Beyond chat: Rethinking how we use LLMs

Large language models (LLMs) have popularized chat interfaces as a dominant form of interaction, largely due to their initial appeal in mimicking human conversation, as demonstrated by platforms like OpenAI's ChatGPT. However, chat interfaces may not always be the most efficient or effective means for utilizing LLMs, as they can add unnecessary complexity and strain resources. Alternative interfaces, such as dashboards or web apps, can better serve scenarios like personalized recommendation engines or business intelligence dashboards, where interactions are more structured. The article suggests a hybrid model that combines automated initial drafts with selective chat-based refinements. This approach balances LLM automation with user-driven interaction, allowing for more efficient workflows and improved user experiences. Examples like Microsoft's Copilot in Visual Studio Code illustrate this model by enabling targeted and controlled interactions, thereby enhancing performance and usability while reducing the over-reliance on chat interfaces for all LLM applications.