In this session, we review thoughtbot's AI Ethics guide in the context of exploring LLMs suggesting tags on blog posts. We'll discuss how this type of problem used to be best solved as a classification problem, and how we can also use embeddings to solve this type of problem without hitting an LLM every time.

We'll then work on the implementation against OpenAI using our existing generic prompt class to send the available tags, the article markdown, and a prompt to get tag suggestions, formatted as a JSON array.