AI & The Creative in Digital Marketing. Part 1

This episode is brought to you as always by The Digital Distillery.

If you would like to get in touch with us you can head to our website or email podcast@the-digital-distillery.com

You can follow us on:
Facebook, Twitter, Instagram & LinkedIn

Written, Produced & Engineered by Phil McDowell
Executive Producer: Nadia Koski
Project Leads: Dennis Kirschner

Transcript

And welcome back to the digital distillery podcast. The show where you join me, Phil McDowell as I take you into the hearts of those meetings of digital marketing minds that we know and love as the digital distillery. 

Now if you are just joining us for season two of the show, these are thought leadership, invite only events held all over Europe. And they bring to the stage key movers and or shakers in the ever evolving digital marketing industry to put forward their ideas and experiences on a particular pertinent topic. 

Today the place is Vienna, the mover is CEO of a consulting agency Michael Katzlberger, and the topic, me myself and ai. 

And without further a dance move, let’s do this…robot

Artificial intelligence, or Kunstlicher Intelligenz, KI as we will hear it referred to from the stage in Vienna.

I don’t need to tell you that a few short months ago this long used, but predominantly fiction residing term, exploded onto the hype scene. Bringing with it a myriad of professional and personal practicalities and concerns. It has been reported on, talked about, demonstrated, griped with and existentially bemoaned in seemingly every corner of the media and especially professions that even get near a computer in their day-to-day. Which is a lot of them. 

The number of reporters, tv presenters, writers and podcasters who I have heard say ‘and guess what, AI wrote that introduction you just heard’ with a smug grin and an eye twinkle, explicitly implying their own creative genius at being able to fool the lowly humans, and to be the first one to do it.

Not me, my introduction was completely 100 percent brain powered as you can surely tell from the unparalleled linguistic creativity that I pour into every single tortured metaphor and twisted turn of phrase…..

[typing into chatgpt. ‘What is a tortured metaphor…. Tortured metaphors are a device used by poor writers at an attempt to make them sound more intelligent. You are not as smart as you think you are… Shutup… stupid robot]

Anyway, like I was saying… my introduction for this episode was all me in my questionably creative glory….but it took me a little while to write, and would anyone notice if I hadn’t? Would my time have been better spent elsewhere? Or are my thoughts and writing style important to the authenticity of the show and worth the time spent? I Dunno… but these are all important questions that our industry leaders and its’ creative’s are, all of a sudden, forced to ask themselves and to start making financial decisions on.

[Chris - I hope whatever the future holds its going to involve me sitting down with words]

Chris cares. But we’ll come back to that. First let's go back to Mr Katzlberger in Vienna and look briefly into where this all comes from, and why its only now we are seeing this rapid and widespread adoption of this technology, even though a lot of the concepts have been around for more than half a century.

Traditionally, a lot of what is in the general consciousness about AI comes largely from film. In the first half of the 20th century , science fiction familiarised the world with artificially intelligent robots. From as early as the ‘tin man’ in the wizard of oz in the 30’s, to HAL 9000 from 2001: a space odyssey in the 60’s, and ASH from Alien in the late 70’s. All of whom were dastardly insidious villiouns out to destroy humanity…especially that Tin Man.

And its because of this perhaps that a lot of the associations we have had with Artificial intelligence are overall pretty negative.

Michael suggests that the first thing we need to understand about AI, is that the whole concept is about trying to sort of, recreate the naturally occurring human brain mathematically with computing power and algorithms. And that nature, as the original and accidental creator of intelligence, nothing is as ingenious and wonderful a designer as her.

Now the term Artificial Intelligence was coined in 1955 by a professor at Stanford, John McCarthy but in the few years leading up to this, there were already a generation of scientists, mathematicians and philosophers playing with the idea. One of these people notably was a young British polymath named Alan Turing who suggested that humans simply combine the information available to them, with their ability to reason in order to solve problems and make decisions. And logically that, given enough computing power, machines should be able to do exactly the same thing.

Computing power however was exactly the problem since at the time, computers were only able to execute commands but not store them which is obviously a prerequisite for intelligence. Long story short, and it is a good story, (I’ll link a couple of articles and papers in the show notes) but essentially it was that fundamental limit of computer storage that has held the technology back for so long. And quite simply put, it's just not anymore. 

Moore’s law has inevitably caught up meaning that we now have the power. After beating Kasparov in the late 90’s in chess, and Google’s Alpha Go besting Go champion Ke Jie in 2017, we have reached a saturation point in computing power that has led to this these impressive feats we have all been hearing about for the last few months; Generative AI and the household names of MrGPT & Friends. Although do note that this gendering of ChatGPT is only in my head, likely because I was around and on the internet in that strange and magical time of Asking Jeeves, so I kinda imagine GPT as a helpful but sometimes mischievous butler. 

And upon catching myself gendering this machine while writing the script  I checked with ChatGPT and it was very clear to tell me that as an AI language model it doesn’t have personality, consciousness or a gender but that it is helpful for me to refer to it in that way then I am welcome to.

GPT Actually stands for Generative Pre-Trained Transformer. Generative refers to these models ability to generate new text based on the input it receives. Pre-trained meaning that the models are trained on large amounts of data from which to draw from, before being fine-tuned to a specific task. And Transformer refers to the specific type of ‘neural network architecture’ used which was released by Google in 2017. Basically it’s the type of machine learning technology that teaches these models how to understand language.

As you all know it’s not just OpenAI that are developing and releasing these kinds of tools but they seem to be cropping up everywhere with all sorts of different specialisations. There are image based tools like DALL:E and DeepArt which can generate entirely new images based on a text input.

There are things like Artbreeder and Journey which specialise in synthesising video content entirely from text based prompts. And for written content there is Open AI’s own GPT 4 and Writer.com. These are really just a couple of examples as there really are just so many.

So that’s a bit about some of what Generative AI can do, but how does it actually work?

But while putting in funny or weird prompts to see what the machine comes up with is fascinating, it really is just the tip of the iceberg when it comes to the current and potential impact that these tools on creative industries like digital marketing.

This is Chris. 

So Chris is a colleague of mine…

He has been writing specifically in the tech industry since 2019…

And he has kinda taken it upon himself within the company to really dig into the burgeoning AI’s impact on the industry and on the creative specifically. And as such I thought he might be an excellent candidate to give us a bit of an overview of what he feels those impacts are going to be.

Now initially it was generally thought that creative jobs were reasonably safe from the developing generative AI technologies. That these technologies would be really good at analysing and summarising data, but that the human brain and human experience was so integral to the act of creating something that it wouldn’t be replaced so easily.

What we are seeing now however, is that generative AI is doing a much better job than we expected. Especially in the professional space.

So with the help of AI tools, graphic designers can now generate logos, typography, mock-ups and product designs. Writes can block out article structures, research or draft texts. And videographers can create demo videos from storyboards alone, or even create the storyboards in the first place from text. All in a tiny fraction of the time, and with a lot less of the skill than was traditionally required. 

As you can see, this could lead to a noticeable reduction in creative specific jobs, especially those in the entry level where the tasks are generally more repetitive and as such more easily automated.

Ad’s are another example. Specifically image based but very quickly becoming video based as well, ads can now be made with a fraction of the organisation and labour and therefore cost.

Let's say a client from the food industry wants to start a campaign using a particular character, in this case they want an old and experienced fisherman. Deadline yesterday. Shooting isn’t possible due to time and budget constraints. Post production would require 8 hours in photoshop and again, deadline yesterday.

It can’t be done. So typically we would turn to a stock images and see if we could find something, anything short notice, that gave the same kind of feel. But it's never going to be exactly what the client wants and you are extremely limited in what you can do with that one image you managed to find if any variations of that campaign are wanted in the future.

Enter AI.

Since the client doesn’t know exactly what they want, I first go to chat GPT and ask it if It could please (I always find myself being extremely polite to these things, partially out of what my mother would call a good upbringing, and partly..you know…just in case…anyway…)

If it could please give me a detailed and colourful description of an old experienced fisherman so that I can then take that description to an AI image generator. And for the purposes of science this exactly what i just did and now, and bloody hell was I impressed. This is what came back from that exact prompt:

Dramatic rendition and background seagulls performed by yours truly… the rendition that is, not the seagulls [clears throat]

[The old fisherman stood on the weathered, wooden dock, a figure etched with the marks of countless days spent at sea. His face, tanned and weather-beaten by the sun and sea spray, bore the deep lines of a life lived amidst the elements. Wisps of silver-gray hair peeked out…]

Really pretty impressive stuff. And it just keeps going.

[His eyes, a piercing shade of deep blue, held the tales of a thousand adventures upon the rolling waves...]

Anyway I have put the whole thing word for word into the shownotes, totally worth a read, but it goes into the details of his thick salt and pepper beard, his strong sinewy hands and his faded and patched blue fishermans jacket

All really descriptive things that I can then take to an AI image generator and create this character for the clients campaign in whatever format they desire. And more than that, once that character is established, we can put him in other environments. Maybe they want a follow up ad where he is out of place in a New York Supermarket, or in a short video, where he is sitting on the docks, fishing rod in hand and turns to wink at the camera as the fishing line goes taught. All stuff that AI can do pretty well now, and is getting exponentially better by the month.

Great right? So then. Who’s the loser in this situation?

Consider also how plagiarism changes. Up until now copyright, IP ownership and plagiarism protection…I mean don’t get me wrong there is a whole episode’s discussion worth of issues with these things, especially in the US, but relatively speaking, it’s pretty clear. I created this piece of content, and registered somehow that I did, and more or less no one is just allowed to rip it off. But when it’s a computer program ripping it off, and although I asked the computer for something but I’m not aware the result is violating someone’s IP then who’s fault is it?

Tricky.

Of course you could ignore this whole technolical revolution thing in an attempt to stay pure in the old ways, but typically doesn’t tend to work out so well. As is so delicately put by one of our panellists Herr Demner in his interview.

To be fair, how much you rely on the tech is another matter. And Mr Demner warned against complete reliance. He proved his point by creating arguably the most memorable moment of the whole event where he got up and jumped up and down rave style to pounding music on stage, only relying on the anolog technology of ‘old knees’ as he put it.

Another concern mentioned on the panel was that of the rise of homogeneity.

and actually this is something that the delightfully insightful Thomas Koch spoke about in his analog vs digital interview from last season of the show. Which by the way if you haven’t caught up on last season yet… what are you waiting for?

Ok. Shameless self promotion… ah  check.

Where were we? Ah yes.

So if everyone is producing high quality content using the same tools as one another, (and tool for want of a better word here since yes the great wall of china, and the statue of David were both probably built using hammers but are clearly worlds apart..)

then we are inevitably going to start seeing a whole lot of campaigns that have a lot in common and therefore potentially stand out less and have a reduced impact, even though they are really high quality.

Of course this shift in the landscape will have a flipside somewhere. What that’s going to be is up for debate but there are a few sound ideas buzzing around the space.

This shift in technology and the way we operate professionally leading to shift in value, is not at all a new thing.

A programmer in recent years has had plenty of job opportunities across a range of industries and even the potential to strike it rich with some neat little idea. But if you were a programmer in the 1960’s for example, you might get lucky and be one of a handful to land a job at NASA or in a mining company like my dad did, who still talks about learning the programming language fortran4 and feeding the computers with punch cards. But it certainly wasn’t a safe bet for employment, and the best you could hope for was drawing a steady middle-class wage.

 Before that even the rise of the industrial revolution favoured physical strength and able-bodied individuals to work the burgeoning technology at the time aka factories. So people with those particular attributes found it easier to get a job. 

With this new technology things are likely to change in what the demand for labour is. With AI certain skills are going to be less valuable to employers. For example the value of an individual's ability to sit and stay focussed to perform a range of tasks using AI is rising whereas the value of vast factual knowledge such as a doctor or lawyer typically had to have might fall. Instead the skills for synthesis and persuasion in the case of the lawyer or empathy and an experienced based intuition  for a doctor might become valuable.

True originality or authenticity may start to be heavily favoured because those are the only things that poke their heads up above the mass of well written, proven format, polished AI pieces or campaigns. And then is the true creative valued even more?

Will the shift go back toward physical types of jobs or more analog types of art, where things that aren’t mass produced rise in popularity. Only time will tell. But one thing is for sure and always been; those willing to read the surf and ride wave, tweaking and twisting with the flow in order to maximise that unstoppable force of nature, of progress or at the very least of change, are going to be the ones that come out standing. And not face-first in the wet sandy, and sometimes slimy and stinging embrace of the jellyfish of the change-averse.

Thank you for joining me on this splash into the 1st part of the Digital Distillery event in Vienna. Still to come we are looking at responsible media, what ever is happening to that damned un-crumbleable cookie, and whether or not CTV has actually successfully killed the linear star.

A big thank you again to everyone who attended the event and made it possible. To the crew in Vienna, to the video team for their tireless efforts and of course to Executive Producer Nadia Koski and Project Leads Dennis Kirschner and Stefanie Leonardi.

I look forward to seeing you all here again next time on the Digital Distillery Podcast.

Au revior