AI has changed the tools. Has it changed marketing?

I can remember when computers first started appearing in the workplace. In one of my first jobs, working in the marketing department of a local council, we had six computers between about 20 of us and had to book a slot if we wanted to use one.

It sounds faintly ridiculous now, but at the time there was nothing particularly unusual about it. Most of the work still happened in other ways. We had filing cabinets, telephones, paper, printers and, if you wanted to send something to somebody quickly, the fax machine.

Despite now spending most of my working life surrounded by technology, I still use a notebook and pen every day. There's something about a nice pen and notebook that helps me think and crystallise my thoughts in a way that typing into a screen sometimes doesn't.

The old tools didn't suddenly stop working when computers came along, but the way we worked gradually changed around them until computers went from something we booked a slot to use to something it would now be almost impossible to run a business without.

I've worked in marketing for more than 30 years, and I've watched that pattern repeat itself: a new technology arrives surrounded by excitement and predictions about how dramatically it will change our lives, we rush to work out what to do with it and, eventually, it either disappears or settles into something genuinely useful.

QR codes are a great example. When they first started appearing everywhere, I thought they felt rather gimmicky. Marketers stuck them on things because we could, often without giving much thought to whether they made the experience any better, and for a while they seemed destined to become another marketing fad.

Now I use them all the time.

We got better at finding useful things to do with QR codes, smartphones made them considerably easier to use and somewhere along the way they stopped feeling like technology at all. I scan one to look at a menu, pay for parking, get some information or book something without giving the little square I'm pointing my phone at a second thought.

Google Glass provokes almost the opposite reaction in me. I understand what the technology is trying to do, but I find the idea of someone wearing a camera on their face while interacting with me intrusive. Perhaps that will change too, but it raises something that gets lost when we talk about technological progress purely in terms of capability: often the issue isn't simply what the technology can do, but what we humans decide to do with it.

I think that's worth remembering as we work out what to do with AI.

There's an enormous amount of discussion about what AI can do, and increasingly about what it should do, but I'm much more interested in working out what it's actually useful for.

I use it increasingly in my own work and, in some areas, it has made a considerable difference. It helps me manage the sheer amount of information involved in running several client projects, interrogate research, organise my thinking, revisit previous decisions, identify gaps and get through some of the administrative work that frankly isn't the best use of my time.

I also argue with it quite a lot. It gets things wrong, occasionally produces something that sounds perfectly plausible until you look more closely and has a tendency towards language that sounds polished but says surprisingly little. If you've spent any time on LinkedIn recently, you've probably started recognising a certain AI-written rhythm too.

Quite often, its value for me is in having something to react to. I can ask it to challenge an idea or organise a mass of information, then apply my own experience and judgement to what comes back. It can help us work faster, think more broadly and make connections we might otherwise have missed. Sometimes it develops the idea further; sometimes I decide the AI has completely missed the point.

That relationship feels much more useful to me than asking it to “do the marketing”.

Just because we can make more doesn't mean anybody wants it

Marketing has been through this before. Email made it extraordinarily cheap to communicate with thousands of people, and as automation became easier we found even more reasons to do it.

Most of us now receive far more email than we could possibly give our full attention to, so we've become remarkably good at filtering it. We scan subject lines, recognise promotional formats and delete things without really consciously registering them.

Ironically, that can make something older feel new again. A well-crafted, properly targeted piece of direct communication landing on your doormat can have far more impact simply because you receive so little of it. I'm not suggesting we abandon email and start buying stamps, but when communication becomes incredibly easy to produce, its abundance changes the way people respond to it.

AI potentially takes that to another level. If every business can produce endless articles, emails, social posts, images and personalised communications with very little effort, I'm not convinced the answer is to produce more of them.

We'll become better at recognising the patterns and filtering out generic material, which means marketers need to think harder about what is actually worth saying and why somebody should give it their attention.

Using AI isn't the same as outsourcing your judgement

Give an AI tool a few instructions and within seconds you can have an article, social media post, image, presentation, marketing plan or page of website copy. That's impressive, but producing the thing was never really the whole job.

If I'm developing a marketing strategy for a client, the value isn't in creating a document with “Marketing Strategy” written on the front. It's in understanding the business well enough to work out what should be in it: listening to people, asking awkward questions, understanding customers and competitors, spotting contradictions and sometimes telling a client that the thing they thought was the problem probably isn't the problem at all.

AI can help with increasingly sophisticated parts of that process, but I still need to decide whether what comes back makes sense and be accountable for the recommendation I eventually make.

If I use a spreadsheet to calculate something, I'm responsible for checking the figures. If I use Google to research a market, I'm responsible for deciding whether the sources are credible. AI shouldn't remove that responsibility simply because the technology appears cleverer. If anything, the easier it becomes to generate plausible answers, the more important judgement becomes.

Useful doesn't mean unproblematic

There are parts of what is happening that concern me, and I don't think we should brush them aside simply because the technology is useful.

We need to understand what happens to the information we put into these systems, particularly where confidential business or client information is involved. There are complicated questions around copyright and the material used to train AI, alongside bias, misinformation and the increasing difficulty of knowing whether something we see, hear or read is genuine.

There is also the impact on work. Some jobs almost certainly will be affected and many more will change; marketing won't be immune to that. I'm increasingly interested in the environmental impact too, because the infrastructure required to build and run these systems uses considerable amounts of energy and resources.

Beyond that sits a bigger question about control. A relatively small number of very large technology companies are developing tools that could become fundamental to how we work, find information and make decisions.

I don't have neat answers to all of those questions, and I'm slightly suspicious of anyone who claims they do. I'd rather understand the uncomfortable bits than ignore them, but I don't think their existence means we should refuse to engage with AI either. If anything, it makes understanding it more important.

Learning to use it well

I suspect the next few years will involve quite a lot of experimentation. We'll find things AI is brilliant at, things it's surprisingly bad at and probably a few things that we initially automate before realising we preferred it when a human did them.

We also need to get beyond the idea that adopting AI means having a ChatGPT account and occasionally asking it to write something. Other AI tools are, of course, available.

The more interesting applications are starting to sit within workflows: connecting information, identifying gaps, reducing repetitive administration and making knowledge that already exists within an organisation easier to use.

That's where I'm finding it particularly useful, but I'm also becoming more selective about when I use it. Sometimes doing the thinking myself is the point. Sometimes talking to another person is the point. Sometimes writing something from scratch helps me work out what I actually think, and sometimes a notebook and pen is still the best tool for the job.

Perhaps learning when not to use AI will become as important as learning how to use it.

Somewhere between the hype and the horror stories

AI is changing the way businesses operate and marketers work, and ignoring that isn't a realistic option. Embracing it unquestioningly doesn't seem particularly sensible either.

I'm approaching this as a marketer who has lived through more than one supposedly revolutionary change in our industry and wants to understand this one properly. The challenge isn't to decide whether AI is good or bad, but to understand where it adds value and where human judgement still matters.

Technology changes, behaviours change and the tools available to us change, but good marketing has always required us to understand people well enough to know when, where and how to communicate with them.

And did I use AI to help me write this article? Of course I did.

I used it to explore the argument, challenge some of my thinking, organise the ideas and help edit the article. I disagreed with it, rewrote it, added my own experiences and removed quite a few things that sounded suspiciously like they'd been written by AI.

Although perhaps I haven't removed all of them.

Did you spot any?