It happens more and more often, and it has become a silent ritual in agency–client dynamics.
We deliver a structured editorial plan, copy for a new website or a rebranding strategy. The work is the product of weeks of analysis, meetings and competitor research.
A few minutes after sending it, we receive a reply. It is a bullet-point list: cold, surgical. It talks about “optimising flow”, suggests using generic terms such as “synergy” or “360-degree innovation”, and criticises deliberate stylistic choices as “unconventional”.
A quick read is enough to recognise the signature. It is not the Marketing Director’s, nor the business owner’s. It is the unmistakable signature of ChatGPT.
Generative Artificial Intelligence is, without doubt, the technological revolution of our decade. At Bliss, we too use it every day to process data or automate repetitive tasks. However, when the operational tool is elevated to the rank of Strategic Judge, it triggers a short circuit that endangers the health of the business: the Hallucination of Control.
In this in-depth piece, we examine why AI, while an excellent executor, makes a poor Creative Director, and why entrusting it with the validation of your project risks condemning your brand to invisibility.
Bliss
Speak to a
consultant
Tell us about your project. Together, we will find the right way forward.
Contact us1. The Mathematics of Mediocrity: How AI “thinks”
To understand why ChatGPT’s feedback is often harmful to strategic creativity, we need to look under the bonnet. Language models (LLMs) do not “think”, do not “understand” and have no “taste”. They are predictive statistical engines.
Their task is to guess the most probable next word in a sentence, based on an immense volume of training data that represents the average of everything written on the internet to date. This phenomenon is technically related to the concept of Hallucination in Artificial Intelligence, where the machine generates answers that seem plausible but are factually or logically inconsistent with the specific context.
Herein lies the paradox. In marketing, success lies not in the average but in the exception. A brand becomes memorable when it breaks the mould, when it creates a cognitive dissonance that captures attention. Asking ChatGPT to “correct” a piece of creative copy means asking a system designed for statistical standardisation to assess a strategic anomaly.
The result? AI will inevitably tend to:
- Smooth out the rough edges: It will remove provocative or ironic sentences because they are statistically “risky”.
- Insert clichés: It will suggest safe, overused words, making your brand indistinguishable from its competitors.
- Regression to the mean: It will turn a brilliant idea into a “correct”, polite and utterly ignorable concept.
2. Contextual Blindness: Data without vision
The insurmountable limit of today’s AI is its lack of deep contextual awareness. When you paste a text into the chat to get an opinion, the AI analyses it in a vacuum. It sees the syntax, it sees the semantics, but it is blind to the real world.
What does the algorithm ignore while it judges your project?
- The Brand’s History: It does not know that this specific choice of words serves to reclaim one of the company’s founding values.
- The Market’s “Unspoken”: It does not perceive the cultural nuances or emotional trends emerging at that precise moment.
- The Long-Term Strategy: They do not know that this content is only the first piece of a complex Strategic Marketing plan of ours that will unfold over the next 6 months.
A human Creative Director assesses a piece of work by asking: “Will this work for the business objectives a year from now?”. ChatGPT assesses it by asking: “Is this sentence statistically probable?”. They are two different sports.
3. The Psychological Trap: Automation Bias
If the technical limits are obvious, why do so many entrepreneurs keep seeking validation from AI? The answer lies in cognitive psychology. It is called Automation Bias: the human brain’s natural tendency to favour suggestions generated by automated systems, regarding them as more “objective” and “reliable” than human judgement.
In a situation of uncertainty, such as launching a new campaign or overhauling a visual identity, the entrepreneur feels anxious. Human feedback (which can be complex, nuanced, sometimes harsh) does not always ease that anxiety. AI, by contrast, always responds in an assertive, structured and reassuring tone. It offers the illusion of a mathematical “quality check”.
But this confidence is a hallucination. Validating a Brand Identity with ChatGPT is like asking a satnav to judge the beauty of the landscape: it will tell you the road is straight, but it will never be able to tell you whether the journey is worth it.
4. Empathy cannot be simulated
There is one final factor, the most critical: Emotional Resonance. Marketing is not pure logic; it is the transfer of emotion. To persuade a customer to choose you, you have to pull irrational levers, evoke memories, stimulate mirror neurons.
An algorithm can simulate empathy by using emotive words, but it cannot feel it or anticipate it. Only a team of professionals, made up of human beings who live, suffer and rejoice just like your target audience, can calibrate the tone of voice to create that subtle resonance that turns a user into a loyal customer.
Conclusions: Returning to the Big Picture
We are not suggesting that technology should be abandoned. On the contrary, at Bliss we are pioneers in the use of advanced data analysis tools. But we draw a clear line between support and guidance.
Artificial Intelligence is an extraordinary tool for running faster, but it does not hold the map to decide the destination. The most valuable feedback for your business remains feedback grounded in the Big Picture: that distinctly human ability to connect the dots between data, psychology, market context and future objectives.
If you want your brand to rise above the background noise, stop seeking the approval of the statistical average. Start trusting strategic intuition, and those who have the big picture to take you where the algorithm cannot go.