I don't know how good it was, but sentiment analysis was definitely a thing pre-ChatGPT.

It was pretty basic though, and even a frontier LLM might struggle to infer that OP is a negative-sentiment link, without sufficient context.