Story

The article argues that generative models can be useful in specific applications, but cautions against relying on 'vibes' or intuition when evaluating their efficacy. It emphasizes the importance of rigorous testing and quantitative analysis in determining the appropriate use cases for generative models.