Fear not, Creatives, AI comes in peace

This week on IS THIS AI?, we’re talking creativity, why 90% of everything is cr*p, and whether AI is here to help or to quietly remove you from your studio.

We’re joined by Derek & Norts (The Peeps) who have been experimenting with machine learning and creativity since before AI tools got slick (and same-y).

This conversation gets into the uncomfortable, strategic, and sometimes joyful reality of creative work alongside AI: where it’s useful, where it isn’t, and what still makes human creativity distinct.

 Tune in for -

1. A more grounded take on the “AI is coming for your job” narrative.

We talk honestly about where the pressure on creative roles is really coming from (hint: not the models), and what agency leaders need to plan for.

2. Where to start with AI and how to avoid the slop

Practical ways to keep your work distinctive when the tools flatten taste and style. This is about creative judgement and filters—not prompts.

3. Where the real creative opportunity sits now.

Why the interesting future is small, strange, hands-on, and niche. And why you should be looking for the things AI enables that were impossible before.

Book tickets to the BIMA AI Roundtable with Be Kaler Pilgrim on AI & recruitment – 19 Nov, 10am UK: https://www.eventbrite.co.uk/e/bima-ai-peer-network-roundtable-tickets-1870338014349

Transcript

Hello, welcome to episode seven of Is This AI? We humans on the BIMA AI Council are back to bring you more myth-busting insights, practical ideas, and a little more clarity for the road ahead. I’m Olly Veysey, co-chair of BIMA AI Council, and I’m joined here as always by Lisa Talia Moretti. Lisa Talia, hello.

Hello, hello, hello. Hi, how are you? Good.

Before I am well, thank you. I am well. Before we get into our episode, our episode six with the brilliant Be Kaler Pilgrim about recruitment really struck a chord and I know you’ve got more to share on that.

Exactly. So the BIMA AI Roundtable is on the 19th of November, which is a Wednesday at 10 a.m. UK time. And Be Kaler Pilgrim will be leading that session and sharing more details on AI and recruitment. So please join if you can and we will add that link to the show notes so you can sign up.

Fantastic.

On to the headlines. It’s been a big week. Which AI story has caught your eye?

It sure has. I mean, we keep hearing that agents are the future, but now some of the biggest technology companies are saying, not your agents, not on our site. So Amazon has sent a threat to Perplexity demanding that the startup block its AI agent from shopping on Amazon’s website on users’ behalf, which I have to say I didn’t see coming, but when you kind of unpack it a little bit more, right? And you start to think, well, agents aren’t susceptible to temptation. Agents maybe aren’t susceptible to mindless buying, aren’t susceptible to some of the perhaps dark design patterns that we know are on some of these retailer sites, which means smaller baskets, which means less revenue. So Amazon are not happy.

Amazon versus the agents was a really interesting one, isn’t it? We’ve been talking about it in relationship AI optimisation for search but user experience and how agents are going to interact with that and we’re finding out. They’re not, as you say, susceptible to the ways that we humans have been encouraged, shall we say, to fill our baskets in the past.

What’s your headline?

I couldn’t go past the news that Stability AI scored a 3–1 victory over Getty in the UK High Courts, which for anyone interested in AI and copyright, which is, you know, anyone who’s ever created anything, as well as their record labels, distributors, film. Everyone’s had their eye on this case making its way through the courts and the precedent that it will set.

Wow.

For those people developing AI and for people who rely on copyright to protect their livelihoods. Getty had alleged that Stability AI in training their AI models on the Getty library had infringed their copyright and that of the creators on there. And the High Court judge found that the acts of copying took place outside the UK and that the AI model does not store or reproduce Getty’s images directly. So on a technicality, it’s not copyright infringement. So it’s a landmark case and it really just highlights the fact that the law is struggling to keep up in its current form with how generative technology is developing.

Yeah. And also reminds me so much of how important language is around these issues, how we define these different technologies, how we talk about what they do, using certain words and not using certain words. Turning images and words into code and then models looking for patterns in the code is not the same as copying and reproducing. And so, yeah, our language as well as our legal system has to really be updated if we are to protect the rights of creators.

On with today’s show. And for episode seven, our penultimate episode of this first season of Is This AI? we have a cracker of a provocation. And it is: Fear not, creatives, AI come in peace. AI come in peace. I’ve spent my life resisting easy puns, but not this one, apparently.

Following the headlines we discussed at the top of the show, there doesn’t seem to be much peaceful at all about the coming of AI for creatives.

Today we’re asking about is creativity still a distinctly human act? Does it matter? How do we protect its value? If the market isn’t sure how it values that human. Helping us today, we have two very productive creatives, fearless and fancy. They’ve been working at the cutting edge of AI, creative technology and marketing since well before the bandwagon started to roll.

They’ve made award-winning AI world firsts from AI gin to AI poetry. Derek and Norts, aka The Peeps. Welcome.

Thank you very much for having us. Thank you very much.

Thank you for having us on. Hi!

It’s fantastic that you could be here and share with us your insight because you really have been working in this field for longer than most, I think it’s fair to say. How do you respond first of all, to that provocation? This AI comes in peace to us creative folk.

Okay, so should I take first a pitch on that? I like to think, don’t necessarily, it’s an interesting thought, but I always think that it’s really not, AI’s never the issue, no technology is the issue. I don’t necessarily think AI is good or bad, right? And in this sense, I’m talking, as we are, about creative AI per se. It ultimately boils down to who’s using it and what is their intent.

Yeah, I come at it from a slightly different point of view and what Norts said is correct if you’re looking at it from the point of view of a creative who has full choice over their tools and full choice over what they’re doing. But what we’re seeing is AI, like any other kind of hyped product before it, like Web3 and Metaverse and Facebook-to-video, it’s sold in at the top and becomes something that you don’t necessarily have a say in. If you’re self-employed, maybe you do. Freelancer, you do. But if you’re a full-time employee in a company and the executive team decides, we’re replacing Department X with AI, you have no choice. AI hasn’t come for your job, but the executives have decided to replace people with AI. And this is happening across the board, and it’s a real problem.

See that’s what you get, when you talk to The Peeps you get two completely different opinions. And yet somehow it works.

Well, that is part of our value, isn’t it? To bring that divergent thinking. So yeah, we’re all for that. Just to step back, would you describe yourselves as early adopters of these creative tools?

Yeah. I’d say we were probably in the 1% of any people who ever thought about using AI creatively since it was created, to be honest. We boldly took up the mantle in 2018.

And when we first kind of stepped out to the world the technology was pretty sort of nascent and clunky and weird and quite a lot of people who knew us kind of thought we’d gone mad really because like, this stuff’s never… how are you gonna make this work for marketing?

And we didn’t even call it AI at the time, did we?

No, didn’t. Yes, that’s the classic thing. We made a very deliberate, conscious effort not to call it AI, because the type of thing we were using was a subset of the larger idea of AI, and we called it machine learning. And the way you used the technology was very different then, because it was all neural networks and coding and all this stuff back then. And what you could do as an output due to computing scale and size was unusual at best.

It kind of, interestingly for us, it then meant we started to get work from people who wanted to be seen as, or businesses, brands, organizations, who wanted to be seen as early adopters and do things with this weird, kooky technology that very few people are using.

For the less initiated, what were the tools then? What were you using? What were the inputs and what were the outputs?

So we were creating, and this is before large language models, we'd create very small neural networks using kind of Python tools on our own machines back when these things were small enough that you could run it locally. ⁓ And using very early generative image tools. This is before, was it before GANs? Or around about the time of when GANs were coming out. And this is before, again,

Dali and Mid Journey and all these kind of ⁓ diffusion tools. yeah, a lot of it was more, you have to assemble things by yourself from open source building blocks. There weren't all these platforms that were kind of ready to just go on and create your content. But then, you know, these things became more user friendly over the years, more powerful.

Yeah, none of these things existed.

A lot of the outputs we would do would be kind of text-based, blending things to make sort of weird outcomes, like what happens if you take Clockwork Orange and you blend it with Pride and Prejudice? What sort of book do you get out of that, right? It's question, but then we started working with people who were experts in their field to make things that these neural networks would suggest. So...at the time we're just a sort of little business that just wants to get noticed. look at us, look at us. We do weird stuff. So we started working with a cake maker. So we started making cupcakes, right? AI sort of like generated, the flavors were generated by AI, but we made them to actually cakes that someone could make. And bizarrely, again, unintended consequences, people started getting us to provide AI food at their events,

We did it as a sort of publicity stunt and then we found ourselves supplying cakes to sort of Henley Business College.

Taking the idea of the machine and then transplanting it into working with a human expert in a different field to do something with it. And that's sort of where we sort of moved on from. I suppose peaking Derek, when we came up with this idea to make this AI gin And we'd do the whole thing, the bottle, the naming of et cetera.

But the gig really wasn't about using AI. The gig was, you take this very small distiller and make them globally famous because they've made this gin? And that's what we were able to do, and that's what it actually did. So they made a thousand bottle run on the gin the botanical combination they decided to go with. But more importantly, because it was the first way back then, you could get it in every sort of tech magazine. It was in Metro newspaper. It was all over the shop. So there was a, for a very small spend, they got this sort crazy global outreach.

This has been such an interesting conversation to listen to an introduction to open up the podcast. you know, it sounds like you had a lot of limitations and challenges and using machine learning to make weird stuff in the very early days. Why did you pursue that? Like, what was it about the early days of AI that really kind of captured your imagination and made you think, yeah, we want to play around with this stuff and see where we're going with it. I'd love to hear about.

Like how you started that experiment and mostly why you started the experiments.

I went to South by Southwest in 2016. I was lucky enough to go and there were two talks on AI that I saw at different ends of the three or four days it was on.

They were the two things I just thought, wow, that's an interesting, that's an interesting idea. Can you really use this technology? And we just thought, yeah, we'll give it a go. And at the same time, we kind of decided to make that leap. I think that was the interesting thing. Because I don't think, correct me if I wronged it, we've never aspired to be like a tech business that wants to scale and have 200 people, et cetera, et cetera. Our kind of...

We've always maintained a smallness because we kind of like doing things quite fast, think just generally as our personalities, when you have a project. also we like to have it to be, it wants to be something that's very positive input to the world, wants to be sort of brings, brings some positive emotions. So we kind of, tend to work small, work fast. Back then it was kind of entertaining because whenever you made AI stuff, whether it was art and we used do a lot of talks and stuff like that and also do exhibitions where people asked us to do stuff. AI was so quirky, it was just weird, wasn't it? It was weird. That's like different from slop, which I'm sure we might talk about later. It was just weird, right? And people liked weird because they hadn't really seen it. And then of course what happens is large language models happen, all these platforms, and suddenly AI isn't weird anymore, is it? It's like something you can use in an associated practicable way.

Well, it is and it isn't. ⁓ What I love about what you're talking about is the weirdness. It's in the distortion, this... It's not trying to be a complete replica of a photo real facsimile of something that...

No, because you had no possible way to do that. So you've got to think, okay, let's just make weird stuff, weird art, weird kind of copy, weird, and blend it all together. So making weird was quite easy at the start of AI, because that's all you could do. Crazy cop, crazy flavors and shit like that.

Yeah, and I love that connection between the digital, the data and the models and the real world, and the tangible, the human experience. There's something very, do you miss that weirdness? What do you think about the weird versus the trying to make it just like a TV ad, a photo real TV ad now?

Well, it's harder to stand out now because everyone's doing it. And also back then, a lot of people were using AI to do fun things, but they weren't taking that extra leap into bringing it into the world with, you know, real people creating it. So that was the difference and that was what made it fun. know, like people were making poetry,

Yeah, now tools they all do very similar things you still obviously have the ability to create your own stuff that's a little weird and those are the things that break out

It's a little difficult because everyone now has access to the same tools and they're all doing the same things because everyone is influenced now by the same set of people, you know, on Instagram or LinkedIn, everyone's saying this is how you make an image, this is how you make a post. It's a challenge to kind of put your own spin on it. But I think, like Nort said, we try to keep ourselves small because we both like doing that work. both, you know, neither of us are into like...telling other people how to do it. We want to get our hands dirty and have fun.

What does a brief for The Peeps I know there probably isn't, there's never just one sort of cookie cutter brief, but what kind of briefs are you getting?

We seem to do a lot of stuff over the last year that's around conference where you can now make pretty good videos, whether you, and that's, I'm talking about the output, the human input and the human decision making at the end of the output and the human editing is something else, but we'll probably touch on that. We seem to do a lot of things around sort of conference events, videos, but again, AI.

think films to make your audience laugh, get inspired, et cetera. Because it is, if you come armed with a good script or a good idea, you can do something and it's not necessarily weird, but you can still write an entertaining, thoughtful, insightful, tell an insightful story. So conference videos seem to be a popular thing, but also what I'd say like those, I'd call them activations. Is that how they're called, Derek? think that's the term. Activation.

I don't know, you're walking at some conference somewhere and some brand has got a stand and wants to talk to you and it wants a reason for you to stop. And so it's perfectly doable for us to create something. Again, it's about the idea ultimately, isn't it? It's not the AI that can use AI to give someone an experience by doing something that is powered by AI that kind of gives them some sort of control.

When we first started, our thought was everyone would come to us to make AI ads. I don't know how we actually thought that was ever gonna happen back in 2018, because they'd have been god awful. But yeah, we don't get, and I know that just the other day, the sort of second iteration of the Coca Cola Christmas ad, the AI Christmas ad came out, didn't it? 

Yeah, what did you make of it?

It just seems to be last year's. It's the same as last year except they've replaced humans with animals. Which might be that you can make animals in a funny sort of way.

Bit underwhelming, I think. Well, it was. It's the same as last year's.

Much more, they've got a little bit more subtlety in that than the human thing. And also that thing about people always stare at human face a bit longer, don't they? And then decide, well, that's not real. It's okay. It's a lot of effort for not a lot of creative reward, I don't think.

It's a PR story isn't it at the end of the day it's successful PR. It's not like a very successful ad in itself.

It didn't make me feel anything at all. ⁓ the PR stories going around it are, well, last year it took 200 people and this year it took five. I mean, and is that what we call success? I can't see it.

Well, this is the, well, it's

part of the problem in this kind of whole AI, you know, it's kind of changing the topic a little bit, but it's going back to what you said earlier, Lisa, and it's people trying to sell a solution. So the whole other kind of AI space, like web three before it, they're going to change the world, but it's being, you know, hyped up by business people and non-creatives and sold in, and everyone has to use these tools.

So these are all basically adverts for the AI products, not adverts for Coca-Cola. It's to get people to get hyped and buy into Midjourney and Gemini and things. ⁓ But yeah, they're not. They don't have an emotional core. They're not something you're going to remember.

You get those people say, Hollywood's dead. Da da da. Art directors, it's over for you. Yeah, and it's like, this changes everything. And it's like, no, shut up, mate. Shut up, mate, you're an idiot. Because the AI doesn't change anything, because it's the input and it's the output. it's the kind of, the creativity ultimately comes from what goes in the first place. Has it got an idea? Has it been thought through?

Hahaha

Does it move people, inspire them, do whatever? And can you ultimately execute something that visually is kind of like true to what your vision was but also doesn't look awful? And like we say Derek, 90 % of everything is crap, isn't it? In all walks of creative life, how true?

Well, I kind of wanted to pick up on that if that's okay.

I wanted to ask a question around, you know, how you stop things from being crap by making sure that you maintain that great creative thinking upfront and creating those great ideas. And has there ever been a brief recently that's crossed your desk where you thought AI is actually not a good thing to be using for this and the kind of thought process around that? Is AI good for everything or is it only good for a few things?

It's probably good for 10 % of things I would think

I mean, if you can afford creatives, humans, to do the real job, employ them to do the real job, that's the problem, is people using these tools because they have them over actual, you know, like the amount of copywriters who have less work now, translators who have less work now, know, illustrators who have less work now. This, you know, it's not funny.

One of the exciting things about how you make it not crap, that's part of the effort of the combo of prompts, workflows, et cetera, I suppose in the modern world now, the problem is it's the ubiquity of social media and the ease of which anybody can publish their creative work because...

creativity is now democratized, that's not a problem if they kept it to themselves. It's the kind of like ability to show it to the world. Therefore, the same Sturgeon's revelation applies whereby it's just more of that 90 % that you actually get to see, don't you? Right, and so that could be in music, that could be in films, that could be whatever. But it so happens the proliferation of it is in social media because that's how the world exists. you know, people think, and it goes back to the point, I guess.

How do you stop things being crap? Well, you've got to be aware, you've got to a filter on you that says, is this crap or not? Right? And most people do not have that filter, it's fair to say. They just put it out there anyway. So, you know, not stop being, I don't want to be elitist or snobby, but you know, that's what some people are better at doing creative things than others. Everyone can do something that they consider creative, but what is genuinely groundbreakingly uniquely creative?

I often think is quite a divisive thing because I think things that become universally popular are probably not creative because they're too vanilla, too meh. Whereas a really great piece of work in any kind of field and that, glues AI, will divide half the audience. I think that's what's quite important because there'll be people who'll be quite offended. Yeah, think filters are very important facet. what do say, like going back to, let's think of something in history, like Stravinsky, Right of Spring.

When that was first performed, the audience rioted, right? Because they'd never heard anything like it What the hell is this? They said, but they were all wrong, weren't they? Because now the basis of all kind of modern cinema music comes from that bloody one piece And the same goes. So I think that's the challenge. And it's like Derek said, this thing about we get people who say, yeah, we'd like to know more because it's a much cheaper way for us to do what we want to do, right? And it always makes us both go.

That's not really why you should use it. You should be using creative AI to probably make some sort of creative output that you couldn't possibly do by any other way or makes it unusual. It's like, it's not the weirdness we talked about. You could be truly weird if you want to, but making weird is much harder work now because you've got to do a lot more things to un-normalise things. But I do think it's great when people come to us and they say, we want to do something that we couldn't possibly do by other means, not because we're saving

95 % of the money we want to spend is because we know that you might come at it creatively from a different way and you might put it together in a different way and it's kind of unusual in a good way.

I love that. loved two things there. One is about that editorial element. And yeah, you've done it and put it out before you've stopped and sort of sat with the discomfort of whether it's actually any good and whether you should put it out. And then

That bit around using AI to make possible something that you can't do any other way. I got excited hearing you articulate like that. That's always the way with things like this. Like what can we do with this that you can't do any other way? And that then...

Well, that's quite hard. That's a quite hard thing. That's like sitting at a piano and guitar, thinking, right, I'm gonna write a song that doesn't sound like 99 % of other music, right? It's the same thing. The process of doing that is quite hard. So it's always quite nice when we get gigs for maybe conference events and they don't really know what they want, but they this is our audience and they want to feel this. How can you make them do that? And that's.

And the thing that we've just talked about there, that you can only do with the AI, is like, so when Norts is talking about these kind conference videos, they want something funny but featuring their executive team. And obviously you can't get their team, know, the amount of time you need to actually film them or get them in one room, or they don't want to do it. And so that's where things like, know, AI tools like deep faking are useful there, because then you can create video with them in. So that's the kind of...practical thing that we're doing around that.

Nice. Yeah, you could never get that performance out of your C-suite, but everybody loves it.

Yeah, they love it.

And also it's like, you know, obviously we're very aware, is it right or wrong? Where is the bar in that? If you can take a likeness of someone and you can make them do pretty much what you want to do, or you can put words in their mouth, yeah, we can clone their voice. We can do all those sorts of things all day long, right? But there is a point where you get into the world and you see people doing it in a wrong way, in the fairest way. But as a practical tool for a corporate organization, they love it, don't they? They get an audience of people.

And the top three people, they're taking the piss out themselves by doing some sort of interesting fun, entertaining, also kind of gets the message over. So those are very useful ways of using that tech in not obvious ways. Consent.

Well, consent, yeah, consent is the key part of that, isn't it? Consent and transparency.

Everyone's knowing what they're getting into. But that's just one small example, isn't it? Which is lovely. If people want to experiment where do they start?

What's one thing they could do to start just doing something a bit more unusual?

I think to do anything in that sort of space is you do, I mean, there are so many platforms that try to sell themselves as a one-stop shop, but I personally, you might have a different view to it to me, don't think you can do the unusual off one platform, right? The trick is to work out how to combine different platforms.

Because they all do things, even if like, you there are 20 things that can make an image that you want of a scene, they'll all make it slightly different. And there'll be one that will aesthetically be, not saying you have to get all 20, but there are just certain ones that work better than others. And then let's go from say prompt to picture from, we like to go prompt to still image to animation, right? Then possibly upscale, especially if we're doing conference things, because you want it to be, look as good as possible. I mean, something that we started doing, which was a sort of time saving device, is like when you're trying to write really great prompts for really, really beautiful images that are very precise to what you want, because it might be about, you've got to think about, if you're creating an image, it's not just about a man, let's take a picture of a man sitting with headphones on in room, right? Because that doesn't really, that's a very general, very general take on what the image is. You really have to detail across a whole parameter of things.

That's about the lighting, the framing, the kind of colours, time of day, all that, the more information you can do. Now that's very time consuming, for example. If you have to be prompt, have to write, say for a video that's gonna have 20 frames in that are gonna be animated, that's quite time consuming, because you won't get it right first time, right? That's the other thing. So one of the things you might do is you might say take a large language model tool, say for example, a ChatGPT - because we do that - and you might make your own bespoke bot that works with the platform to write the type of prompts that will work in that platform. So it's sort of time-saving, and of course, you don't just drop it in, you look at it, you tweak it, and that just speeds it up because it will then designate you. So this is the 15 parameters I wanna look at. Can you write a prompt like that? You put it in, is it working? No, what am I changing about it? I might change that, that, that, and that. So that's useful, so I think you go from like, some way of all.

not automated prompts, but working with a tool to help you write more complex, sophisticated prompts, because you can do it as a human being, but if you're doing lots, it's a lot of time, right? Which is fine, we used to do it like that, and it was a lot of time, right? But then you go to platforms like your Mid Journeys, your Ideograms, all your of your Runways, we tend to use it to make the still image, and then for the still image, we then take it into something that might animate it. I mean, probably favourites.

Higgsfield is one we use a lot, Mid Journey as well, but again, Runway, they all do these things very differently. ⁓ Mid Journey is very cinematic. If you want a very cinematic looking thing, you probably get that Mid Journey to Higgs Field. Ideograms really good for likenesses, for example, that seems to be the best model for doing that, but you can do it on other things. Just depends what you're looking for. I suppose our sort of gamut, is we try to make people on screen look as much like them as they possibly can. And we know there are certain platforms to do that. So I think the simple advice is, if you're getting started in it, don't do everything because that can be mindblownly confusing. Just pick something to help you write good prompts, something to animate or something to make a video. So just a still image and something to animate and try to three, one, two, three and so on. And also as Derek will always say, you can always get to try these things out for free at first to see what works for you. You don't have to buy 45 subscriptions and find yourself paying two and a half thousand pounds a month for a lot of platforms you never use and stuff like that.

Well, thank you. Derek, final question to you, where do you see AI and creativity going? What are you excited about? What are you excited about? Let's finish on that one.

I think what's really, really exciting is there are now so many tools that are very narrowly focused towards one kind of creative endeavour. So, you know, if you're a coder, there are now so many really good coding assistants. And, you know, we kind of disparage like chat GPT and large language models in general, because they're not good for everything. But for some things, they're very, very good at. And so, you know, you mentioned vibe coding earlier.

That's where you ⁓ can just give an instruction and then receive some working code back. So coding tools are really cool. If you're a musician, they're really interesting tools, plugins to the software you're using. If you're a videographer, there's really interesting tools that will plug into After Effects and stuff. So I think the future, the exciting future is niche things that are very geared towards helping you be creative and because this is nearly all open source and the ability of coding tools to help you do things that you wouldn't have been able to do before, this is unlocking lots of potential new sectors, new software. So from my technical side, that's what's it's absolutely exciting for me.

Yeah, I love that. And that makes me that speaks to the provocation at the start around how how creatives, whether AI is coming in peace. And actually, if it comes and makes those niche skills more interesting, unlocks opportunities for human creatives in those spaces. That's absolutely brilliant. That's what that's what we're here for.

I keep coming back to, you know, to your point earlier, this idea that you raised earlier, Derek and Norts about, you know, what can AI allow you to do that you've never been able to do before? And that's what we should be using it for. And it's made me think a lot more about like how AI is not just a tool, but perhaps also a new medium. And I, yeah, thoroughly loved this conversation with you. So thank you so much. I learned a stack.

Thank you very much. Oliver and Lisa Talia, thank you. Thanks for having us.

Thank you.

Bye bye.

Thank you.