ChatGTP has put Generative Artificial Intelligence in the spotlight, but for many of us, it looks like a solution in search of a problem. What should professional services marketers make of this exciting new technology, asks James Lumley.

“When you look at AI through the lens of professional services marketing, you have to ask
two questions,” says Lee Grunnell, Chief Marketing Officer at national law firm Womble Bond Dickinson.

“Those questions are: what do we mean by AI, and what do we mean by marketing? And for most people, right now, by marketing they mean ‘marketing communications’ and by ‘AI’ they mean the sort of AI people are using to write blogs and articles. There’s a lot more to both.”

His department will experiment with AI, he says, to see, for example, if they can use it to write the first drafts of newsletters and potentially turn them into audio.

More interestingly, they also want to see if they can leverage AI for bids and tenders.

“We have a huge database of case studies, credentials, quotes and biographies. We are going to look at whether we can point the AI at it and see if it can, for example, write a tender to an HR director at an investment bank.”

Again, it would be a first draft that would have to be improved by humans, but it does have the potential to save a great deal of time to free team members up for other things, says Grunnell.

Differentiation?

“For me, there is a whole tranche of marketing – and I mean marketing in the widest sense, across all of the four Ps, not just the marcomms part of promotion – that will remain unaffected, at least for some time to come,” says Grunnell.

“We have just finished a host of qualitative client interviews that we are using to inform our marketing strategy
in different ways, not least to inform our quant research, segmentation, targeting, and positioning. Could AI do all that? I’m not sure. We could give the interviews to an AI tool and ask it to pick out key themes, and it might do it more quickly than a proper marketer, but would the outcome be improved by AI? I really don’t think so.”

In fact, he suspects that right now, using AI for anything other than the simplest of writing tasks and heavylifting large amounts of data could be counterproductive to good marketing.

“It really could have a detrimental effect,” he says. “Professional services firms already struggle to differentiate. It’s already an industry full of the same generic words and phrases, despite the fact we know clients don’t want that. Things are already too homogenous, and AI could just make it worse. That’s my fear,” he says.

It’s a great sandbox for creativity

It’s a great sandbox for creativity Simon Marshall, founder of professional series marketing agency TBD, has been using ChatGTP and other packages as creativity and ideation tools, and has been enjoying every minute.

“Right now, it isn’t going to put any good writers out of business, but it is a great tool for brainstorming,” he says.

“Once you start using it, you learn how to ask better questions to get more interesting results.”

marketers could ask AI to write an invitation to an event aimed at a 37-year-old banker who gets invitations all the time. Not, says Marshall, to cut and paste the result into an email, but to see what it throws up.

“It can really augment your thinking and come up with some ideas that you would never have come up with on your own.”

Marketers should disclose when they use AI to generate copy.

He also uses it for silly, fun, throwaway blog posts such as ‘what do lawyers have in common with cats?’ It is, he says, the sort of thing you wouldn’t waste an afternoon on, but it isn’t a bad use of a spare 15 minutes.

Marshall has a serious reason for using it for less serious things.

“When marketers are using AI, they need to take a step back and look at where the data comes from,” he says. AI builds, broadly, on the online archive of humanity. Which means that not only is it not always right, but it has well-documented tendencies towards humanity’s biases. It can, for example, be sexist, racist, homophobic and much more besides.

“If you ask AI to draw you a picture of a CEO, it’ll most likely draw a man. A nurse? It’ll probably draw a woman,” he says. “Asking it to write an invitation has a relatively low-risk factor, so just consider potential risk versus potential reward.”

Email marketing

“AI can be really good at summarising things,” says Alexander Low, UK Managing Director at DCM Insights and host of the Death of a Salesman podcast.

“For example, you can give it some complicated text and say: turn this into 10 bullet points.”

Working with software engineers from Siemens he decided to see how he could harness this, and he came up with the idea of using it for email marketing.

“We asked the AI to simplify marketing copy for use in an email campaign, and then simplify it some more. They are seeing greater success rate in terms of opening rates, cold emails, and improved engagement with clients, which is a really interesting dynamic.”

There are many other uses, he says, for AI’s summarising prowess.

For example, a bot could turn a long earnings call, podcast or article into a series of digestible bullets, or summarise a potential client’s financial results over a period of time.

“That sort of thing could take ages. Now you can ask the AI to do it, and, for example, tell it to summarise the key points, the impacts, and the key legal implications.”

This is closer to the way that AI is being used in life sciences and medical research. If it can summarise one earnings call or annual report, it can summarise two, three or a thousand giving the user a level of analysis that simply couldn’t be achieved by any normal hard-pressed marketing team.

Marketing personas

Low has also found it an interesting tool to help develop marketing personas – fictional characters created by marketers to represent a target audience. With practice, he’s learned to ask the right questions and has had good results. But, he says, as with many processes
the same old principle of ‘garbage in, garbage out’ applies. AI bots, at the moment, can only be as good as the data it is working from, and the user.

Low also has his concerns. He believes that marketers should disclose when they use AI to generate copy and don’t add to it; so it is 100% AI-generated.

“Imagine if your AI-generated text leads to a client conversation, and they discover that until then, they haven’t been speaking to a real human. They are likely to feel a little aggrieved.

“My view is that if you are using content that is 100% AI-generated to create business relationships with people in the public domain, you should declare that it has been created by AI.”

A digital assistant

Heather Murray, who runs content and training agency Beesting Digital, has also been experimenting heavily with AI, and she has found many uses for it. In her opinion, the mistake many people make is that they expect too much from it, so don’t appreciate what it can bring.

Some of her clients have asked her if it could be used to get rid of whole marketing teams, she says. Others don’t want anything to do with it. In her opinion, both approaches miss the point. To her, AI is a useful virtual assistant.

“Ask it to check if some copy is in-house style,” she says. “It’s great for that.”

We are probably going to see a tsunami of identical marketing content coming out… and it is all going to be as dull as dishwater.

As for creating copy: writers and editors can check to see if they have missed anything in their article or see what the differences are between their article and an AI-generated one. It can provide ideas for structures and hooks, and break writer’s block.

“Using AI like an assistant is like having someone on the team with a tonne of potential, who is also really, really quick but can also make really, really dumb mistakes,” she says.

Direct well, and you will get results.
If you don’t, you won’t.

“AI does not create talent,” she says. “It augments talent. That’s the thing that lots of people fundamentally misunderstand. They think that they can use it to get rid of their copywriter or editor. But in fact, the copywriter or editor should be using it to create even better copy.”

Meanwhile, Murray is an enthusiastic user of AI solutions such as Sybill, which analyses video meetings. There are, she points out, plenty of labour-saving online software solutions regularly used by marketers that leverage AI, such as apps that work with video calls to minute meetings, to products that make podcast and video editing relatively easy.

A spur to creativity – and a threat to mediocrity

In the longer term, Murray is an AI optimist. She predicts it is going to drive quality up and lead to more human interaction, not less.

“Yes, we are probably going to see a tsunami of identical marketing content coming out. Comments, posts, articles and blogs all heavily AIassisted, and it is all going to be as dull as dishwater.

“So, to stand out, copy will have to have a human touch. It is going to be really interesting to see how creative people rise to the challenge.”

Murray also thinks that it is likely to lead to more in-person meetings because, as we use AI more and more, with its slick blandness, we are likely to prefer interacting with messy, imperfect humanity.

In this brave new world, the people who will likely be challenged the most are those who have got used to providing middling, characterless marketing material.

“If you can get adequate mediocrity on tap for £20 a month, who is going to pay any more?” she says.

Death by AI?

From the myth of Prometheus to today’s newspaper headlines, humans have always been afraid of new technology. What is AI, and what is AI in the marketing context? And should we be afraid? James Lumley explains more.

“AI chatbot ‘encouraged’ Windsor Castle ‘assassin’ to carry out Star Wars plot to kill Queen,” ran a Daily Mail headline in July 2023. It is the sort of scare story that has a long history. Articles predicting death, destruction, mass unemployment and other specified and unspecified bad things have been regular media staples since the newspapers lost interest in the destructive power of the Metaverse.

In the same month, we saw Hollywood using AI to make Harrison Ford 50 years younger in the latest Indiana Jones movie and The Beatles releasing a new single that used AI to clean up John Lennon’s vocals. AI is everywhere we look.

And, as Barney O’Kelly, a marketing director at Alix Partners, points out, humans have always feared innovation and change, and catastrophe stories about tech killing us all have been a Hollywood staple for decades, from 2001 to Terminator. Equally, promises that labour-saving tech will lead to a utopia in which the bots do all the work are also long in the tooth.

“Neither has happened yet, so the best we can say is ‘so far so good’,” he says. “What we can probably expect is that AI will fit in and be useful, rather than kill us.”

But what is AI generally, and how can it help marketers?

Artificial Intelligence has been around in the tech world for decades.

“There’s nothing new to it,” says Alexander Low, UK Managing Director at DCM Insights and host of the Death of a Salesman podcast.

“How do you think your car knows how to turn on the wipers when rain hits the windscreen? One can argue that’s AI. It’s a computer program, following a set of instructions. Water touches windscreen, turn on wipers.”

“When talking about marketing, we need to be clear about what sort of AI we are talking about. We are talking about generative AI, like Chat GTP, Bard, Claude etc.”

Scientists have been using computer modelling for years but in the last decade machine learning, artificial intelligence and cloud computing have come together to create incredibly powerful computing solutions that can interpret massive datasets in a fraction of the time it used to take. It has revolutionised medical research, for example, and some say it might one day cure cancer.

Generative AI has also been around for some time in the form of ‘chatbots’ that might guide us on a customer journey or to the appropriate operator in a call centre. However, recent advances in so-called ‘large language models’ have supercharged the chatbots into things that can write engaging text, create convincing images and craft sounds.

“OpenAI and Microsoft have absolutely played a blinder,” says Low.

“They have made it mainstream and accessible to non-tech people, and now everybody is talking about it.”

Even so, generative, content creating AI still looks to many like a novelty – something that could be useful, but for what?

The reason why it looks a bit like a solution in search of a problem is, according to Jason Cox, also a marketing director at Alix Partners, because generative AI is still probably in its infancy, compared with more technical uses of artificial intelligence.

Content-creating AI still looks to many like a novelty – something that could be useful, but for what?

“Right now, the best-case use is still around data analysis,” he says. “We are all looking at how it can create content, but I think of ChatGTP as being a bit like the old Netscape browser in the 1990s. It gave us all the opportunity to interact with the internet, and that led to the dot.com boom.

“I think we are going to see a lot of tech companies getting into the AI space,” he says.

And when that happens, generative AI will no longer be a solution in search of a problem. It may well instead be the main way that we all interface with the huge amounts of digital data we all now experience in our everyday lives.

Helpful AI tools

Content generation
Syllaby – https://www.syllaby.io/
Copy – https://www.copy.ai/
Storybird – https://storybird.ai/

Design
Booth – https://www.booth.ai/
Eluna – https://www.eluna.ai/
Leonardo – https://leonardo.ai/
D-ID – https://www.d-id.com/

Client insights & posting
Repurpose – https://repurpose.io/
HolyCow – https://holycow.ai/

Transcription
Otter – https://otter.ai/
Monica – https://monica.im/
Metaview – https://www.metaview.ai/

Web and SEO
Framer – https://www.framer.com/ai
10Web – https://10web.io/

Ads Transparency Center –
https://adstransparency.google.com

AI. What is it good for? Quite a lot actually…