To me, LLMs feel like "low-code" tools in most applicable domains. They're very, very good at creating a new, novel solution - but specially trained ML models

To me, LLMs feel like "low-code" tools in most applicable domains. They're very, very good at creating a new, novel solution - but specially trained ML models will rule.