AI and brand naming – do humans still have the edge?

With all this going on you may have missed the UK Government launching an AI Opportunities Action Plan to ‘turbocharge artificial intelligence and deliver a decade of national renewal’ – estimating a productivity boost of 1.5% a year, worth up to £47 billion to the UK each year over the next decade. But how many jobs might AI replace in the process? And will this put the UK’s world-leading creative industries - which generate >£124bn in gross annual value and employ 2.4m people - at risk?

AI is certainly transforming businessand reshaping our lives, initially by increasing the efficiency and productivity of the human workforce; and the speed of adoption is unprecedented. But generative AI, like DeepSeek-R1, appears to threaten core human skills, including writing, still and video imagery (the image for this article was created with AI in less than a minute), coding and much more. Could machines monopolise creativity in the future and replace what we thought of as a uniquely human attribute?
 
To explore whether AI is a threat or a productivity booster to creative human endeavour, we have taken a look at something generative AI, or Large Language Models as they are also known, should be really good at; language and, specifically, brand naming.

AI and brand naming

At first squint, AI appears to be an immediate threat to brand naming agencies like our own, with online marketplaces such as Squadhelp (now Atom.com), GoDaddy, Namelix and Crowdspring already offering ‘AI-powered’ brand naming and even brand design free, or for a small fee. 
 
In principle, their AI brand naming propositions are compelling, especially for small businesses and start-ups, where money is tight or non-existent. But, when using these tools, many appear to have been conceived to sell domain names and don’t have the rigour to create powerful brands by searching for trade mark availability, company names, double-meanings, or to think abstractly, emotionally or, err, creatively. 
 
Ever since ChatGPT was launched we’ve been using it to supplement our human naming activity, and even with highly nuanced prompts the results are usually too literal, too descriptive and too generic, with rarely a result making it through to our naming shortlists, although that does happen from time to time. 

Here’s an example of a typical output for a project we recently completed comparing various AI’s shortlists with our own for naming an AI geo-data platform that brings proprietary and public geo-data together to form a real-time digital twin of earth (read full case study here). DeepSeek's outputs to the same prompts have been added  post completion for comparison:

Case study results

All the human results were screened for trade marks, unregistered rights, domain names, social media handles, pronunciation, double-meaning and visual appeal, Go Daddy and Namelix only appeared to check for available domain names and both generative AIs returned very literal candidates and both needed further checking as the screening wasn't reliable. Virgeo was the client’s preferred candidate. You can read the full case study here.
 
As a result of these experiments, our naming methodology now uses structured AI conversations for rapidly developing naming themes, rather than just the names themselves, which we then brainstorm, tweak and refine through our existing matrix of validation filters. AI does some of the legwork, but rarely comes up with a winner.
 
Don’t get us wrong, ChatGPT, DeepSeek and their peers are impressive tools, that can expand human creativity in the initial stages of a naming project and then rapidly check name options and permutations, associated domain availability, and cultural relevance (checking on the feasibility of Chinese translations for example). But everything always needs to be double-checked as, frequently, you don’t get the same answer twice. The better we get at “prompt engineering”, of course, the better the outputs, but the inspiring 'gold dust' ideas we are always looking for, seems largely to be missing.
 
We also use AI in brand naming for identifying linguistic patterns; testing names for memorability, emotional response, and cross-cultural effectiveness. And then there’s ‘synthetic’ market research. which could rapidly test concepts in future by using synthetic personas. This all provides a powerful development tool, but is it going to replace human naming endeavour and inspiration?

Conclusion, for now

For now we have found that human creativity still remains critical for infusing intuition, personality, humour and storytelling into brand naming that generative AIs cannot yet replicate. AI is getting better and better as the technology advances, but its evident weakness is its lack of human empathy and social awareness, so it is unable to infer or imply moods and rhythms, or add optimistic notes and emotional depth as humans naturally do. Often the results are just plain ugly. While this will no doubt improve, it doesn’t change the fact that human input remains essential when it comes to brand naming.
    
So, for these reasons, we perceive AI as a competent and capable co-pilot on today’s journey, but not the creative force some forecast it to be. At least, not yet.
 
Brand names ‘made by humans’ still appear to have the edge and command a premium when it comes to something as critical as defining the right brand name for your business.

Next we will look at how AI can contribute to or threaten brand design...

Peter Matthews and David Gilbert

Get in touch if you wish to comment on this article, or if you have a brand or brand naming project you would like to discuss with us.