There’s clearly a difference between someone who types queries into ChatGPT and someone who builds LLMs from the ground up, but both will tell you at a pub that they “work with AI.” (I may need to reconsider my local.)
When someone tells me they work with AI, my clarifying questions are quietly designed to sort them into one of three categories.
Type 1: AI Researcher. You are working directly with or building core LLM models. Architecture, training, alignment. For me this is the actual scientific and engineering basis of the field. If your employer is one of five labs or you’re working in an academically aligned field and your calendar is full of training schedules (not at the gym), this is you
Type 2: AI Practitioner. You are using AI to build things for you: documents, code, presentations. The typical ChatGPT end-user. This category is enormous, growing fast, and awesome. Using AI to write your quarterly report is cool. It is not, however, the same as engineering the system that wrote it.
Type 3: AI Engineer. You are building something that consumes or leverages an LLM. A pretty commonly accepted term at this point. You’re writing code that calls AI APIs, doing prompt engineering, wiring models into larger pipelines. The thing you ship is a product or solution that incorporates an LLM somewhere.
After testing this, I think the outcome matters too… Take fine-tuning: if you are adapting a model to advance science or push a research frontier, that’s Type 1 work. If you’re fine-tuning because the application you built needs a better engine under the hood, that’s Type 3.
HIH - Matt