Can you turn your thoughts into typed words? Congratulations, you are a prompt engineer. But here's the catch: the LLMs are updated so frequently that prompts which gave you good results today won't give you consistent results tomorrow, so you've got to constantly update your scripts. Furthermore, every AI company aims to improve the "reasoning" capabilities so much that the idea of prompting will one day be unnecessary. The end goal is that humans will use very basic natural language and the LLM will intepret your needs and "just work". So prompt engineering is a very, very short lived career. You can probably skip it.
I created a shorter podcast from the transcript using notebooklm to get quicker insights.
https://notebooklm.google.com/notebook/60e2c25d-555a-44a4-bf53-d2a8a8bc75b7
Fun to see sander pop up here. We first crossed path in 2022 I think doing open source RL work. Look where we are now.
Can you turn your thoughts into typed words? Congratulations, you are a prompt engineer. But here's the catch: the LLMs are updated so frequently that prompts which gave you good results today won't give you consistent results tomorrow, so you've got to constantly update your scripts. Furthermore, every AI company aims to improve the "reasoning" capabilities so much that the idea of prompting will one day be unnecessary. The end goal is that humans will use very basic natural language and the LLM will intepret your needs and "just work". So prompt engineering is a very, very short lived career. You can probably skip it.
in the pod sander agrees its not a standalone job. but its a skillset for all of us working with LLMs today.