Finding The Pulse In AI
Industry Insights from Pete Durant
Unless you’ve been living under a rock (not under The Rock - that would be entirely different), then you’ll have read about artificial intelligence and its impact on society in some way. From ad tech and the new wave of language model tools like Chat GPT to generative AI content creators where you can get a set of head shots for $20, it’s the in-vogue conversation.
AI (or, as most parents have called it at some point, the major road A1) is having and will continue to have an impact on all our lives. Whether that be the dawn of some totalitarian fear-ridden Matrix future where we have no jobs to the facilitation of a naturally integrated, seamless experience that enriches our lives, I’ve read it all.
Whatever your opinion, the unbelievable pace of AI development and integration is making it this generation’s arms race, with new product, software and use cases appearing every day.
But where is it having an impact for marketers specifically, and what do we need to do to make it an all-round positive experience?
We’re at the point where we can impact the evolution of AI support. As buyers and users in our organisations, we can impact the tech stacks we use and state what we need and expect from AI. In a world of uncomfortable digital tracking and advertising (I read an AI article, mentioned it over a coffee and now I’m unable to scroll my Instagram feed without hitting AI-sponsored content every other post), personal data sharing and some quite dreadful marketing automation (I’m looking at you, LinkedIn DM and InMail messaging), we can influence AI by making considered choices to ensure it develops in the right way.
It’s been around for a while.
It may be all the rage now, but AI has been in use for a long time. In fact, you’re almost certainly currently using it and most likely relying on it in your everyday life. In its basic marketing form, AI is based on machine learning, using language and data processing to create an output such as a critical marketing insight, decision and action.
For example, your social media post scheduling may be powered by AI data thinking, using engagement success metrics to define what posts should be and when they should go out. It’s already built into influencer identification and management platforms, where AI finds and interprets data and tells us which influencers we should use, taking weeks off an executive’s research time. Ever wondered how a search engine on an ecommerce site delivers you just the right white t-shirt when you ask for one? That’s AI-powered learning. Google, Ecosia and Jeeves? AI powered search engines, not kooky English(ish) butlers.
AI is already deeply embedded into the Google ecosystem through their basic search, offensive content identification on YouTube, Google Maps’ route finders and smart scheduling tools in Calendar, to mention just a few aspects. We just haven’t seen it as something game changing for our livelihoods. These instances of AI have been exciting tech developments, not mainstream juicy “woah!" moments.
AI has also driven major changes to digital advertising in the last few years. It hasn’t made the Six O’clock News (probably because there was a big This Morning announcement that day) but it’s promised smart optimisation of ads, creative and delivery. Programmatic advertising is based on intelligent delivery of advertising digitally by a programme. AI has made the buying and placement of this more effective, while ensuring these ads aren’t prone to ad fraud. Your social ads on Meta are being delivered by AI, which identifies the best performing ads and prioritises them when serving creative to potential buyers. AI can already power up new ads from existing creative based on performance. Performance Max was heralded by Google as a genuine game changer, where you can essentially leave AI to run and deliver your always-on campaigns without the need for human intervention.
These are just a few examples of how, for years, AI has already been a part of our lives. Where people have raged about kids writing essays via Chat GPT, they’ve also delighted in their recent purchases made after a brilliant combination of display, Pmax and email automation; ultimately AI influencing they might shudder at if they realised how it had happened. We use AI because it does deliver better return-on-investment and because it delivers faster, smarter decision-making in digital advertising. Ultimately, refusing to use AI would cripple our effectiveness and efficiency. It still has its limitations. The above notwithstanding, AI is currently still extremely limited. Influencer marketing best practice dictates, rightfully so, that you must combine the tech with a human when choosing influencers. At the time of writing this, no AI can fully understand the nuance of a brand and how appropriate somebody is for a project. I’ve heard tales of Tik Tok’s influencer tools simply not being up-to-scratch because the AI can only interpret numbers and will aways serve the same influencers to a user. Pmax struggles to work for ecommerce businesses that bring in completely new products or lines on a seasonal or regular basis. It can’t deal with a new product being added to its learned system because all the signals it uses says that no one has bought said new product, so it’ll keep pushing spend towards the old established categories. This mentality leaves gaps for seasonality and encourages the simple shifting of stock regardless of how cheap socks, pants or a hat might be in terms of CPA. It doesn’t allow for something groundbreaking to come along and makes it difficult for new products to succeed while still maintaining volume. In fact, best ad practice now dictates that most campaigns have a mix of human controlled and Pmax campaigns built within it, specifically to navigate things like this. Programmatic and marketing automation is good but, unless tightly managed, can be spammy, irrelevant and why 40% of the global population (Hootsuite) use ad blockers. The ad creative at scale tools provided by Google and Meta, for example, where they can generate ads based on basic provided creative, aren’t that exciting. This, of course, is the personal opinion of someone with a creative background, but what they make feels off to me: never quite… human. You need to check and amend copy and videos are never as good as they could be - yet. Even creative tools that claim to make a presentation for you just deliver grey creative without personality, as found by a BBC journalist trialling Microsoft’s new Copilot tool. What was lovely and fine for another AI tool would’ve had you snoozing in the back of a darkened conference room. Tools like Synthesia can create endless training videos at very little cost using an AI-powered human recording. But taking the heart and soul out of delivery and understanding sounds like the opposite reason to create training. Why not just animate it instead of creating Red Dwarf Holly-fuelled content that will… okay, Holly would be great, but you get my point. In recruitment, AI is used to sift through those 600+ applications that hiring managers and recruiters receive for every job. Obviously this is helpful, as you can never go through 600 applications. However, it’s let down massively by a lack of understanding of people and nuance. AI isn’t the solution to replacing a good talent manager or recruiter who can find the right people for you and understand what words mean rather than basing hiring decisions on who uses the right keyword. AI-driven chatbots have never been a great experience. If we’re going to a chatbot, we’re likely wanting a human experience. Players like Deepdesk are trying to use AI to evolve this so we have actual verbal AI customer service agents instead, but the success of this is yet to be seen. Still, these tools are powerful and they do deliver some great results when used in a considered way. What’s exciting is where they will be in two years’ time… It's a tool, so find the pulse. What’s most important to think about when considering AI? The human behind it. Even in data science, we need to find the pulse within. The why and how that number is created. The emotional context and personality of your target consumer is the most important thing to consider when marketing and AI simply cannot deal with that, no matter how many times I make a Ken Barbie version of myself online. It’s our sentience and tendency to be put off by something spammy that needs to be understood in order to sell an expensive coat. That might cost more, but it’ll have a bigger overarching organisational impact and allow us to play with language that brings the target market to life. We need to approach AI as the brilliant technological revolution it is and take responsibility for our role within it. Whether as creators, editors, curators, strategists or insight miners, we have the responsibility and privilege to inject emotion, sentience and humanity into the tools. I wrote the novel #Blessed in which AI was making our life choices for us, according to data read from health apps and signals from social media sites on how other people enjoyed these choices. What AI can’t do is taste, touch, feel and connect your experience to memory. Yet. But by combining the emotional left side and data-driven right side of our brains, we have the potential to deliver great results. To ensure the people we market to are valued and that the balance of influence never tips too much. After all, who wants to be targeted, hit, stalked, cloned, retargeted and converted, except a cyberman superfan? So, as marketers, how do we lead the industry away from the gnarly end of capitalism which pushes AI that is driven only by profit and not by people, planet and progress? AI for right now and for tomorrow. Apart from taking back some control from the creepy automation you’re already using, where should you be looking right now to make impact? What mountains can AI help us make molehills of? Processing large data sets and being able to translate that into actionable insights is a huge opportunity for marketers. As much as we like to brag about how data and maths-led we are, business analysts are worth their weight in gold. When budget or access constraints mean you can’t access these, or teams to create large complicated multi-channel attribution models, AI has a huge role to play and presents a huge opportunity. Today BI is the top AI market according to Data Bricks (2023) and data integration is the fastest growing part of that market, with 117% growth. Our data worlds are fragmented, complicated and it’s often hard to truly understand the relationship between them. AI can help us do this and combining that with an insight lead to use the thinking to challenge our business status quo is the most powerful and exciting sort of strategy. Do you even have a BI unit in your business? Alongside one strong hire, AI tools across data integration, migration and business intelligence (from kepler, great expectations and looker to esri, Qlik dbt and Fivetran) could be a brilliant place to start and scale quickly. Wherever you are in your journey, you should really be implementing an AI-based data systems approach from today. Workplace productivity is a huge area of improvement, especially for agencies and services businesses. Sometimes at agency level, meetings, writing up notes from meetings, getting sign off on notes, interpreting these notes and then combining them with an updated status sheet takes time away from the valuable service an agency should be providing. Volume businesses that sell brain power and expertise need to focus on giving more of the human element and less of the admin to make impact and, importantly, profit and scale. AI tools like Motion can organise your diary to help you be more productive, group meetings together and create space to deliver work. Otter and Fireflies are great examples of note takers and list creators that can make somebody’s 30-minute job into a two-minute check. Although I’ve been negative about some of the creative AI tools out there, there’s still a load of brilliant uses that we can implement. Outside the gimmicky tools to remove backgrounds and audio noise, you can really scale and power SEO strategies. Being able to use ChatGPT to generate a constant stream of articles and content that would take months to create can transform your business results. Simply turning your freelance writer budget to editor, turning it into brand voice and reviewing pacing, will bring down costs and take away the heavy lifting while still having the human guiding hand there. How many times have you felt like that there’s just too much to do on the website and not enough manpower to create the code to make it happen? With the right brief, AI can create the code that your internal teams can use and implement. You still need the team, but lots of the hard graft is done to speed up delivery of projects. I’m sure I'll get beaten down here by friends and foes, but how much originality goes into media planning? Can an AI tool do this job for me? When looking at the bigger spends of businesses that have similar plans every year (ITV, digital, Channel 4, outdoor and some press standard) can I speed up this process and use my planners to add difference, nuance and change? Often the challenge in planning is integrating change, because of the volume of tasks necessary to properly understand it. AI could remove this challenge. In all of these examples, AI can do the heavy lifting freeing us up to add value, thought, strategy, storytelling, emotion and nuance. It should mean better work and additional space for the type of progress we want marketing to make on society. Where can your business use AI to unlock this powerful human impact? Future skills and culture. If we’re going to implement AI in this way, how and where do our skills change? What is it that we need to ensure colleagues truly benefit? The first crucial skill is being able to write a proper brief. To get the most out of an AI tool, you need to ask the right questions and train it. Communicating complex ideas or simply being able to write exactly what we need and want is a rare skill in our industry. Let's be honest, a lot of people are bad at it. But not only does AI need it to be able to create code, SEO content, but to get the output we want, we’ll need to be able to express the value we’ll add and be brilliant at briefing our tools. Much like social listening’s Boolean search, it’s a new key role to learn. Simply put, we’ll need to be better thinkers and our teams will need to evolve. I’d love to see an agency put 10 people in the room with degrees, A-levels or even btecs in psychology and sociology. We’re in the business of behaviour change and we need to hire more people obsessed with cultural and human change. With the rise of digital, we got a rise in direct mail catalogue usage. With MP3s and streaming came the return of records and collectables. I think we’ll also see a reaction where tactile personal experiences are privileged as AI becomes more prevalent in our lives. More value will be put onto human creators, writers and artistry and that will then impact the way our marketing creative evolves. We will need to fight for our culture. The danger of language-based AI created by a large general public, or AI controlled by big American corporations, is the ideologies and cultural norms it pulls from. Solely pulling data might mean it processes a certain body type in an Instagram post as the right way to edit your selfie. AI may look at skin tones of existing CEOs and decide that lighter skin edits on headshots is the right way to help someone get a job. Language and cultural expressions could quickly be homogenised by pulling from the largely US-based internet culture and news outlets. Being able to act as the filter for this and translate output, almost as a gatekeeper, will be key to keeping local alive and some of the more unsavoury cultural views dead. Policing what AI pulls from the vast content world and helping it to understand and process this data responsibly will be almost impossible. We as humans apply media/opinion bias to our own sources of information subconsciously, so helping AI to decide whether a report from The Sun or The Guardian is unbiased or ‘fake news’ will need that human touch. Even academic sources aren’t always objective. They can be funded by someone pre-ordaining a result. Something we see a lot in the dairy, agricultural and natural resources industries, funders will try to manipulate academia to manufacture results that prove that they’re one of the good guys. How AI navigates that and surfaces genuinely ground-breaking and often opposing theories will be key. So, most importantly, the big question in the broad Google, Apple and ChatGPT styled AI platforms will come to ethics. Behind every AI is a human, programming it, training it and using it. If the future is putting us at the heart of AI, we’re challenged by exactly the same thing I’m advocating for. Emotion, values, personal opinion. AI is still working within the parameters and rules you set. What values will it promote? Who is the owner and architect of this world? To provoke an action, AI harvests inputs and data, and can choose (or be programmed to choose) and interpret whatever it chooses. This will raise huge governance questions for governments, schools and people. Future outlook. It’s exciting! The use cases, whether macro or micro, are fascinating and seeing the improvements in real time is a reason to get excited about ‘tech’. The next mainstream leap will come when Apple integrate AI into their ecosystem in a purposeful way. Google exec Matt Brittin says, “it’s too important not to get right”. They claim that this won’t be until at least 2025, but it’ll push the rest of the market on. If my #Blessed nightmare/dream can happen, at least within device walled gardens, the possibilities could be endless. China specifically will be taking the lead with their more open data culture. AI will become man’s best friend. Just as my dog completely impacted my values and view on sentience and life, AI will change us. But how is up to us. For now, we need to take baby steps. Focus on our data infrastructure, empower our productivity and focus on our humanity and values, so we can impact our businesses positively and control the narrative and development of the industry. We already need to put heart and soul into our data and advertising industries, so this is the prod up the arse to get us going! After all, it will all lead to creating better work. And lastly, as a snippet of ultimate digital craziness, while loading an AI tool’s website, I was asked if I was a robot before proceeding. Is this AI dating? Is this how AI will meet local robots in their area? A new spin off for First Dates? I’ll tell my AI diary tool to put some time aside to email the Channel 4 Commissioning Exec and find out.