Artificial intelligence & naming.
The advent of generative AI opened a huge number of doors. Many are about automation. But some are related to helping the human mind do what it does in a better, more efficient way.
And that’s where naming fits into the AI equation.
Asking today’s generative AI models to name products, services and brands typically results in the delivery of strategy-vacant, highly generic names. This is because large language models (LLMs) are trained using next word prediction, and as such are trained to provide the most predictable answer to a question. Thus, because LLMs excel at providing predictable responses, they are unable to deliver uniquely valuable names.
However, where generative AI adds enormous value to naming is in its potential to mine focused, high-quality language at scale. When employed in a deliberate, guided and bounded fashion, generative AI has the power to deliver Language Clouds with exceptional specificity, breadth and relevance – beyond what the human imagination could achieve on its own. Language Clouds that reveal a landscape of unexpected yet high-potential words that otherwise would have gone undiscovered. And that provide a more robust and unique springboard for a human-led naming exploration.
And, given our years of early investment in building up generative AI experience and internally-developed software, this is exactly what Story Style Sound can do.
How we incorporate the best of AI in our naming process.
Using GPT3 and GPT4 from OpenAI, which we believe are currently the best performing, publicly available LLMs for creative naming purposes, we have built two complementary libraries. One is a series of proprietary prompts specifically engineered for the purpose of strategic language mining.
The other is a toolbox of proprietary software that actions these prompts to produce a series of Language Clouds. Language Clouds built in a way that optimizes their utility in the context of name generation, as well as reflective of an individual project’s specific parameters, considerations and inspirations.
These proprietary libraries encode the branding industry’s best practices around identifying and evaluating language, including the application of linguistic theory and statistics to inform language selection. And the Language Clouds they produce contain distinct vocabulary related to the nuances surrounding a particular naming challenge. Nuances like category perceptions and drivers. Target audience need states and engagement aspirations. And relevant functional and emotional benefits, as well as their perceived value in the eyes of end users or consumers.