Opinion: LLMs Should Not Be the Default Engine for Every AI Task

A blog post published on Unmeshed.io argues that large language models are being misapplied as a catch-all execution tool in AI systems. The author contends that thoughtful decisions about when and how to use LLMs should be made before any prompting begins. The piece suggests that defaulting to LLMs for every task can lead to inefficiency and poor system design. The argument centers on the idea that AI should be used wisely, with LLMs reserved for tasks where they genuinely add value.