You can't unit test for taste

A developer discusses their experience of using AI-powered tools to build a feature for their running app, where they used a Large Language Model (LLM) to enrich data and provide a subjective rating for points of interest.