Use AI to Stand in for a Domain Expert (kerrick.blog)

A recent development in AI tools emphasizes the use of large language models (LLMs) to uncover domain-specific jargon, bridging the gap between software developers and the experts in the fields they're targeting. When developers lack familiarity with the language used by domain experts—often including field-specific jargon—they risk creating solutions that misalign with user expectations, leading to unreliable software. The inability to communicate effectively with expert vocabulary can hinder the knowledge acquisition essential for developing a successful product. The proposed solution involves leveraging LLMs to extract the pertinent terminology by crafting targeted prompts that utilize few-shot learning, a method that provides examples for context. By querying multiple language models for semantic nouns that describe relationships within the desired domain, developers can gather diverse insights and validate them through further research. This approach not only enhances the accuracy of the software but also ensures it resonates more effectively with users, ultimately improving usability and user acceptance in various applications.