Brands in the age of generative AI
The Internet regularly throws up paradoxes. My paradoxical experience in recent months has been trawling through the now-saturated market of AI-powered brand and logo generators. They are all much of a muchness, which creates a problem for them. While they all promise to deliver the customer a beautiful AI-powered identity for a small fee, none of them offer anything different to each other. As such, not one of these generators actually stand out, and I cannot remember the name of a single one off-hand. So much for differentiation.
But, of course, a logo does not a brand make. A brand is part of an organisation’s DNA; it is in how its employees — both front and back-office — think, feel, and act. Brand and culture are two sides of the same coin. You can’t really have one working in opposition to the other. Culture should be empowering the brand experience, while brand should add a context and framing to organisational culture.
The ways in which brand are manifest are therefore infinite. The sensorial, experiential experiences of a brand are often referred to as touchpoints, but there’s much more to it than that. It’s the implied that adds to the feeling that you get when interacting with that brand, as well as the context added to it from external stakeholders and sources. Digital brands can be polarising in this regard: one might consider Facebook and Twitter as having powerful brands with global reach, offering a “public town square” opportunity to people the world over; others might be less favourable through negative coverage of those brands through the media and elsewhere.
So, if AI — specifically, generative AI — can create a logo, can it create a brand? Can it create its architecture, visual identity, behaviours, guidance on touchpoints?
Of course, generative AI could produce the visual and textual guidance required to implement a brand’s identity. It would instruct the reader as to how the brand should be expressed in literal form: which Pantone shades to use, the size of the logo’s exclusion zone, and how to write both the brand name itself and how to write about the brand in given settings. AI logo generators handle a subset of these requirements in the same way.
The constant in generative AI brand work is that all of the generated brands are new. There is no highly contextual input to inform a particular type of output. If the input is nothing, then you could cycle through 200 differently-generated brands and logos until you get to the right one. It’s fast and cheap because you are cutting out all of the human research that has gone into classical brand projects. Therefore, a generated brand becomes a game of chance in the market: you feel that the brand will be resonant and successful, probably because it looks similar to other brands in the market (and the AI has been trained to that effect) but you don’t necessarily know that it will perform as such. As I mentioned at the beginning, if you don’t feed in the deep context and guardrails in terms of differentiation, don’t expect differentiation to miraculously come out at the end.
As such, where generative AI really helps is in the technical production of a brand, rather than its cultural production. Brand strategists and agencies should really be thinking about how to create foundation models that take this into account. A generative AI model that understands brand and market research in its ingested data will create something that is much more likely to be appropriate but also differentiated. That data should also inform a large language model specific to that brand, which could provide samples of how the brand should textually perform in given scenarios.
That is where generative AI could really add value. While I have been talking about how brand strategists and agencies could benefit, that doesn’t cater for anyone else in terms of the brand’s stakeholders and touchpoints. A generative AI service that offers examples and guidance for how the brand should be used could be really democratising. We have all seen examples in global businesses where local brand executions have either mis-interpreted or even completely ignored the brand that they should be referencing. Offering anyone in the business a chance to understand the brand conversationally allows for a transfer of power — the brand isn’t owned by the brand team, it’s owned by everyone. Empowering people in this way through a conversational interface requires a certain element of trust, but if trust isn’t part of your emotional brand architecture already, perhaps you’re doing it wrong.
I’m not suggesting the immediate rollout of a conversational interface to thousands of people. There would, of course, be due process, a degree of fine-tuning, and the prevention of hallucinatory productions to prevent the guidance from going off-piste. But, after working through these factors, there should be nothing to stop local teams putting tactical campaign requests to generative AI — “Give me a fun look to the brand, using a vinyl wrap of a van by a beach” — which would massively reduce incorrect executions while freeing up central brand teams to do their highest value work.
Overall, there is huge potential for generative AI to be part of brand inception, development, and adherence. Gen AI services don’t just have to be at the end of the process — they can be at the start too, throughout a brand’s development, and for many years beyond the brand’s launch. However, like the introduction of any enterprise software, it has to be done properly, with considerable and continual consultation with mapped processes and an unrelenting focus on operational optimisation for maximum effect. The time is right to examine, experiment and experience what this genuinely transformational technology can be capable of.
