Founder EASTWEST PR and Host of the SPEAK|pr Podcast

Dr. Stylianos Kampakis, or Dr. Stelios for short, is an expert data scientist on a mission to educate the public about the power of data science, artificial intelligence, and blockchain. He is a member of the Royal Statistical Society, an honorary research fellow at the UCL Centre for Blockchain Technologies, and a data science advisor for the London Business School. He is also the CEO of Tesseract, an organisation with a vision of making data science available to everyone, even non-technical people. On top of that, he is an author, with his book, The Decision Maker's Handbook to Data Science, currently selling on Amazon. 

AI works in multiple ways in marketing, and Dr. Stelios shares that one of the main ways is in analytics and trying to answer traditional marketing problems, from improved ways to performing A/B testing to understanding execution in a better way as well as content generation of articles, titles, images, and even videos. A large part of Dr. Stelios' work over the last few years has been around explaining to decision makers from startups and bigger organisations, and helping them understand how they can implement AI in the best possible way, how they can get the most value out of AI and data science, and strategies on designing it. That being said, many of the marketing tools are more focused and easier to use, because the idea behind these tools is that you can use them for very specific purposes. For instance, when it comes to AI generating titles, there's less risk in terms of implementation because the purpose of this is clear.

Companies have their own style, and so one of the best things about AI in marketing is that AI writers can craft content with personality. The AI can be taught these skills so that the content doesn't sound generic. Dr. Stelios says it's not very difficult to do because in natural language generation, there are models like GPT-3 and GPT-2, and these can simple be fine-tuned so that the model learns how to speak in the language of those texts. If you want them to make the network speak like Shakespeare, all you would need to do is feed it some of Shakespeare's work, and then it will learn that language and start speaking in that manner. That's an entirely solvable problem with AI. 

Generative models in AI can create fresh content, eliminating possible issues of plagiarism or copyright lawsuits. Research and development in this area have advanced a lot to the point that algorithms can now write pretty realistic content, albeit with minor improvements needing to be made. Nevertheless, copyright is not considered a major issue when using AI to create content. Realism and writing relevant content might probably even be bigger issues than plagiarism. The network can also be trained in any language or even multiple languages simultaneously, so it can be used multinationally. Training a network does require resources, so if you want to train a network on a language it's not familiar with, you'll need to spend time training it but also collecting the data set, which is where the challenge presents itself. 

The state of the art model in natural language generation is GPT-3, which is a model developed by OpenAI, unlike GPT-2 which was developed by Open Source AI. If you type GPT-2 and Python, you will find some implementations in Python, and it's fairly easy to use. The GPT-3 model is pre-trained, so it knows English, basically. Dr. Stelios actually fed his book called The Decision Maker's Handbook to Data Science into this algorithm, and because he had a certain manner of speaking, the AI generated sentences which sounded like his tone of writing and were somewhat realistic. The text that was produced was factually correct, although there were no deep insights, and it uncovered the relationships between AI as a subfield of computer science, databases, and more.

There's a certain humor and nuance in advertising that isn't in a long-form article, but with AI, that's no problem. One can use similar technology, and it'll work just as well. The technology is there; it's just not as refined yet to be used generally. Another thing is that the models are big, so they may not necessarily be easy to sell to users, but it won't take long before AI content generation becomes widely available. That's not to say that humans don't have a role. AI learns from humans, but for now, it won't be able to come up with novel ideas. In the near future, one can expect to see great content generation services based on AI for content, which is largely "vanilla."

When it comes to technical write-ups like brochures, manuals, or handbooks, which are a big body of work as well for most companies, AI may not be the best approach. These are facts and systematised knowledge that is built-in and put into boxes, and when you're reading a guide, you want it to be very precise. Whereas when talking about AI, it produces content which looks as if it was created by a human. The blurred lines to using AI, too, are in intellectual property rights. If you use AI to create an article, technically, you didn't write it yourself, so do you own the copyright, or can you copyright content that wasn't generated by you? 

AI is great at reaching out across various bodies of information, compiling,  synthesizing, and generating a narrative. From the three main data formats, text, audio, and video, it's possible to create an article from which a video or an audiogram could be made from that same piece of content. And these days, AI is increasingly being used by companies big and small, as it's not just about efficiency. With the pandemic still going on, it's also about using technology in domains where humans might not be able to work because of a virus. 

The impact of AI on marketing would be lower barriers to entry into the game for smaller agencies and companies. AI tools for marketing would become cheaper, making it easier for a smaller company to do the job that a bigger company with a marketing department is doing. This is going to level the playing field, because problem that smaller companies have is the continuation of content, because people are trying to do so many things at once. But with AI helping them out, they are able to focus on other important aspects of the business. There will also be more and more tools as well as better ones continuously being developed to go out into the market. The first step to this is creating cheaper tools, because currently, some of the solutions out there are meant for big companies, but eventually, they will get cheaper and more accessible.

AI is proving to be a global innovation, with it taking place in many countries. The world leader is the US, and China is a close second. People are witnessing a global arms race around AI and data science, and over the next few years, people will see many different pockets of innovation as more countries realise the hidden value in these technologies.

Dr. Stelios' Tesseract Academy has a goal to serve the content needs to help decision makers, and these decision makers can be anyone from a CEO, to an entrepreneur with a startup, to a company manager to help them better understand how they can implement data science without having to go through the all the technicalities and details, which can sometimes seem obscure or esoteric. The Tesseract Academy has worked on topics like data strategy and scoping out AI projects, and they worked with the US Navy and Vodafone. 

A recurring theme in this line of work involves explaining to stakeholders how AI and data science can be used and helping them achieve their goals because, quite often, people lose sight of the big picture. They make poor decisions, or they might not have the right plan in place, and so they teach them about building the right culture for data science adoption, because these are the things that the decision maker needs to deal with early on to get the most value out of their data. And as data science and AI become more and more widespread, it's also a question as to whether you can do this as a competition. Sooner or later, everyone will need to make the change and implement AI if they haven't already in their organisation. So, the Tesseract Academy has been helping decision makers get the most out of this technology in the simplest way possible and in the most efficient way possible.

This article is based on a transcript from my Podcast SPEAK|pr, you can listen here.

Cover Photo from