the legal teams really have the head around the technology side of it to know what the correct answers are.
So, that’s part of the training that we give, which is to the marketing team normally, so that they’re aware. If the marketing team are aware of these considerations, then at least if the legal team come and ask them, they have an answer and they’re doing the right thing.
Anton:
Yeah, yeah. We’re seeing similar and same as you mentioned before about procurement. So, procurement get excited when they see cost savings. And I think this is one big area which is a double-edged sword.
We’re seeing cost savings absolutely, as you said, speed, the ability to produce a lot of material at scale is then from a procurement perspective. Equating to, well, we should cut the agency fees, or we should start to look at lower fees from our agencies in terms of what they’re charging us.
One of the areas we’re working on with procurement is to say, “Well, that’s fine if that’s the angle you’re taking.” But marketing is also, looking at what’s the strategic use of AI?
So, it’s not about just putting lots and lots of juniors onto the business. It’s not about taking juniors off the business. It’s about how AI can be utilized from a strategic perspective and then the generative AI solutions that come out of that use case, or scenario, or problem-solving area.
But yeah, I think that procurement is definitely an area that has gone cost, cost, cost in our initial probably year of analysis. Have you guys seen the same or are you seeing an engagement with procurement more around value of AI?
Jason:
Well, first of all, I just want to share a quick anecdotal story with a client who I met with a couple of weeks ago. We’re talking about doing some proof of concept work for them.
But they were commenting on the agency networks that they’re using, very clearly saying, “I know that they’re using generative AI and I expect them to pass on the cost benefit to us. I know that they’re using it, we can all see it.”
And wasn’t necessarily interested in how the tools are being used but was interested in the dollar value saving that that should amount to for the brands.
Which I don’t necessarily agree with that way of thinking. I do think it’s part of the benefits that AI can bring are definitely around the optimization side. But that’s really just one side of the coin.
I think that the much more exciting side of it is like you were pointing out, so what are the new creative routes? What are the new experiences that can be created using this tech? And what are the new opportunities that it affords us? And it’s definitely worth exploring.
I’d probably say it’s unfortunate timing for all this tech to emerge. Like we’re going through global recessions are pretty much everywhere and cost of living crises, so there’s a lot of focus on optimization and efficiency plays.
I wonder if in better times, if we would’ve been a bit more focused on some of the better, stronger attributes that generative AI can bring.
Anton:
Well, conventional wisdom says spend when we’re in the trough. So, as a Ritson or a Byron would say to us, now’s the time for brands to stand up and really stamp their purpose, their mark, their positioning. So, it may be a great opportunity.
But I think that brings us back to that marketing department, that use, and you both touched on this. Marketing’s excited. It’s everything from how do we manage our website and journeys better? How do we manage our customer comms better?
And then how do we connect experiences that previously were disconnected to now, connect them seamlessly and give customers, consumers a much more interesting experience?
We’ve obviously seen that around for a while, but I think this hyper opportunity is exciting marketers. You talked about the training for marketing. Where are you focusing on the training in terms of either quality, or speed, or opportunities?
Tim:
The training is focused on, like first thing we do is undercover what are the potential use cases? So, what are the challenges that they have in a workflow, or a task, or a process.
For example, is it pulling the insights out of a brief or is it responding with new creative ideas to the brief or how do the teams work together and identifying what are the potential opportunities that weird generative AI might help. But starting with challenges and opportunities.
And this is sort of a part of our process as a design workshop with whether it’s the marketing team or any part of the business, sometimes it’s the exec team.
But if we talking about a marketing team, with these design workshops, we then uncover 50 ideas. There’s never any shortage of ideas and we prioritize those with the client.
And at the end of the design workshop, there’s 10 to 12 use cases prioritized for, these are the different things that you could do with generative AI to help the marketing team be better or more interesting or more creative with their roles.
And some of those use cases can be solved with off the shelf product. So, it might be, “Cool, that’s a really good idea. You know what, there is a tool that does exactly what you’re describing. You might never have heard of it, but we have, and here it is, and here’s some guidance or training and how to use it.”
So, how to use … I mean, I won’t go into examples of tools, there’s too many. But image generation tools, for example, like, okay, you want to do image generation, there are eight good tools that you could use. They each have their pros and cons.
We’ll help you evaluate which one is best for your needs, and then we can provide outside of the design workshop training and how to use that tool like if it’s Midjourney, or Adobe Firefly, or whatever it might be.
So, there’s always huge amount of ideas and the ideas then, as I said, sort of run the focus training, which is hands-on training and specific tools as well as sort of looking at an entire process.
Anton:
And Tim, on that point, is there a way you’re helping prioritize or helping the marketers prioritize on the impact of those use cases or the value of those use cases?
Because we all know that it may be a tiny area or a small area to focus on first to prove it out and then business case it, and then seek budget from the CFO.
But what are you seeing with AI? Is it being a bit of a blanket solution to a use case, or is it taking one specific smaller area and trying to crack that first?
Tim:
I mean, so certainly when prioritizing the business value and impact is yeah, that’s one of those axes. And then a big consideration is how mature is the technology for solving that problem. That’s a kind of exciting thing.
But also, difficult thing with generative AI is everything’s moving so fast, it’s really hard to keep abreast of what works well now, and what doesn’t work well now, but what might maybe work well in the future.
So, like copy, for example, has worked well for the last 12 months or so. Images have worked well for the last six months or so. Video doesn’t really work well now, but soon, like we’re seeing hints of Sora, for example.
So, they’re moving at different speeds. And there might be a really great use case that comes out of a client workshop that has huge business value, but we have a pretty good sense of going, you know what, let’s not try to tackle that now.
Like if we make six months, there’s probably going to be better technology so we don’t have to custom build something that will be done within Microsoft Copilot or done for Adobe Firefly in the next few months.
So, that’s a big consideration is the business value and the maturity and also, the maturity of the general technology to meet the use case.
Anton:
Yeah, I think that’s a good point. And I’m hearing as you talk the concept of personalization. So, when you say video might not work, or we know copy might not work to that degree, it might not work for one customer, of course it does work for another customer.
So, training marketing teams on which customers are engaging with this, or in what way are they engaging and therefore, have you got the right benchmarks? And are you proving that you’re lifting cohorts or segments or individuals, whatever the objective is, I think it’s going to be critical.
Otherwise, it just becomes, well, here’s another implementation of a technology that’s interesting and fun and we did some really good stuff and won some awards. But did it shift the dial on business, customer value, customer profit?
Have you seen much around the benchmarking and putting the metrics and some rigor in place with clients? Or is it still a bit exciting and test and learn?
Jason:
I think on the part of our helping clients prioritize is, as an example, can we take an existing process where we might be able to measure certain elements of what’s the speed of, say, creating a campaign without AI?
And are we able to then introduce AI components into the workflows and clear metrics around what sort of efficiencies did it drive?
I think it’s still quite early on the sort of creation of new experiences and delivering sort of incremental value rather than efficiencies.
I think it’s still early days to be able to see that because there’s a whole bunch of other considerations like consumer trust in speaking to an avatar as an example. It’s very early days to see whether people love, hate, or just need time to get value out of these new things that can be created.
But yeah, that needs sort of trying to attribute metrics is definitely part of the value definition process.
Anton:
Yeah. I had an interesting chat with a client the other day and they were talking about their chat bot and saying no millennials want to use the phone anymore.
And had to pull them up and say, “Well, yeah, maybe the majority of the sentiment is that, but some do. If you’ve got a very clunky web experience, then some people do want to pick up the phone and still speak to someone as much as they might like to use a chatbot and search or ask or get some information that way.”
So, I think that, yeah, some of the risks I’m hearing in this is it’s not a blanket one size fits all. You’re right, maybe there’s efficiencies and productivity gains on one side, but will get to value, and ethics and morals, and pretty more rigor around contracts on the side.
Jason:
I like Sora’s new tool from OpenAI for video creation, and they smartly released that to a bunch of creatives as opposed to sort of letting it go crazy, which I think was a really smart move to show what can be done with the tool in the right hands.
And there was one of the videos in particular Air Head, which I think we shared through our LinkedIn because we very impressed with the storytelling that they achieved there.
The behind the scenes insights into how the directors and producers and creatives thought about building that out was a true sort of human plus machine can create a result, which is exceptional and emotive and powerful for …
So, what a new way that brands can think about generating content, basically. Where I think the climate that we’re in at the moment, again, maybe it’s because of the economy, but maybe it’s because it’s the lower hanging fruit, really around efficiency.
So, instead of talking about what’s the new type of creative we can put out there, it’s how might we chop down the creative production process, video editing.
So, I think yeah, having that education piece is so important because without it, some people in the business will think it’s just about one or the other and not necessarily see the full potential.
Anton:
Yeah, yeah. And I think like many other mediums or other channels, there weren’t many restrictions on channels. If I go back to magazines or outdoor or even TV, there weren’t restrictions for brands. So, brands could test and do anything to try and make impact.
I think we’re excited in a weird way to go, “Well, we’ve got to be careful of this AI thing. We don’t really understand what’s in the black box and we’ve created a bit of a monster.” It’s the Frankenstein effect.
Versus saying, “Let’s get it out there. Let’s test some things, and sure, there’ll be some fails, there’ll be some shockers, but it’s a great chance to push some boundaries and maybe we just can’t see those boundaries or what’s beyond the boundary in the coming 12 months.”
Tim:
Yeah, I think that’s super interesting, the boundaries. There are so many gray areas and the gray areas are moving like so quickly as well.
There was a really interesting article, I think it was a couple weeks ago in Ad Age in the US, which was talking about how it was kind of agencies complaining a little bit about this trend they’re seeing in procurement processes from brands, and that the brands are effectively banning AI for the agencies without express permission.
So, during a contracting process, the brand will say, “You’re not allowed to use generative AI in any part of producing work for us, unless you ask us first and give permission.”
And interested to hear from a TrinityP3 hat on, if that’s what you’ve seen happening here in Australia.
Anton:
We definitely haven’t seen it to creep in all discussions. It’s popped up once or twice, but it’s not anywhere near being embedded in thinking. But it’s logical.
So, it’s one of those areas that we see it as it’s going to be a logical next progression. Whether it should be banned or not, we don’t tell anybody what to do, just to be an independent arbiter.
But our view would be that you need to have some serious discussions around it. And if one group is saying it needs to be banned, well, why does it need to be banned? From what perspective are you concerned? Is it legal, or moral, or ethical, or a cost perspective?
And then from an agency perspective, well, is it a way of delivering better creative? Is it a way of us improving and delivering an even better product? Is it a way of measuring better and proving we can do better work?
So, there are arguments around the whole sphere. I just think it’s one of those areas that we need to approach it materially. We need to obviously have all the watch outs and concerns raised but no one’s got the answers.
So, there’s no crystal ball saying AI will be like this in 12 months time. I mean, we haven’t even got social media guidelines yet, and it’s about 15, 20 years in. So, I can’t see AI getting regulated to that degree.
But look, you’re right, US is starting to talk about it. But the US is on a different path as well, on a bit of war path with China in terms of ownership over social media platforms and all sorts of other risks and challenges.
So, I think we just need to be careful in terms of how we’re assessing and weighing our pros and cons. But I’m sure legal will get their teeth into it, Jason, as you said. Not quite yet, but I can’t see it too far away before they’re at the table asking more questions.
And IT are there already from a data perspective. That’s probably the one we are seeing the most with data breaches, data leaks, and you talked about this earlier, the lack of trust from consumers.
I think my name has been taken, I don’t know, 15 million times, stolen and taken and used, and God knows where Anton Buchner has gone. But do I worry about it? Not so much. I’m out in the ether. My data is out there in the ether.
But that is an issue for brands in terms of, again, how much consumers are going to give over to their footprint to sit in an AI engine, or am I going to start wiping my footprint and saying, “Well, actually within 14 days, or 7 days, or 24 hours, I don’t want my answers or my use of AI to even be kept.”
So, again, who’s going to have the power of this? Will it be consumer driven, customer driven, or will the brands and companies set the benchmark? That’s all to play out.
Tim:
Yeah, I think one, is that there’s more awareness needed on both sides of agencies from the top down, educating their teams on if you are doing some copy creation for this client, make sure that you’re using the right version of ChatGPT if you’re using ChatGPT.
And you’ve got the data security stuff switched on to the right level so you’re not leaking the confidential brief into the training data or image generation for example.
A lot of people don’t know that if you’re using Midjourney, which is really popular in agencies because it’s amazing in terms of image generation. But by default, every image that you generate is publicly available.
And now, with Midjourney moving out of Discord and onto the website, it’s actually even easier to find the library of other people’s create prompts and the generations.
Just this morning I was experimenting. Super exciting, just got access to the new Alpha Web version of Midjourney and playing around with that and then realized, “Oh, there’s a library in there.” Which has actually always been there, I’ve just never bothered to really go and play in the library.
It’s a little bit fiddy, but basically, I accidentally came across a very famous Australian creative director, highly awarded, works at one of the Australia’s best creative agencies. And it’s his entire Midjourney history for the last year.
It’s like, “Oh, I can see exactly what that person’s been working on for the last year.” Literally view every single image that they’ve created in Midjourney, I can see. So, assume that they’re doing a pitch for this brand, and here’s all the images, here’s all the comps they’ve done.
So, people don’t know that there are these gotchas, some of them are well known, like ChatGPT and others are much less known. And agencies need to be aware from the top down and clients need to keep agencies on their toes about this.
Anton:
Yeah. And asking the right questions. And I think that is the conundrum on both sides. Agencies absolutely need to bring to the table the watch outs, the transparency. We’ve had challenges in media before around transparency. So, I think the agency landscape has learned from that.
And then clients, who do you trust? Who do we turn to? Do we trust a Meta? Do we trust a Google? Do we trust the agency in what they’re telling us? So, it’s just about probably pragmatically building a path together.
Something TrinityP3 always talks about, shouldn’t be either or. Marketers shouldn’t tell, agencies shouldn’t tell. There needs to be good, solid, robust conversations and then decisions to move forward together.
What about a final takeout? I think you’ve touched on it, but is there a one piece of advice you’d say, Jason, for brands thinking about this space, what would you leave them with?
Jason:
Oh, well, we open up all of our presentations with a very positive quote from Bill Gates who invested heavily in sort of OpenAI. We feel quite positively about where this technology can take us.
I think the takeout for brands is that you need to be starting somewhere. That can be very different for different brands.
Definitely think that we need to take the conversation away from the technology people, Silicon Valley and all those that are building these large language models and start ingesting that and defining what it means for us.
I like to reference the CSIRO’s AI ethics frameworks. It’s been around for, I think, over five years now, and largely ignored by most people, certainly marketers. There’s been no need to look at them.
But I think that having a look at those principles and defining what that might mean for your brand, for your organization as you move into that world, what does transparency mean? What does accountability mean?
And be social with that, share that with your staff, share that with your agencies and plan a path forward together. Yeah, I think there’s a lot of exciting potential.
The optimizations are definitely there, but I think the ability to create new experiences and things, surprise and delight, taking things back to what marketing is always about, trying to create emotive connections and getting people to act on our messaging. I think there’s a lot of potential to create some exciting experiences on the tech.
Anton:
Right. It’s exciting times.
We don’t normally always do a little plug, but anyone listening, Time Under Tension, get their perspectives, get their views, touch base with Tim, Jason, and their team if you’re starting out in this area, or you are partly down the track and wanting to ask questions. Sounds like pragmatic sage advice.
Thank you, guys. Thank you both for your time. Tim, Jason.
Tim:
Thanks very much, Anton. Pragmatic from sage. I love it.
Anton:
Yeah. I don’t know which one’s pragmatic or sage. I’ll let you find it out.
Tim:
Sounds like a brand name for some candles, maybe.
Anton:
Look, we’re out of time, but if you’ve enjoyed listening to this, please like, review, or share this episode and spread Tim and Jason’s words of wisdom.
But guys, one final question. You talked about three co-founders. I’m wondering, is the fourth going to be a part share from your chat bot?