This is the first in a series of articles that explores Generative Artificial Intelligence (Gen-AI) tools in marketing: policy and principles, their tactical or operational deployment, strategy and planning. This first article explores key foundational elements (background, concepts, strengths, limitations, and risks) to provide an understanding of these emerging tools, within a context of Marketing, when designing policy (to be covered in article #2).
Gen AI is creating a wave of change and it’s becoming clear that we are only at the start of it. It will empower the marketing generalist with significant capability and save a huge amount of non-value added time – but they are not the only ones that stand to gain. How organisations approach the use and adoption of these tools will be key to building competitive advantage.
Gen-AI tools in marketing will extend far beyond the current use cases of AI, such as analytics, chatbots, and automation of communications and ads. These tools will allow fast and accurate work with segmentation and targeting, market research, data augmentation, knowledge management, brand consistency, marketing management and through to even strategy and planning. So, how exactly do these new AI tools work, what are the limitations and dangers, and where do you start?
AI: the ultimate in abstraction (and a brief lesson in computational abstraction)
The journey from Colossus, the pioneering British code-breaking computer, to modern AI, showcases the evolution of abstraction in computing. Operating the Colossus (see image below) required the setting and moving of physical switches and cables to configure it to complete a task. The development of silicone chips and machine languages (e.g. Fortran) further distanced human operators from the underlying calculations. These low-level languages enabled the use of human readable code – translating it into something the machine would understand. Further abstraction then arrived via improvements to user interfaces, operating systems and software, mobile devices and Apps.
Today, Gen-AI (in various formats of LLM, MLLM & Gen-AI / GAI) represents a new tier of abstraction. Not just abstracting users away from the specialist skills needed to operate IT hardware and systems, but also the specialist skills and in-depth knowledge from other fields. The marketing generalist can now deploy Gen-AI to perform complex statistical-analysis for market research, rapid video and animation for content marketing, or write code to develop website and apps – all without having the knowledge and experience that would have been required just a short time ago. It might just be the ultimate in abstraction.
AI will drive new levels of effectiveness and efficiency in Marketing Operations
These new AI tools are a paradigm shift in progress. Executing complex tasks is as simple as issuing instructions in plain language. The result is a kind of ‘democratisation’ of technical proficiency. ‘Marketing generalists’, empowered with AI, can execute plans without having specialist skills in the team, or a reliance on agency support*.
For example, deep knowledge of SEO is no longer a prerequisite for incorporating SEO best practices into your content strategy. Generative AI tools like Multimodal Large Language Models (MLLMs) now ‘comprehend’ and apply the rules of these practices as instructed. These “co-pilot” tools provide the marketing generalist with best practice whilst providing suggestions for new content ideas, writing articles and choosing headlines / metadata. Such tools are already available in platforms like WordPress and HubSpot. The result is access to broader strategic capability but without buying in the required deep vertical expertise.
*This is not to say that agencies will become redundant; however, they will have to undertake some shapeshifting and reinvention to stay relevant.
AI elevating Marketing Strategy and Planning
AI is also capable of improving the effectiveness and efficiency of the Marketing Strategy process. Marketing Management and Strategy is steeped in well-established (and not-so well established) models and tools. These have defined rules for their application. Thus, AI can use and apply these tools in the Marketing Strategy process.
Data-led approaches are best-practice. However, there are still large amounts of gut feeling or assumption used in the strategy and planning processes. The labour-intensive nature of data validation, or a lack of relevant skills, can hinder the testing of such assumptions. AI’s capacity for data analysis and strategy formulation can help to ground decisions in validated insights. Unfounded assumptions could be automatically flagged by AI tools, as well as any strategic risks that arise from them. Individual tools like the SWOT analysis, or processes such as competitor research, can be streamlined and implemented with more robust process.
The limitations of Gen-AI tools
Despite its vast capabilities, AI’s creative potential remains limited. It may produce novel combinations of existing ideas (which can sometimes feel entirely new) but cannot originate unique concepts. Human reasoning and creativity will remain indispensable in generating distinctive content and branding strategies. Likewise, while AI can streamline data analysis and strategic planning, the infusion of human insight and innovation is irreplaceable in key elements of marketing activities such as strategy development and branding.
‘Hallucinations’ are also a problem. Sometimes the new MLLMs can appear to make up facts. In time these will become fewer, but it is still a real risk that needs to be considered in how these tools are applied.
MLLMs are *usually* capable of providing good quality outputs when given the right instruction or prompt. The same can not be said for the generation of images and video. The results can vary wildly and can be amusing – and that might work for some advertisers looking to churn out images and video for ads (see the gallery below!). Getting consistent images for your brand will require much more consideration.
The risks arising from ever greater abstraction
Increasing the level of abstraction has the affect of distancing users from the underlying instructions, ‘knowledge’ and logic that provides the result. This presents an opportunity for biases and falsehoods, inherent in AI’s learning process, to have a negative affect. LLM Model training data is often scraped from the web. The web contains largely historical data replete with biases, opinion, and evolving and nuanced language (e.g. sarcasm, satire, parody etc). So they may not always accurately reflect truths and current realities.
Add to this the complexity of human context interpretation, intertwined with emotional cognition – this helps to underscore the challenges AI models face when trying to mimic human judgement.
Pragmatic adoption of AI in Marketing
It’s worth noting that a generalist performing specialist tasks without relevant in-depth knowledge or expertise exposes certain risks. Is the tool performing as expected? Is proper process being followed? Is key terminology understood? How do you know the results are accurate? What safeguards are in place?
Understanding AI’s operational mechanics and the risks inherent in their use is the first step toward establishing appropriate governance and policy for deploying these new tools. Recognising a rapidly evolving landscape, the transformative potential, and the inherent limitations of AI in marketing is essential for leveraging its benefits while maintaining standards, transparency and integrity.
The evolution of Marketing with AI will move very rapidly – and it requires a balanced approach. It’s important to quickly realise the new opportunities and recognise the risks and limitations. By combining AI’s computational power and access to knowledge with human creativity and insight, marketers can achieve new heights. But there must be guidelines and safeguards in place. The first step is to establish clear policy and principles of use and a programme of training in AI developments. This approach will support an ethical, de-risked and effective adoption of new AI tools that deliver significant value for the organisation.
Read next (coming soon): Developing policy and principles of use to support succesful AI adoption in Marketing.
Leave a Reply