There’s been a lot of talk recently about artificial intelligence applied to communication and marketing, with ChatGPT, Midjourney and many other tools all hot topics. My personal view, for what it’s worth, is that this is no fleeting phenomenon, but rather a combination of knowledge, tools and paradigms that will surely revolutionise the way we communication experts work.
Creativity, campaigns, persona design, market research and much more besides: nothing seems to have escaped the grip of AI, with its data, analytics, algorithms and machine learning.
But what potential and prospects can we (and by that I mean everyone, not just large business employees) unlock with the help of AI marketing? To answer this, we have enlisted the help of Alessio Pomaro.
Alessio is head of AI at the Search On Media Group and a software engineer, who has always had a passion for web marketing and technology. He is a LinkedIn Top Voice for Italy and the author of Brand Voice (published by FrancoAngeli) and Voice Technology (published by Dario Flaccovio).
I hope you will find this interview useful and interesting, and a good starting point for adopting the right mindset in this area.
Happy reading!
Hi Alessio, welcome back. At a certain point in your career you ended up going down ‘the AI road’. How did that happen? What piqued your interest, and what is your vision for the future of AI?
I’m a software engineer, and I began my professional career as a full-stack developer. Later I changed direction and focused on SEO, the area of marketing dedicated to obtaining free results on search engines.
I first came across artificial intelligence when looking for new ways to hone the various SEO analysis stages, and the skills I had learnt previously allowed me to get into the swing of things quickly.
I started off with Natural Language Processing systems, and from that moment on I never stopped studying, experimenting and creating services based on AI, aimed predominantly at marketing, e-commerce, advertising, automation and coding.
What are artificial intelligence and machine learning, and why are people talking about them so much at the moment? Can you help us get our heads around these concepts?
Let’s go back to basics. First of all, it’s worth noting that artificial intelligence is an academic discipline, like physics, for example. And it’s a branch of IT that aims to create machines with characteristics that are typically thought of as human, such as visual perception, decision-making skills and the ability to act independently.
But how does a machine develop artificial intelligence? The answer is through machine learning, which is part of the discipline, a subcategory of AI.
Machine learning is a system that can train a model to make useful predictions using input data. It therefore basically allows the machine to learn without specific programming.
Take visual perception, for example: this application of AI uses machine learning to train the system to recognise certain elements in visual content.
In this case it uses deep learning (an additional subset of machine learning), since efficient structures like deep neural networks are required to process the millions of pieces of input data.
Nowadays, when AI and machine learning are mentioned, it is generative algorithms like GPT-4, ChatGPT, Claude, PaLM 2 and Midjourney that immediately spring to mind. They’ve been on everyone’s lips for the past 6 to 8 months or so, although actually they are not that recent an invention.
What has changed is their accessibility. Now these algorithms are available to everyone, as you well know.
Let’s move on. What are generative algorithms, what different types are there and what opportunities do they offer to communication and marketing professionals?
Generative algorithms are systems based on machine learning algorithms trained on an enormous quantity of data, which can generate text, images, audio, video… basically any type of digital output. And they do it in much the same way as a human being would.
How they work can be summarised in three bullet points:
- They generate content based on probability by completing our input: they’re not great ‘wordsmiths’ , they’re clever statistical calculators.
- They’re not intelligent nor do they have understanding (or at least not in the way we understand these terms): they are just well trained in the ways we express ourselves.
- They’re not search engines: they can produce ‘hallucinations’, in other words content that is factually incorrect.
These models offer numerous opportunities. For example, you can use them for project analysis, for clustering search queries or for processing your competitors’ content by automatically extracting the topics they cover. You can use them to extract data from videos and documents, to write content or to transform tables of technical product data into textual descriptions for product pages on online shops. And that not all… You can also get interesting results when analysing reviews, user surveys and project data.
Micro, small and medium-sized businesses are the lifeblood of the UK economy. What advice would you give to business owners and managers to help them make the most of the AI revolution?
The first advice I would give is to experiment as much as you can and try to apply the systems to your personal productivity: this, I think, is the simplest way to understand their potential. Here’s a deliberately trivial example from my own personal experience.
Google Search Central often posts videos and/or podcasts with updates on the latest technical aspects of SEO. I’ve always listened to every episode in the series. Now, however, using GPT-4 to process a transcript of the contents (in English), I can extract a detailed summary (in Italian) divided up into points, as well as the minute of the video where they discuss each individual topic. This saves me a lot of time. I use the same process for many types of content I need to examine in depth, and to quickly create detailed social media posts.
The use of AI in a business should be based on two key concepts: integration into workflows and automation.
AI-based systems can be used, for example, to develop code for automatic corrections.
Another example is content production. If used with plenty of common sense and with careful consideration of the context, you can achieve some excellent results.
Another interesting use is for extracting business data using very simple actions (data democratisation). And then there’s data analysis: machine learning models, if trained and configured well, can be extremely valuable for business decisions in various areas, and therefore provide a competitive advantage.
Finally, the million-dollar question: can machines be creative?
When we think about creativity, we tend to focus on artistic disciplines, such as design or visual campaigns. I’d like to share an alternative viewpoint, through two examples.
Here’s the first. We are currently used to the paradigm of people training algorithms that then perform tasks. In certain areas, however, techniques are used that allow the machines to train themselves. Like a chess player learning not from the books written by the grand masters, but by millions of games played against themselves, in other words from experience. The advantage of this is a reduction in the cognitive bias derived from having learnt from other people. Removing this burden can help you embrace solutions that you probably would never previously have imagined.
Now for the second. During an experimental project I used generative AI to develop the code I needed to analyse a large dataset, create a dashboard and visualise certain metrics. I was astounded to discover that on several occasions the algorithm offered me a different – and better – visualisation from the one I had in mind.
So… can machines be creative? The answer depends entirely on your definition of creativity. But we can definitely use artificial intelligence to enhance our abilities. So let’s do it! And create great things we would never have imagined before.