I could not agree less with the idea of prompt engineering being an emerging field, or really one that will be here more than a couple of years at best. Right now it’s a stopgap needed because LLMs aren’t actually good enough to translate conversational English into the expected intent. The idea that prompt engineering is here to stay suggests a massive and fundamental inability to conceive of the most likely realities of an actual AI-assisted near future.
I have no idea how someone can simultaneously believe AI is going to become unimaginably capable, but also that that in order to effectively use this extraordinarily capable AI, we will need a highly skilled human to interface with it. If AI really is that capable, it will be able to ask the right questions back and solicit feedback such that it can give satisfactory results to even the lowest common denominator of users.
That is a fair point and I can see how the amount of prompt engineering would reduce. However, even a human has to be told how to solve a task, what tools are available to them, and how to stay within guardrails. Prompt engineering is how that is done.
ML Ops engineers have been real things for a while. It is a specific subset of SRE or DevOps or whatever. I’ve noticed there are particularly useful subsets of the ops skillset useful here.
Prompt Engineers will definitely not be “real jobs” any more than certified scrum masters are real jobs. But they may exist for a while until the industry gets wise. Someone with an interest in this should pay Gartner to get the job title on their magic quadrant somewhere and then all the legacy companies will be falling over themselves to hire this title.
Sorry but I don’t want to read any article that finds equivalence between LLMOps engineering (perfectly reasonable, albeit overly specific) and Prompt Engineering (a real thing, but not something that would ever require studying engineering to be good at).
I’m not sure it requires studying, but it isn’t easy and it definetly requires practice.
You need to write incredibly detailed descriptions as to how the agents should behave, how they should reason and how they should use the tools available to them.
The language is quite esoteric, and every model behaves differently.
8 comments
[ 2.9 ms ] story [ 31.5 ms ] threadI would be interested whether or not you agree with that central point?
I have no idea how someone can simultaneously believe AI is going to become unimaginably capable, but also that that in order to effectively use this extraordinarily capable AI, we will need a highly skilled human to interface with it. If AI really is that capable, it will be able to ask the right questions back and solicit feedback such that it can give satisfactory results to even the lowest common denominator of users.
Prompt Engineers will definitely not be “real jobs” any more than certified scrum masters are real jobs. But they may exist for a while until the industry gets wise. Someone with an interest in this should pay Gartner to get the job title on their magic quadrant somewhere and then all the legacy companies will be falling over themselves to hire this title.
You need to write incredibly detailed descriptions as to how the agents should behave, how they should reason and how they should use the tools available to them.
The language is quite esoteric, and every model behaves differently.