Artificial intelligence is being spoken about as though it has arrived carrying a weapon of mass anti-creativity. Some say it will replace writers, artists, filmmakers and make marketers obsolete. Fill the internet with identical content. Make every company sound the same. Destroy trust. And quite possibly finish off civilisation before the next quarterly campaign.
Some of those concerns are indeed legitimate as the speed and scale of generative AI create real risks, particularly when companies introduce it without proper controls. But the truth is that AI is unlikely to single-handedly destroy a brand that is already clear, distinctive and well governed. However, brands that are not so well-defined are far more likely to have their weaknesses be exposed. Maybe that’s what really scares people.
The sameness did not begin with AI
Companies were producing near-identical marketing long before ChatGPT arrived. Banks talked about trust, technology companies promised innovation, professional services firms offered excellence and healthcare companies cared. Across industries, everybody was customer-focused, committed to quality and ready to exceed expectations. Generative AI did not invent those phrases; it inherited them from the industry lexicon.
Large language models generate responses by working from patterns in the material on which they were trained and the information supplied to them. So when a company gives the system vague instructions such as “make this sound professional” the model has little reason to produce anything except a polished version of what professional communication commonly sounds like.
Now that competitors have access to many of the same models and tools, the advantage increasingly comes from the context an organisation can provide. Proprietary knowledge, specific judgement, internal standards and a clear understanding of the customer become more valuable when the technology itself is widely available. The real differentiator is the quality of thinking the organisation brings to the tool.
AI scales whatever it is given
Let’s face it, one of AI’s most attractive qualities is its speed. With it, a marketing team can produce ten headline options, several email versions, campaign concepts, social posts, presentation copy and variations for different audiences in a fraction of the time needed just a few years ago.
McKinsey identifies this ability to create personalised copy, imagery, tone and experiences at high volume as one of generative AI’s major marketing opportunities.[1] The thing is, that opportunity also contains the problem. Speed is useful when the organisation knows what it wants to multiply but dangerous when nobody has agreed on what the output should reinforce.
Basically, if the positioning is sound, AI can help the team explore more ways to express it. If the positioning is generic, it will produce generic work at an industrial scale. AI is an amplifier. Before increasing the volume, the company should inspect its own signal.
A weak brand voice will become obvious
Many Caribbean businesses think they have a brand voice because there’s a “Tone of voice” section in their brand guidelines. It usually contains instructions such as: be conversational and warm, avoid jargon and instil confidence. None of this guidance is useless, but it does not necessarily create a recognisable voice when other brands follow the same guidance.
Employees who have worked for a company for years understand its temperament, its customers, the kinds of claims leadership will tolerate and the phrases that would never sound right coming from it. However, this collective knowledge may never have been codified. Then, with AI, the company begins producing larger quantities of communication and this is where the exposure starts. Various employees will use different prompts; one may ask the system to sound authoritative, another for “elevated but relatable” and maybe a third requests a light-hearted tone. The result is that the company’s voice changes from one touchpoint to the next and AI simply brings those existing inconsistencies to the surface.
A usable voice system needs more than adjectives. It should include sentence patterns, vocabulary preferences, words the brand avoids, its level of formality, how it handles humour, how it explains difficult information, examples of what is right and wrong and a point of view. AI does not have that accumulated organisational instinct unless the company supplies it.
Weak positioning becomes generic content
However, the most serious exposure will not be grammatical. Scarily enough, it will be strategic. When AI is asked to create a campaign for a business that has not settled its positioning, it has to fill the gap, and it will. To do so, it may reach for category conventions, familiar benefits and the messages most commonly associated with that type of company. The result can sound perfectly competent and that is what makes it dangerous.
The headline reads well, the paragraphs flow and the call to action sits neatly at the end. But there is nothing that gives the customer a stronger reason to choose the company. This is not an AI writing problem. Instead, the system was asked to express a distinction that the business never made.
So before a company uses AI to increase its marketing output, it should be able to answer the following questions:
- Who are we trying to matter to?
- What are those customers comparing us with?
- What can we credibly offer that affects their choice?
- What do we want to be known for?
- Which claims could any competitor make just as easily?
AI can help explore possible answers, challenge assumptions and organise research, but the final strategic choice still belongs to the humans in the business.
Contradictions will surface faster
Brands do not communicate only through advertisements and social posts. The mix also includes proposals, sales conversations, service replies, onboarding materials, policy documents, invoices, chat systems and the words employees use when the approved script runs out.
AI is increasingly being introduced across many of those functions. Stanford’s 2025 AI Index reported that the proportion of surveyed organisations using generative AI in at least one business function rose from 33% in 2023 to 71% in 2024.[2] In the Caribbean, as adoption spreads, the contradictions that once remained separated by department can multiply across the entire business.
For example, human resources may describe a company culture that customers and employees do not experience. Marketing may programme one promise into its content tools while customer service works from a different knowledge base. Sales may use an AI assistant trained on old presentations and an automated chat may use language that is warmer, colder or more certain than the organisation can support.
Automation allows the inconsistencies to be repeated continuously and at scale. This makes AI governance part of brand governance. Someone must decide which sources are authoritative, which materials are outdated, which claims require approval and where human review is mandatory. Without that ownership, every department can create its own efficient version of the brand. But efficiency without consistency can create a very efficient mess.
Confidence is not accuracy
There is another uncomfortable possibility: AI can say the wrong thing in a most convincing manner.
We thought that sentence needed to sit by itself.
The US National Institute of Standards and Technology identifies confabulation, information integrity, privacy, intellectual property, harmful bias and homogenisation among the risks organisations should manage when deploying generative AI. NIST defines confabulation as confidently presented false or erroneous content and recommends structured testing, documentation and human oversight.[3]
For brands, the consequences are practical. An AI-generated article may invent a statistic. A service assistant may overstate a policy. A proposal may include a capability the company does not offer. A post may adopt a cultural reference the brand does not understand. A generated image may create an intellectual property problem or represent the company in ways nobody intended.
Guess what? The brand still owns the result.
Customers will not excuse a misleading statement because software wrote it. Nor will they separate the AI from the organisation that decided to publish or deploy its output. Human review therefore cannot mean checking only for spelling and tone. The reviewer must understand the subject well enough to question the content, verify claims and recognise when the system has produced something plausible but wrong. Accountability for what that material says still rests with humans.
The brands with stronger foundations will gain more
None of this is an argument for Caribbean companies to avoid AI. Generative AI can help smaller teams conduct preliminary research, organise information, generate alternatives, adapt content, identify gaps, create first drafts and reduce the time spent on repetitive work. Used properly, it can give a business capabilities that previously required a much larger department.
Keep in mind that the benefit will not be distributed evenly. The companies with stronger positioning, cleaner information, clearer approval systems and more disciplined brand standards will be able to move faster without losing themselves. Companies without those foundations may also move faster, but in several different directions at once.
Audit the brand before automating it
Before introducing AI across marketing, sales and customer communication, the company should audit the information the system will be asked to use.
- Are the positioning and value proposition settled?
- Do the website, sales deck and service materials describe the company in the same way?
- Are there approved sources for facts, product information, policies and claims?
- Does the tone of voice provide enough detail to guide someone who has never worked with the brand?
- What information must never be entered into a public model?
- Which decisions still require human judgement?
- Who reviews AI-generated material, and what are they checking for?
This is especially relevant in Caribbean markets, where marketing departments are often small, employees carry overlapping responsibilities and a limited number of people hold a great deal of institutional knowledge. AI may relieve some of that pressure, but it can also reveal how much of the brand depends on unwritten knowledge held by one or two individuals. If the brand changes when those people are absent or leave, the brand has not been properly embedded across the business.
Companies should treat generative AI risk management as an ongoing process rather than a one-time technical check. That means continually identifying risks, assigning responsibility, testing systems and monitoring their effects throughout use.[3]
AI will make the truth harder to avoid
AI will make acceptable marketing easier and cheaper to produce. But when every company can produce it, simply being acceptable will no longer help a brand stand apart.
The advantage will belong to companies that can bring something to the technology that their competitors cannot: a clear position, useful knowledge, a recognisable voice, disciplined judgement and an organisation capable of delivering what the communication promises.
The company that knows what it stands for can use AI to express that position with greater speed and range. The company that does not will keep asking the technology to generate a position it has not yet decided on. And no prompt, however carefully written, can magically fix that.
References
[1] McKinsey & Company, Unlocking the Next Frontier of Personalized Marketing, January 2025.
[2] Stanford Institute for Human-Centered Artificial Intelligence, The 2025 AI Index Report, 2025.
[3] National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1), July 2024.
